Bridging Built Environment Attributes and Perceived City Images: Exploring Dual Influences on Resident Satisfaction in Revitalizing Post-Industrial Neighborhoods

Abstract

The deterioration of physical spaces and changes in the social environment have led to significant challenges and low life satisfaction among residents in post-industrial neighborhoods. While resident satisfaction is closely linked to the built environment, physical attributes alone do not directly influence human feelings. The perception and processing of urban environments, or city images, play a critical mediating role. Previous studies have often explored the impact of either city image perception or physical space attributes on resident satisfaction separately, lacking an integrated approach. This study addresses this gap by examining the interplay between subjective perceptions and objective environmental attributes. Unlike previous studies that use the whole neighborhood area for human perception, our study uses the actual activity ranges of residents to represent the living environment. Utilizing data from Shenyang, China, and employing image semantic segmentation technology and multiple regression methods, we analyze how subjective city image factors influence resident satisfaction and how objective urban spatial indicators affect these perceptions. We integrate these aspects to rank objective spatial indicators by their impact on resident satisfaction. The results demonstrate that all city image factors significantly and positively influence resident satisfaction, with the overall impression of the area’s appearance having the greatest impact (β = 0.362). Certain objective spatial indicators also significantly affect subjective city image perceptions. For instance, traffic lights are negatively correlated with the perception of greenery (β = −0.079), while grass is positively correlated (β = 0.626). Key factors affecting resident satisfaction include pedestrian flow, traffic flow, open spaces, sky openness, and green space levels. This study provides essential insights for urban planners and policymakers, helping prioritize sustainable updates in post-industrial neighborhoods. By guiding targeted revitalization strategies, this research contributes to improving the quality of life and advancing sustainable urban development.

1. Introduction

In 1973, Harvard sociologist Bell first introduced the concept of the “post-industrial society”, noting that the economy would shift from goods production to a service-based economy in the post-industrial era [1]. This transition led to the decline or relocation of many industries to suburban areas, leaving urban landscapes filled with abandoned factories and creating post-industrial neighborhoods [2]. In regions such as Germany’s Ruhr area and the Lehigh Valley in the United States, this transformation not only altered urban landscapes but also promoted socio-economic restructuring, presenting challenges and opportunities for urban revitalization and heritage preservation. A similar trend has emerged in Asia, although at a later stage. For instance, neighborhoods in northeast China exhibit a complex interaction between industrial heritage and contemporary urban development [3].

Post-industrial historical areas are the result of the collision and fusion of industrial civilization and contemporary development, carrying the era’s memories of the worker community. Improving the living environment of post-industrial neighborhoods has become a significant issue in urban studies [4]. Residents face significant problems such as deteriorating road quality [5], severe industrial heritage decay, poor landscape appearance [6], low green space levels, and insufficient open spaces. These neighborhoods, established during the early stages of industrial development, now face a fundamental conflict between the low-quality, high-density buildings and the modern urban residents’ desire for an improved quality of life. This situation underscores the pressing need to renew and revitalize post-industrial neighborhood environments [7].

Renewing post-industrial neighborhoods focuses on improving residents’ quality of life and subjective well-being. Resident satisfaction is closely linked to the quality of the built environment; however, physical space attributes alone do not directly influence people’s subjective feelings. Instead, the perception and processing of urban imagery play a crucial mediating role [8]. Variations in living ranges and activity areas lead to different perceptions of physical space among residents. Additionally, demographic characteristics result in varied impressions of the spatial environment after cognitive processing. Urban image, which reflects urban spaces as experienced and processed by the human brain, serves as a bridge between physical space and residents’ feelings.

Since the 1960s, when Kevin Lynch proposed the method of urban image cognition based on five spatial elements [9], cognitive maps of urban images have been an important means of understanding urban spatial structures [10] and have been instrumental in urban design. Researchers have successively applied classical methods or models of urban image studies, continually innovating research methods and expanding the scope of urban image theory from different angles. Urban image research tools and methods have evolved significantly over time. Initially, qualitative methods such as questionnaire analysis [11] were used to capture the subjective image and cultural connotations of neighborhoods. These early approaches focused on residents’ perceptions and experiences. As the field advanced, researchers began to incorporate spatial syntax to measure two-dimensional topological relationships [12] like spatial accessibility. Recent developments in urban studies have introduced advanced quantitative tools such as fully convolutional networks (FCNs) [13,14] and DeepLabv3 [15] for image segmentation, integrating these with POI data [16] and remote sensing [17] to analyze city perception and urban renewal. Mobile signaling data has also been used to reflect residents’ activity areas [18], aiding in urban planning by aligning actual activity zones with administrative boundaries [19]. The integration of these tools has led to a comprehensive approach that combines qualitative methods, such as in-depth interviews and field observations, with quantitative data analysis. This evolution from subjective, qualitative assessments to a mixed-methods approach has significantly enhanced the understanding of urban image, improving both the credibility and persuasiveness of the research.

Most previous studies have explored the effects of image perception and physical space attributes on residents’ satisfaction from either a subjective or objective perspective. However, research has shown that the mechanism affecting residents’ satisfaction involves physical space first influencing people’s perceptions, which then further affect their feelings [20]. Therefore, it is necessary to systematically integrate the influences of image perception and physical space attributes on residents’ satisfaction. However, existing studies have rarely focused on this integration.

This study takes Shenyang, China, as a case study. By matching residents’ subjective perceptions with their objective activity ranges and using image semantic segmentation technology and multiple regression methods, we analyze the impact of subjectively perceived image factors on residents’ satisfaction and the influence of objective spatial indicators on perceived image factors. We further integrate the mechanisms of both subjective and objective aspects to rank the capability of objective spatial indicators in influencing residents’ satisfaction. This research aims to reveal the subtle relationships between subjective image perceptions, objective physical elements, and residents’ satisfaction within post-industrial historical areas, helping urban planners and policymakers propose targeted revitalization strategies for these areas.

The research framework (Figure 1) involves three steps. First, 27 indicators based on Kevin Lynch’s 5 elements of urban image were assessed through a survey, and exploratory factor analysis identified subjective image factors. Next, 7 objective and 20 planning indicators were selected using literature and image segmentation techniques. Finally, multiple regression analyses linked these subjective factors to both resident satisfaction and objective spatial indicators, determining the impact of objective indicators on satisfaction and highlighting the interaction between subjective perceptions and urban environments.

The key questions this study aims to address are: (1) How do subjective image perceptions affect residents’ satisfaction? (2) How do objective spatial indicators influence residents’ subjective image perceptions? (3) How do objective spatial indicators influence residents’ satisfaction under the regulation of subjective image perception?

The following Materials and Methods Section first provides a detailed introduction to the urban image evaluation indicators and objective image evaluation indicators in post-industrial neighborhoods. It then outlines the methods employed in this study, followed by a description of the study area and data collection procedures. The Results section elucidates the impact of subjective image factors on residents’ satisfaction, the influence of objective spatial indicators on residents’ subjective image factors perception, and the potential impact of objective spatial indicators on residents’ satisfaction. In the Discussion section, we analyze the limitations of this study and discuss the policy implications of our findings. The paper concludes by summarizing the study and emphasizing key insights derived from the survey. Figure 1 illustrates the research framework used.

2. Materials and Methods

2.1. Evaluation Indicators

2.1.1. Evaluation Indicators for Perceived Urban Image

Defining what makes a better city has always been a question pondered by contemporary architects and urban designers. Kevin Lynch proposed five elements of urban image: district, edge, node, path, and landmark [9], along with five basic indicators of urban image, including vitality index, livability, suitability, accessibility, management level, and two additional indicators: efficiency and equity [21]. Based on the above elements of urban image, subjective perception indicators for residents were determined in this study. Since Kevin Lynch suggested the possibility of forming standard theories to evaluate urban spatial value, contemporary urban designers have made considerable efforts to construct an evaluation system that can measure basic urban indicators.

Because there is no standardized scale to measure residents’ subjective perception of the built environment in post-industrial neighborhoods, this study established a customized set of dimensions and attributes. In order to design and improve this measurement scale, we convened a focus group consisting of four experts in urban planning and urban renewal, and two graduate students majoring in urban and rural heritage conservation. This scale is based on the 5 major elements of urban imagery, which are subdivided into 7 main imagery elements and 20 secondary imagery elements.

Table 1. Subjective Perception Indicators of Residents in Post-Industrial Neighborhoods toward the Built Environment.

Categories	Primary Indicators	Secondary Indicators
Path	Road Quality [22]	Pedestrian Comfort [23]
		Pedestrian Convenience [24]
		Pedestrian Safety [25]
Landmark	Quality of Industrial Heritage and its Ancillary Buildings [26]	Historical Significance of Industrial Heritage [27]
		Richness of Industrial Heritage Interface [28]
		Protection and Restoration of Industrial Heritage [29]
		Coordination between Old and New Buildings [30]
		Accessibility of Industrial Heritage [31]
		Openness of Industrial Heritage [32]
Node	Quality of Plazas, Parks, and other Open Spaces [33]	Recognizability [34]
		Accessibility [35]
		Number of Rest Facilities [36]
		Feeling of Rest [37]
District	Appearance Feeling	Overall Appearance Change in the Area
		Environmental Quality of the Area
	Sky Feeling [38]	Sky Openness [39]
		Aesthetic [39]
	Green Feeling [40]	Green Space Ratio [41]
		Distance to Green Space [41]
Edge	Degree of Regional Features	Uniqueness of the Area

provides a comprehensive overview of these attributes as a basis for further analysis.

Paths play a crucial role in shaping urban image. Paths typically refer to urban roads, including vehicular roads and pedestrian walkways. Road attributes involve width, connectivity, pavement materials, and so on. Correspondingly, at the perceptual level, these attributes are reflected in the aspects of walking comfort [23], convenience [24], and safety [25].

The primary distinction between post-industrial neighborhoods and typical industrial neighborhoods lies in their historical built environments. Consequently, elements of post-industrial heritage have become significant landmarks in these areas, playing a crucial role in identifying relevant influential attributes [42]. The quality of industrial heritage and its associated buildings [26] can reflect the level of preservation and heritage revitalization in post-industrial neighborhoods to a certain extent. The historical significance of industrial heritage [27], the richness of industrial heritage interfaces [28], the state of preservation and restoration of industrial heritage [29], and the harmony between new and old buildings [30] are four secondary indicators closely linked to the uniqueness of the environmental landscape and the overall atmosphere. The accessibility of industrial heritage [31] and openness of industrial heritage [32] are two secondary indicators that play important roles in influencing the interaction experience between people and the environment.

Nodes are key locations in neighborhoods that attract foot traffic, promote interaction, and become centers of urban activities. Open spaces such as squares and parks typically constitute urban nodes [33]. People’s perception of these nodes generally includes two levels: at the overall level, people focus on the recognizability [34] and accessibility [35] of open spaces such as squares and parks; at the level of resting facilities, people are concerned with the quantity of resting facilities [36], and the quality of the resting experience [37].

The evaluation of appearance perception is a crucial indicator for evaluating the quality of a region. Areas with distinctive appearances highlight unique natural and cultural environmental features, creating high-quality living and working environments that meet people’s needs. For residents living in historical environments, their perception is influenced not only by the quality of the overall appearance but also by changes in it. This involves psychological factors such as a sense of identity and familiarity. Therefore, people primarily perceive changes in the overall appearance and the quality of the regional environment. The perception of the sky was not initially included in the original urban image theory, but later research has identified it as an important indicator for evaluating urban imagery. Therefore, the perception of the sky is included in the evaluation of the five elements of urban image [38]. The perceived indicators include the openness of the sky and the aesthetic quality of the sky interface [39]. Similarly, greenery perception has also been proven in some articles to be an important indicator for evaluating urban image [40]. The secondary indicators are the extent of greenery coverage and the distance to green spaces [41].

The characteristics of a region are mainly manifested in its internal homogeneity and external heterogeneity. The boundaries of the five elements of urban image usually delineate the internal and external areas of a region. Therefore, the distinctiveness of the region is chosen as a primary indicator for boundary imagery. Its secondary evaluation indicators include the assessment of the uniqueness of the region compared to other areas.

2.1.2. Objective Evaluation Indicators Based on Visual Imagery

In the evaluation system of computer image segmentation, some objective physical indicators can be derived from the calculations of segmented image pixels [43]. This study employs frequency analysis to conduct statistical analysis using street view images. The evaluation indicators were initially selected based on a review of 50 relevant studies and include the Green Visual Index (GVI) [44,45], Sky Visibility Index (SVI) [39,46], Street Feasibility Index (SFI) [47], Vehicle Interference Index (VII) [22], Traffic Sign Index (TSI) [48], Accessibility of Industrial Heritage (AIH) [49], Open Space Index (OSI) [50], and Social Venue Attributes. Due to limitations in image recognition algorithms and data availability, the Social Venue Attributes indicator was excluded, resulting in seven final objective computational indicators.

Table 2. Objective Calculation Indicators for Post-Industrial Historic Districts.

Indicators	Formulas	Indicator Interpretation
Street Feasibility Index	$SF{I}_n={\displaystyle \frac{{\Sigma }_{i=1}^i{W}_i}{{\Sigma }_{i=1}^i{R}_i}}$	The Road Feasibility Index can be intuitively reflected in the results of street view imagery segmentation by the ratio of the pixel areas of pedestrian pathways (Wi) to vehicular roads (Ri). A higher ratio (SFIn) indicates better walkability [47].
Vehicle Interference Index	$VI{I}_n={\displaystyle \frac{{\Sigma }_{i=1}^i{M}_i}{{\Sigma }_{i=1}^i{R}_i}}$	The Vehicle Interference Index is based on the premise that the presence of numerous vehicles reduces the perceived safety of the environment. The more vehicles there are, the lower the safety. This can be intuitively reflected by the ratio of the pixel areas of motor vehicles (Mi) to vehicular roads (Ri). A higher ratio (VIIn) indicates a higher vehicle interference index, making the environment feel less safe [22].
Traffic Sign Index	$TS{I}_n={\displaystyle \frac{{\Sigma }_{i=1}^i{T}_i}{{\Sigma }_{i=1}^i{S}_i}}$	The Traffic Signage Index can be intuitively reflected in the results of street view imagery segmentation by the ratio of the pixel areas of traffic lights and traffic signs (Ti) to all elements (Si). A higher ratio (TSIn) indicates a greater number of required traffic signs in the traffic environment, implying increased road complexity and lower pedestrian safety [48].
Accessibility of Industrial Heritage	$AI{H}_n={\displaystyle \frac{{\Sigma }_{i=1}^i{A}_i}{{\Sigma }_{i=1}^i{W}_i}}$	The Accessibility of Industrial Heritage can be intuitively reflected in the results of street view imagery segmentation by the ratio of the pixel areas of industrial heritage (Ai) to pedestrian paths (Wi). A higher ratio (AITn) indicates better accessibility to industrial heritage [49].
Open Space Index	$OS{I}_n={\displaystyle \frac{{\Sigma }_{i=1}^i{O}_i}{{\Sigma }_{i=1}^i{S}_i}}$	The Open Space Index can be intuitively reflected in the results of street view imagery segmentation by the ratio of the pixel areas of open spaces (Oi) to all elements (Si). A higher ratio (OSIn) indicates a higher proportion of open activity spaces like squares [50].
Green Visual Index	$GV{I}_n={\displaystyle \frac{{\Sigma }_{i=1}^i{G}_i}{{\Sigma }_{i=1}^i{S}_i}}$	The Green Visual Index in street view imagery segmentation results can be intuitively reflected by the ratio of green vegetation (Gi) to all elements (Si) pixel districts. A higher ratio (GVIn) indicates higher comfort [44,45].
Sky Visibility Index	$SV{I}_n={\displaystyle \frac{{\Sigma }_{i=1}^i{V}_i}{{\Sigma }_{i=1}^i{S}_i}}$	The Sky Visibility Index can be intuitively reflected in the results of street view image segmentation by the ratio of the pixel areas of the sky (Vi) to all elements (Si). A higher ratio (SVIn) indicates higher spatial visibility and greater comfort [39,46].

provides a comprehensive overview of these indicators as the basis for further analysis.

The Street Feasibility Index (SFI) evaluates the visual environmental impact from a horizontal viewpoint, taking into account the proportion of pedestrian-related infrastructure (such as sidewalks and fences) relative to roadways [47]. The Vehicle Interference Index (VII) assesses the concentration of motor vehicles on the street, affecting travel patterns, safety, and public evaluation of street landscapes [22]. A higher Traffic Sign Index (TSI) indicates a greater need for signs to explain the traffic environment, correlating with increased road traffic complexity and decreased walking safety [48]. The Accessibility of Industrial Heritage (AIH) reflects the ease with which people can use industrial heritage sites and serves as a measure of their openness [49]. The larger the Open Space Index, the higher the proportion of open activity space such as the square [50]. The Green Visual Index (GVI) represents the visibility of green space from a human perspective, calculated as a proportion of all the greenery in the image [44,45]. Lastly, the Sky Visibility Index (SVI) quantifies the visibility of the sky as observed by the human eye, calculated as the proportion of the sky within the imagery [39,46].

2.2. Research Methods

The research employed exploratory factor analysis (EFA), multiple regression analysis, and fully convolutional neural network image semantic recognition. The aim was to explore the influence of objective spatial indicators of post-industrial historic districts on subjective perceptual impressions and the impact of objective spatial indicators on resident satisfaction.

2.2.1. Exploratory Factor Analysis

Exploratory factor analysis (EFA) is a statistical method for reducing the dimensionality of indicators, used to identify a set of latent factors and effectively summarize a series of indicators [51]. This method determines the number of latent factors underlying the indicators and examines the relationship between each indicator and these factors, known as factor loadings. By qualitatively assessing the content of indicators that strongly influence the factors, researchers can deduce the conceptual meaning of the identified factors [52]. Due to its capacity to streamline datasets and unveil hidden structures, EFA has found extensive use in social sciences, market research, psychology, and various other fields to enhance researchers’ understanding of the underlying structure and dimensions of data [53].

In previous studies, exploratory factor analysis (EFA) has typically been used to reduce the number of indicators, achieving dimensionality reduction and making data processing more convenient [54,55].

This study employed 27 subjective indicators, which is relatively large, thus necessitating the use of EFA for dimensionality reduction. Prior to applying EFA, the Kaiser–Meyer–Olkin (KMO) test and Bartlett’s sphericity test were conducted [53]. Factor rotation can be used to improve the interpretability of common factors, making it easier to understand the relationships between these factors and observed variables.

2.2.2. Multiple Regression Analysis

Multiple regression analysis explores the relationship between a dependent variable and multiple independent variables. The multiple regression model is defined as:

$$E\left(y\right)={\beta }_0+{\beta }_1{x}_1+{\beta }_2{x}_2+\dots +{\beta }_p{x}_p$$

(1)

where ${\beta }_0, {\beta }_1, {\beta }_2, \dots , {\beta }_p$ are the parameters and $x_1, x_2, \dots , x_p$ are predictor variables.

The advantages of multiple linear regression lie in its ability to simultaneously handle multiple independent variables, offering robust predictive power and intuitive model interpretability. By quantifying the impact of each independent variable on the dependent variable through regression coefficients, and supporting a variety of statistical tests (such as F-tests and t-tests) to assess the significance of the model and variables, it provides a comprehensive analysis. Additionally, it possesses excellent scalability, allowing for the inclusion of interaction terms and nonlinear terms to capture more complex relationships [56]. Key aspects of multiple linear regression involve assuming a linear relationship between the independent and dependent variables, the use of the least squares method to estimate regression coefficients, and conducting residual analysis for model diagnostics. It also involves identifying and addressing multicollinearity issues and uses metrics such as R-squared and adjusted R-squared to evaluate the model’s goodness-of-fit and predictive capabilities. However, it has high requirements for data quality and sample size [57].

In this study, multiple regression analysis was used to investigate the effect of various objective spatial indicators on residents’ subjective perception factors, using the former as independent variables and the latter as dependent variables. Additionally, the influence of subjective perception factors on overall resident satisfaction was investigated, using subjective perception factors as independent variables and overall resident satisfaction as the dependent variable. Multiple regression analysis was conducted with the help of IBM SPSS Statistics software (Version 27).

2.2.3. Semantic Segmentation of Objective Imagery Indicators

With the advancement of machine learning technology, the integration of street view imagery alongside deep learning techniques gained popularity and credibility in urban studies [58]. Numerous scholars have conducted research in this area, such as utilizing street view data and deep learning approaches to study street canyons [59], visual characteristics of streets [60], residents’ living standards [61], and residents’ perceived preferences for the built environment [13,47]. This integration of street view images and deep learning has proven crucial in identifying and evaluating urban spatial characteristics, as well as understanding residents’ psychological perceptions.

Fully convolutional network (FCN) imagery semantic segmentation is a computer vision technique that classifies every pixel in an image into predefined categories. Convolutional neural networks (CNNs) are the main DL architecture for image classification [62]. By transforming traditional CNNs into a fully convolutional format, FCNs can produce output maps that match the size of the input imagery, enabling pixel-by-pixel classification [63]. As illustrated in Figure 2, each CONV block represents a convolutional layer that processes feature maps to extract relevant patterns, and the CONCAT block indicates the concatenation of feature maps from different scales, which are then used for further processing to produce the final segmentation map. FCN proposed a method for end-to-end image segmentation using fully convolutional layers, which has the advantages of accepting input images of any size and being more efficient compared to traditional CNN methods. This technology is widely used in various fields, including autonomous driving [64], street view imagery analysis [65], and satellite imagery processing [66].

The fully convolutional network (FCN) program used in this study was provided by the High-Performance Computing Laboratory (CUG.HPSCIL) and developed by Professor Guan Qingfeng and his research team from the School of Information Engineering at China University of Geosciences, based on the ADE_20K dataset [67]. FCN imagery segmentation does not require extensive preprocessing of input imageries, allowing for direct analysis of scraped street view data. The output includes 16-bit bitmap imageries and feature data for all 150 object categories.

In the field of urban and rural planning, not all elements can serve as analysis objects. Guided by the framework of urban imagery theory, elements were manually integrated and categorized, resulting in 20 major element categories for subsequent data analysis. These categories include buildings, sky, trees, grass, roads, sidewalks, people, plants, cars, houses, open spaces, walkways, huts, trucks, traffic lights, traffic signs, doors and windows, and railings.

2.3. Study Area and Data

2.3.1. Study Area

Shenyang, located in the central part of Liaoning Province in China, is the provincial capital and an important central city. Since the founding of the People’s Republic of China, Shenyang has served as a significant industrial base in the Northeast region, experiencing rapid industrialization and urbanization. Numerous factories and production facilities were established within the city, leading to the development of worker residential districts around these industrial sites to accommodate the workforce.

As time progressed and economic development strategies evolved, Shenyang’s industrial structure shifted toward modernization and higher-end industries. This transformation resulted in the emergence of many abandoned factories within the urban space. The remaining industries either relocated to surrounding districts or moved entirely to other regions. Consequently, former industrial neighborhoods have evolved into post-industrial neighborhoods, facing various challenges.

In Shenyang, the worker villages and residential areas once inhabited by working-class families represent typical post-industrial communities. This study surveyed six such communities located in the central urban area of Shenyang, as depicted in Figure 3.

Despite having distinct historical and cultural backgrounds, industrial contexts, and unique contributions to Shenyang’s cultural and economic structure, these neighborhoods face significant challenges during the post-industrial era. Issues such as deteriorating road quality, dilapidated landscapes, idle industrial heritage, and insufficient public service facilities have led to a decline in residents’ quality of life.

2.3.2. Research Data

The questionnaire data comes from surveys conducted in the previous research [68]. When designing the questionnaire for the earlier study, we included additional questions regarding urban image perception. To ensure the validity of the questionnaire and identify potential biases, we conducted a pilot survey with 19 residents from the target neighborhoods. Based on the shortcomings identified in the pilot survey, necessary adjustments were made to the survey content. After refining the questionnaire, we conducted the formal survey. Respondents were residents living within a 500 m radius of the surveyed communities to ensure they were directly impacted by the local environment.

To ensure the randomness of the survey questionnaires and the comprehensive representation of the surveyed population, interviews were conducted at various times of day and on different days of the week at locations within and around the post-industrial communities. These locations included residential entrances, open spaces, commercial areas, and transportation hubs. A total of 348 questionnaires were collected both online and offline, with 307 responses considered valid. After excluding samples with incomplete information on daily activity locations, 301 samples were retained and used for subsequent analysis.

This study utilized three categories of data from questionnaires: satisfaction with the surrounding environment, demographic characteristics, and levels of urban image perception. Satisfaction with the neighborhood environment was assessed using a single-item measure, where respondents rated their attitudes on a five-point scale ranging from “very dissatisfied” to “very satisfied”. In terms of the perception level of urban image, the questionnaire included items related to the indicators mentioned in Table 1. Respondents were asked to provide feedback on positive descriptions of each indicator on a five-point Likert scale ranging from “strongly disagree” to “strongly agree.”

Table 3. Demographic profile of respondents, sample characteristics (n = 301).

Variable	Value	Percentage
Gender	Female	46.58%
	Male	53.42%
Age Groups	Under 18	6.84%
	18–40	19.21%
	40–60	25.74%
	60–80	42.35%
	Over 80	5.86%
Ethnic Groupe	Han	92.83%
	Manchu	5.54%
	Mongolian	1.63%
Political Identity	Young Pioneer	2.28%
	League Member	3.58%
	Party Member	14.01%
	Reserve Party Member	0.65%
	The Masses	79.15%
	Else	0.33%
Education	Junior high or below	49.19%
	High school	20.52%
	Vocational school	12.38%
	Bachelor	14.98%
	Postgraduate or above	2.93%
Income	Below CNY 50,000	39.09%
	CNY 50,000–100,000	30.29%
	CNY 100,000–150,000	23.45%
	CNY 150,000–200,000	4.89%
	Over CNY 200,000	2.28%
Work Experience 1	Oneself	37.46%
	Family Member	15.63%
	None	46.91%
Residency Duration	Below 5 years	43.17%
	5–10 years	15.31%
	10–15 years	14.32%
	15–20 years	10.91%
	Over 20 years	16.29%

Notes: ¹ This indicates whether the respondent or any of his or her family members had previously worked at a neighboring factory.

outlines the demographic profile of the survey respondents, providing context for respondents’ urban image perception and satisfaction levels.

In terms of gender distribution, the gender ratio of the respondents is nearly equal, with close to 50% representation for both genders. Therefore, there is no significant gender bias that could affect the questionnaire results. Regarding age demographics, 48% of respondents are aged 60 and above, indicating an aging population skewed toward older and potentially economically disadvantaged groups. This suggests a relatively high level of aging within this type of area, with a noticeable decline in the younger population. Concerning ethnic composition, the Han nationality constitutes the vast majority at 92.8%. In terms of political affiliation, the majority of respondents (approximately 79%) are ordinary citizens, commonly referred to as “the masses”, while about 7% are affiliated with a political party. Regarding educational attainment, approximately 82% have education levels below university, indicating a predominance of lower education levels among residents in these historical districts. Regarding income distribution, 69% of households earn less than CNY 100,000 annually. Regarding work experience, over half of the respondents or their relatives have worked in nearby factories. Concerning length of residence, about 42% of respondents have lived in these neighborhoods for over ten years, reflecting their origin as housing built by factories for workers. Despite changes over time, many retired workers and their families still choose to reside here, further emphasizing the authenticity of residents’ perceptions of the urban image within these areas.

Table 4. Descriptive statistics of residents’ perceptions of various indicators of urban image and overall satisfaction.

Attributes	Mean	Very Low	Low	Neutral	High	Very High
Road Quality	4.14	1.99%	12.96%	52.82%	1.33%	30.90%
Pedestrian Comfort	3.55	0.00%	7.64%	36.54%	49.17%	6.65%
Pedestrian Convenience	4.06	0.00%	2.65%	10.30%	65.79%	21.26%
Pedestrian Safety	3.85	0.04%	0.62%	29.57%	53.49%	16.28%
Quality of Industrial Heritage and its Ancillary Buildings	3.31	0.00%	12.96%	44.85%	40.53%	1.66%
Historical Significance of Industrial Heritage	3.36	0.00%	11.63%	43.19%	42.86%	2.32%
Richness of Industrial Heritage Interface	3.59	0.00%	16.28%	27.24%	37.54%	18.94%
Protection and Restoration of Industrial Heritage	3.33	0.00%	23.59%	25.91%	44.19%	6.31%
Coordination between Old and New Buildings	3.31	0.00%	14.62%	40.86%	43.19%	1.33%
Accessibility of Industrial Heritage	3.94	0.66%	5.32%	19.27%	49.17%	25.58%
Openness of Industrial Heritage	4.23	0.00%	4.32%	22.26%	19.93%	53.49%
Quality of Plazas, Parks, and other Open Spaces	3.57	0.00%	10.30%	30.90%	50.83%	7.97%
Recognizability	3.91	0.30%	2.59%	23.72%	53.12%	20.27%
Accessibility	3.64	0.35%	12.62%	20.60%	54.80%	11.63%
Number of Rest Facilities	3.67	0.66%	12.96%	17.94%	55.48%	12.96%
Feeling of Rest	3.57	0.66%	10.63%	26.91%	54.49%	7.31%
Sky Feeling	3.60	0.66%	11.96%	21.93%	57.81%	7.64%
Sky Openness	3.65	0.66%	9.30%	22.26%	59.47%	8.31%
Aesthetic	3.62	1.33%	10.63%	21.93%	56.48%	9.63%
Green Feeling	3.25	6.31%	24.25%	19.27%	38.54%	11.63%
Green Space Ratio	3.27	6.31%	23.92%	17.94%	40.53%	11.30%
Distance to Green Space	3.06	8.30%	19.27%	38.87%	25.25%	8.31%
Appearance Feeling	3.86	0.00%	2.66%	22.92%	60.47%	13.95%
Overall Appearance Change in the Area	3.56	0.00%	7.30%	40.21%	41.53%	10.96%
Environmental Quality of the Area	3.84	0.00%	5.32%	16.61%	67.11%	10.96%
Degree of Regional Features	3.72	5.98%	7.31%	21.26%	39.53%	25.91%
Uniqueness of the Area	3.94	3.65%	6.31%	16.61%	39.21%	34.22%
Resident satisfaction	2.89	0.00%	14.62%	82.72%	1.99%	0.66%

provides descriptive statistics on residents’ perceptions of various urban image indicators and overall satisfaction. The mean perception scores for most attributes are above the midpoint of 3, indicating generally positive perceptions. However, attributes such as green space ratio, distance to green space, harmony between new and old buildings, industrial heritage protection and restoration, and the quality of industrial heritage and its auxiliary buildings have lower mean scores.

The survey included questions on urban image indicators and urban space satisfaction, and questions regarding the locations where respondents conduct their daily activities. This was conducted to collect data on locations that meet their material and non-material needs, such as places of residence, workplaces, shopping districts, entertainment venues, and other relevant places. The place of residence was a mandatory question, and each respondent was required to identify at least three additional key locations. Using the respondents’ daily activity location data, we utilized ArcGIS 10.6 to generate minimum convex polygons (MCPs) to represent their activity spaces. This polygon serves as a spatial representation of their daily activity range [49].

Panoramic street view images are crucial data for this study. Baidu Maps (https://map.baidu.com/, accessed on 6 May 2024), one of China’s largest online mapping service providers, offers 360-degree panoramic street view imagery. By overlaying OpenStreetMap (OSM) road network data with the MCPs generated in ArcGIS, sampling points were established along major roads within the study area. The spacing of these sampling points was determined based on the density of the streets. Python scripts were written to capture 360-degree panoramic imagery at each sampling point, as illustrated in Figure 4. A total of 2636 sampling points were selected, resulting in the collection of 2636 street view panoramic images.

3. Results

3.1. Exploratory Factor Analysis of Subjective Perception Indicators

Before conducting factor analysis, prerequisite data condition tests were performed. The data underwent KMO and Bartlett’s tests, resulting in a KMO score of 0.712, indicating a high correlation among the variables. The significance level was 0.000, demonstrating a strong significance of correlations between the data points. These results strongly suggest that the collected questionnaire data is suitable for factor analysis. The results are shown in

Table 5. KMO and Bartlett’s Test Results.

Kaiser–Meyer–Olkin Measure of Sampling Adequacy.		0.712
Bartlett’s Test of Sphericity	Approx. Chi-Square	3199.841
	df	351
	Sig.	0.000

.

After verifying the data, exploratory factor analysis (EFA) was conducted on the scores of 27 items from all 301 questionnaires. The results are shown in

Table 6. Total Variance Explained.

Component	Initial Eigenvalues			Extraction Sums of Squared Loadings			Rotation Sums of Squared Loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	4.938	18.290	18.290	4.938	18.290	18.290	2.544	9.422	9.422
2	2.652	9.821	28.110	2.652	9.821	28.110	2.421	8.968	18.390
3	2.002	7.416	35.526	2.002	7.416	35.526	2.290	8.481	26.872
4	1.617	5.990	41.516	1.617	5.990	41.516	2.051	7.596	34.467
5	1.453	5.383	46.899	1.453	5.383	46.899	1.853	6.863	41.331
6	1.431	5.299	52.198	1.431	5.299	52.198	1.734	6.422	47.752
7	1.323	4.900	57.098	1.323	4.900	57.098	1.685	6.241	53.993
8	1.093	4.047	61.145	1.093	4.047	61.145	1.570	5.816	59.809
9	1.037	3.840	64.986	1.037	3.840	64.986	1.398	5.177	64.986

Note: the extraction method is principal component analysis.

. The components with variances greater than 1 sum up to 9 in total, which collectively account for 64.986% of the total variance. Extracting these 9 factors captures 64.986% of the information from the original 27 items. In Figure 5, the horizontal axis represents the number of factors, while the vertical axis represents the eigenvalues. Observing the trend of eigenvalues with the number of factors, a clear “elbow” occurs at the ninth factor, indicating that these nine factors capture the main information of all 27 question items.

Figure 6 presents the rotated component matrix, showing the correlation between each factor and the 27 environmental elements in the survey questionnaire. For each principal component, we selected items with factor loadings greater than 0.5 and summarized them to identify the main type of environmental element associated with each principal component. Subsequently, factors were classified as follows: Y1 represents the perception of greenery, Y2 represents the perception of the sky, Y3 represents the perception of industrial heritage texture, Y4 represents the perception of open space, Y5 represents the perception of regional uniqueness, Y6 represents the perception of street quality, Y7 represents the perception of overall regional appearance, Y8 represents the perception of recreational facilities, and Y9 represents the perception of industrial heritage reuse.

Table 7. Component Score Coefficient Matrix.

Attributes	Component
	Y1	Y2	Y3	Y4	Y5	Y6	Y7	Y8	Y9
Road Quality	−0.079	−0.212	0.020	0.111	−0.172	0.386	−0.069	0.076	0.169
Pedestrian Comfort	−0.128	0.143	−0.091	−0.125	−0.007	0.365	−0.065	0.056	0.089
Pedestrian Convenience	0.024	−0.041	−0.018	0.005	−0.009	0.480	−0.030	−0.100	−0.102
Pedestrian Safety	0.010	0.013	0.008	−0.140	0.076	0.348	0.070	0.030	−0.173
Quality of Industrial Heritage and its Ancillary Buildings	0.029	0.035	0.282	0.045	−0.010	−0.018	0.047	−0.143	−0.039
Historical Significance of Industrial Heritage	−0.015	0.027	0.226	−0.021	0.236	0.001	−0.043	−0.064	−0.146
Richness of Industrial Heritage Interface	0.069	0.020	0.286	−0.034	0.003	0.016	−0.117	−0.174	0.126
Protection and Restoration of Industrial Heritage	−0.049	−0.132	0.417	−0.019	−0.111	−0.059	0.056	0.118	−0.067
Coordination between Old and New Buildings	−0.018	−0.045	0.299	−0.172	−0.138	−0.002	−0.043	0.394	−0.047
Accessibility of Industrial Heritage	0.089	0.000	−0.009	0.168	−0.014	−0.002	−0.041	−0.299	0.498
Public Accessibility of Industrial Heritage	−0.059	−0.008	−0.037	−0.119	−0.037	−0.081	0.028	0.196	0.566
Quality of Plazas, Parks, and other Open Spaces	−0.062	−0.003	0.081	0.218	−0.020	0.027	0.117	0.056	−0.255
Recognizability	−0.025	0.021	0.040	0.417	−0.047	−0.056	−0.050	−0.179	0.028
Accessibility	−0.004	−0.103	−0.081	0.477	−0.023	−0.057	−0.027	−0.064	0.075
Number of Rest Facilities	0.021	−0.037	−0.109	0.169	0.097	0.016	−0.073	0.318	0.008
Feeling of Rest	0.034	0.045	−0.069	0.070	0.060	−0.015	−0.094	0.353	−0.008
Sky Feeling	0.024	0.313	−0.034	−0.043	−0.051	−0.048	−0.035	−0.012	0.084
Sky Openness	0.033	0.443	−0.036	−0.085	−0.044	−0.085	−0.054	−0.046	−0.010
Aesthetic	−0.021	0.336	−0.021	−0.007	−0.030	0.001	0.004	−0.033	−0.105
Green Feeling	0.361	0.008	−0.016	−0.067	−0.043	−0.031	−0.021	0.064	0.040
Green Space Ratio	0.367	0.008	−0.017	−0.058	−0.032	−0.037	−0.023	0.052	0.035
Distance to Green Space	0.331	0.009	0.058	0.035	0.035	−0.022	0.013	−0.240	−0.061
Appearance Feeling	−0.078	0.052	−0.058	0.002	0.038	−0.045	0.291	0.193	0.100
Overall Appearance Change in the Area	−0.037	−0.031	0.022	−0.080	−0.032	−0.089	0.581	−0.019	−0.043
Environmental Quality of the Area	0.061	−0.052	−0.024	0.052	0.007	0.079	0.407	−0.191	−0.036
Degree of Regional Features	−0.024	−0.086	−0.102	−0.038	0.455	−0.034	−0.016	0.134	0.099
Uniqueness of the Area	−0.001	−0.031	−0.032	−0.011	0.505	−0.009	0.021	−0.049	−0.120

Notes: The extraction method used was Principal Component Analysis (PCA), and the rotation method was Kaiser Normalization with Varimax rotation.

presents the component coefficient score matrix obtained from factor analysis. The values in this matrix should be multiplied with the 27 evaluation scores of the attributes, respectively, to derive scores for each factor, which ultimately consolidates into a subjective perception dataset; see Appendix A for the example formula for Y1.

3.2. Objective Imagery Indicators Based on Semantic Segmentation

Semantic segmentation of street panoramas was performed using FCN (fully convolutional networks). The segmented images, totaling 2636, were imported into ArcGIS for symbol recategorization, as illustrated in Figure 7.

Each MCP usually involves multiple street view images. The indicators of an MCP involving multiple images should be calculated based on the average values of all images within that MCP. The objective image dataset is compiled by summarizing the proportions of 20 categories of objective elements obtained directly through semantic segmentation, such as Building, Sky, Tree, etc., as well as 7 objective indicator values calculated using formulas.

Table 8. Statistical values of the objective imagery indicators.

Category	Type	Mean	Median	SD	Minimum	Maximum
Twenty directly obtained indicators	Building	0.1034	0.0973	0.0416	0.0268	0.2706
	Sky	0.2829	0.2798	0.0398	0.0728	0.3762
	Tree	0.1126	0.1105	0.0506	0.0136	0.3723
	Roadway	0.1445	0.1412	0.0280	0.0473	0.2327
	Window	0.0001	0.0000	0.0004	0.0000	0.0055
	Grass	0.0023	0.0012	0.0037	0.0000	0.0299
	Sidewalk	0.0466	0.0443	0.0178	0.0098	0.1043
	People	0.0008	0.0007	0.0009	0.0000	0.0117
	Door	0.0000	0.0000	0.0002	0.0000	0.0021
	Plant	0.0021	0.0016	0.0022	0.0000	0.0105
	Car	0.1087	0.1086	0.0270	0.0215	0.2029
	House	0.0003	0.0000	0.0010	0.0000	0.0083
	Open Field	0.0001	0.0000	0.0003	0.0000	0.0025
	Fence	0.0032	0.0028	0.0024	0.0000	0.0187
	Runway	0.0014	0.0006	0.0024	0.0000	0.0268
	Hardwood	0.0001	0.0000	0.0002	0.0000	0.0012
	Truck	0.0007	0.0005	0.0008	0.0000	0.0066
	Traffic signs	0.0001	0.0001	0.0001	0.0000	0.0014
	Small Locomotive	0.0009	0.0007	0.0009	0.0000	0.0055
	Traffic Lights	0.0001	0.0000	0.0001	0.0000	0.0009
Seven indicators obtained through calculation.	SFI	0.3482	0.3183	0.1859	0.0498	1.7228
	VII	0.1159	0.1166	0.0272	0.0297	0.2129
	TSI	0.0002	0.0001	0.0002	0.0000	0.0014
	AIH	0.1120	0.1069	0.0496	0.0268	0.4935
	OSI	0.0481	0.0455	0.0180	0.0098	0.1162
	GVI	0.1171	0.1150	0.0521	0.0148	0.3723
	SVI	0.2829	0.2798	0.0398	0.0728	0.3762

presents 20 directly obtained indicators and 7 indicators obtained through calculation.

3.3. The Influence of Subjective Image Factors on Residents’ Satisfaction

We conducted a collinearity test using SPSS before performing multiple regression analyses. We used the Variance Inflation Factor (VIF) and Tolerance (TOL) as indicators. A VIF value greater than 10 or a TOL value less than 0.1 suggests potential multicollinearity. In our analysis, all VIF values were below 10, and all TOL values were above 0.1, indicating no significant multicollinearity among the independent variables. Multiple regression analysis was conducted with the nine perceptual factors as independent variables and residents’ satisfaction evaluations obtained from the survey questionnaire as the dependent variable. The aim was to determine the impact of residents’ subjective perceptual factors on their satisfaction levels.

The multiple regression results show that the p-values for resident satisfaction with the 9 image factors are all below 0.05, indicating a strong correlation between these factors and satisfaction, with all correlation coefficients being positive. Based on the unstandardized coefficient B, it can be observed that the three factors Y7, Y8, and Y5 have the greatest impact on overall resident satisfaction, while the three factors Y4, Y6, and Y1 have the least impact, as depicted in Figure 8a. In Figure 8b, |Beta| represents the absolute value of the standardized coefficient, which is used to assess the influence of various objective spatial indicators on resident satisfaction in subsequent analyses.

3.4. Impact of Objective Spatial Indicators on Residents’ Subjective Perception Factors

To identify the key objective spatial factors influencing the perception of each image factor, we conducted multiple regression analyses. Before conducting multiple regression, a collinearity test was conducted, and the results showed that there was no significant multicollinearity relationship between the independent variables. Given the potential mutual influence among subjective perception factors, in each regression analysis, 1 of the 9 image factors (Y1–Y9) was used as the dependent variable, while the remaining 8 factors, along with the 27 objective indicators, were used as independent variables. Finally, the impact of objective spatial indicators on residents’ subjective perception factors was obtained, as illustrated in Figure 9.

The results indicate that Y2, Tree, GVI, Grass, SFI, and Plant have significant positive effects on Y1, while Traffic Light has a significant negative effect on Y1. Y1, Y4, Sky, SVI, Grass, and OSI have significant positive effects on Y2. Building, Gate, and Window have significant positive effects on Y3, while AIH, Car, Small Locomotive, and Trail have significant negative effects on Y3. Y2, Sidewalk, People, and OSI have relatively significant positive effects on Y4, while Traffic Sign and Car have significant negative effects on Y4. Roadway and Hardwood have significant positive effects on Y5, while Trail and Small Locomotive have significant negative effects on Y5. Roadway, Sidewalk, People, Car, TSI, and SFI have relatively significant positive effects on Y6, while VII has a significant negative effect on Y6. House and Fence have significant positive effects on Y7. Open Field has a significant positive effect on Y8, while People have a significant negative effect on Y8. People, Car, Small Locomotive, and VII have significant positive effects on Y9.

3.5. The Capacity of Objective Spatial Indicators in Influencing Residential Satisfaction

Standardized coefficients, compared to unstandardized coefficients, eliminate the influence of units of measurement and can, therefore, be used for calculations. Figure 10 presents the absolute standardized coefficients |Beta|_OVS obtained from multiple regression analysis between objective spatial indicators and residents’ subjective perception factors, while Figure 8b illustrates the absolute standardized coefficients |Beta|_SVS obtained from multiple regression analysis between residents’ subjective perception factors and residential satisfaction.

Multiplying |Beta|_OVS and |Beta|_SVS yields the ability ∂ of each objective spatial indicator to influence resident satisfaction. The formula is as follows, and the results are shown in

Table 9. Impact Capacity of Objective Spatial Indicators in Influencing Residential Satisfaction.

Perceptual Factor	Objective Index	Objective Index|Beta|	Overall Satisfaction|Beta|	Objective Index Satisfaction Score
Y1	Tree	0.209	0.288	0.060
	GVI	0.464		0.134
	Grass	0.626		0.180
	SFI	0.113		0.033
	Plant	0.112		0.032
	Traffic Light	0.079		0.023
Y2	Sky	0.392	0.261	0.102
	SVI	0.334		0.087
	Grass	0.219		0.057
	OSI	0.226		0.059
Y3	Building	0.187	0.237	0.044
	Gate	0.201		0.048
	Window	0.150		0.036
	AIH	0.176		0.042
	Car	0.163		0.039
	Small Locomotive	0.141		0.033
	Trail	0.176		0.042
Y4	Sidewalk	0.151	0.165	0.025
	People	0.135		0.022
	Car	0.138		0.023
	OSI	0.115		0.019
	Traffic Sign	0.681		0.112
Y5	Trail	0.150	0.362	0.054
	Small Locomotive	0.166		0.060
	Roadway	0.169		0.061
	Hardwood	0.143		0.052
Y6	Roadway	0.275	0.162	0.045
	Sidewalk	0.150		0.024
	People	0.136		0.022
	Car	0.210		0.034
	TSI	0.136		0.022
	SFI	0.169		0.027
	VII	0.422		0.068
Y7	House	0.277	0.286	0.079
	Fence	0.161		0.046
Y8	People	0.588	0.304	0.179
	Open Field	0.771		0.234
Y9	People	0.126	0.279	0.035
	Car	0.271		0.076
	Small Locomotive	0.130		0.036
	VII	0.177		0.049

.

∂ = |Beta|_OVS × |Beta|_SVS

(2)

Figure 11 shows the top 10 objective spatial indicators that influence resident satisfaction: People, Grass, Open Field, Car, GVI, Small Locomotive, VII, Traffic Sign, Roadway, and Sky. These indicators are classified as primary impact indicators due to their high rankings. The bottom 6 objective spatial indicators, classified as tertiary impact indicators, are Building, AIH, Window, Plant, Traffic Light, and TSI. The remaining indicators are classified as secondary impact indicators.

4. Discussion

This study integrates image semantic segmentation techniques and multiple regression methods to align residents’ subjective perceptions with objective activity spaces. On one hand, it analyzes the impact of residents’ subjective perceptual factors on satisfaction. On the other hand, it examines the influence of urban objective spatial indicators on residents’ subjective perceptual factors. Furthermore, it integrates both subjective and objective aspects to rank the capabilities of objective spatial indicators in affecting resident satisfaction.

This study acknowledges several limitations. Firstly, urban image perception scores were obtained through survey questionnaires, reflecting the overall urban image level based on various subelements. However, due to factors such as residents’ perceptions, geographical locations, psychological states, and building types, there are inherent limitations in this approach. Given the specific socio-economic backdrop of the post-industrial neighborhoods under study—areas that thrived during the Northeast’s industrial peak but now face economic downturns—the applicability of our findings to other contexts is not guaranteed. Secondly, sample selection may be subject to potential biases as it might not include certain groups, such as individuals who spend long periods at home. Thirdly, during the acquisition of urban street view data, uncertainties in the data collection time by Baidu’s open platform led to issues in image semantic segmentation, resulting in errors in some indicators. For example, images captured in summer may have more foliage compared to those captured in winter, leading to seasonal biases. Future research could utilize street view collection tools to manually collect images during the same season to reduce semantic segmentation errors. Fourthly, this study focused mainly on post-industrial neighborhoods with specific socio-economic backgrounds when selecting subjective and objective indicators, thus limiting its applicability to other contexts. Additionally, despite an extensive literature review and inclusion of numerous indicators, there remains a possibility of overlooking certain factors. In the regression analysis between objective spatial indicators and subjective perceptual perceptions, some individual results showed relatively small R² values. Future studies could explore alternative semantic segmentation methods to enhance the accuracy of objective indicators’ representation.

This study innovatively introduces the concept of residents’ activity spaces. Using survey questionnaires, we obtained the daily activity locations of residents and connected these locations to generate the minimum convex polygon as each resident’s daily activity space. Street view images were then obtained within this space. Compared to the conventional method of obtaining a large number of urban street views based on administrative boundaries [24,69], this approach more accurately represents the street views seen by residents in their daily lives, thus, further enhancing the experimental accuracy.

The impact of residents’ subjective perceptual factors on resident satisfaction is as follows. There is a strong positive correlation between subjective image factors and residents’ satisfaction. Residents are particularly concerned about changes in the overall landscape of the area [49,70], consistent with early research findings. Additionally, our study revealed that residents place high importance on the quality of recreational facilities and the uniqueness of the area. These concerns are closely linked to a sense of place identity, reflecting residents’ opposition to homogenous urban redevelopment and their desire to maintain the area’s original characteristics while enhancing the quality of everyday leisure activities. Past research emphasized the importance of the quality of industrial heritage [7,71]. In addition to this, we found that residents are also highly concerned with the reuse of industrial heritage and their perception of the sky. These concerns underscore the uniqueness of post-industrial historic sites. Some residents, having participated in industrial production, are particularly interested in industrial heritage. Proper protection and utilization of industrial heritage can help preserve residents’ historical memories. The sky, as a compositional backdrop for urban architecture, directly influences residents’ moods through their perception of it. However, the perception of open spaces, street quality, and greenery has the least impact on residents’ satisfaction.

Three factors influence residents’ perception of regional elements in the urban image: the perception of greenery, the perception of the sky, and the perception of the overall regional landscape. The interpretation of the results for these three factors reveals that four indicators are directly related to the composition of greenery, showing a positive correlation with greenery perception. This finding is consistent with the studies conducted by Smith et al. [41] and Li et al. [72]. Our research also discovered that greenery perception is correlated with sky perception, road surface feasibility, and traffic signs. The correlation between sky perception and green space perception is due to the fact that in more open environments, the green space ratio is typically higher, and the sense of sky openness is also stronger [73]. Locations with higher road surface feasibility tend to have more street trees and shrubs, thereby enhancing the perception of greenery. Traffic signs are typically located in paved areas, hence showing a negative correlation with greenery perception. SVI are direct components of sky perception, showing a positive correlation with sky perception, which aligns with earlier research findings [38,46]. The positive correlation between the perception of open space, the OSI, and sky perception is due to the relatively low presence of high-rise buildings in open areas such as parks and squares, resulting in a greater sense of sky openness. The adjusted R² value of the multiple regression results for the perception of the overall regional landscape is relatively low. Only the indicators of houses and fences show a positive correlation with the perception of the overall regional landscape. This indicates that houses and certain architectural elements have a positive impact on the perception of the overall regional landscape [68].

Since industrial heritage is an important component of post-industrial historical sites, the perception of industrial heritage quality and the perception of industrial heritage reuse are classified as factors related to landmarks. The interpretation of the results for these two factors indicates that indicators such as buildings, doors, and windows are direct components of industrial heritage quality, showing a positive correlation with the perception of industrial heritage quality. This finding aligns with the conclusions of earlier studies [26]. Additionally, we found that AIH (Artificial Intelligence Heritage), cars, small motorcycles, and pathways are negatively correlated with the perception of industrial heritage quality. The reason is that these indicators disrupt the historical environment, leading to a decrease in the perceived quality of industrial heritage. Based on the analysis results, the greater the number of tourists and vehicles near industrial heritage sites, the higher the accessibility [74] and publicness of the industrial heritage. Therefore, the presence of people, cars, small motorcycles, and the vehicle interference index are positively correlated with the perception of industrial heritage reuse.

Open spaces and their internal rest facilities are provided for people’s leisure and recreation. These locations are generally arranged as important nodes within the city. The interpretation of the results for these two factors indicates that sky perception, sidewalks, people, and the OSI are direct components of open space, showing a positive correlation with the perception of open space. This finding corroborates previous research results [34]. Additionally, we found that cars and traffic signs can degrade the environment of open spaces such as parks and squares, thereby diminishing residents’ recreational experience. Thus, these indicators are negatively correlated with the perception of open space. Similar to the perception of the overall regional landscape, only two variables are highly correlated with the perception of rest facilities: people and open spaces. Given that the number of rest facilities remains constant within a certain space, an increase in the number of people leads to fewer rest facilities per capita, resulting in a negative correlation between the number of people and the perception of rest facilities. Conversely, more open spaces indicate a higher total number of rest facilities, thereby showing a positive correlation between open spaces and the perception of rest facilities [36].

The perception of street quality is positively correlated with indicators such as roadway, sidewalk, people, cars, TSI, and SFI; higher values in these indicators correspond to better street quality [23]. Conversely, the VII, the lower the perceived street safety, leading to a negative correlation between VII and the perception of street quality [22].

Compared to previous studies on urban boundaries that were limited to administrative divisions, we found that boundaries can be described by internal homogeneity and external heterogeneity. For example, each region has its unique species of street trees. In the urban area of Shenyang, elm trees are commonly planted, making broadleaf trees a good indicator of regional uniqueness. Additionally, roadways, especially highways, are generally located on the outskirts of the city, reflecting the regional boundary. Thus, these two indicators are positively correlated with the perception of regional uniqueness. In contrast, internal landscapes such as small motorcycles and pathways are generally similar across different cities and typically do not delineate the city’s boundary. Regression results show that these factors are negatively correlated with the perception of regional uniqueness.

Integrating both subjective and objective influences, we have ranked the ability of objective spatial indicators to affect resident satisfaction. Grass, Open Fields, the Green Visual Index (GVI), and Sky rank in the top 10. This is likely because the population in these areas is predominantly middle-aged and elderly, who frequently use and value green spaces and open spaces. As a result, they pay significant attention to open spaces, greenery levels, and sky openness. Additionally, People, Cars, Small Locomotives, Traffic Signs, Roadways, and the Vehicle Interference Index (VII) also rank highly. This may be because post-industrial historical communities are primarily inhabited by former factory workers and their families. After the factories closed, work locations changed, leading to longer commuting distances, making road safety a significant concern. Although factors such as pedestrian flow, traffic flow, and the protection and utilization of industrial heritage are important, some minor details have a smaller impact on satisfaction. These include Traffic Lights, the Traffic Sign Index (TSI), the Accessibility of Industrial Heritage (AIH), Buildings, and Windows. Plants also rank lower, indicating that while sufficient greenery significantly improves residents’ satisfaction, the specific types of plants are not a major concern. While this study prioritizes these factors based on their influence, practical urban renewal efforts should also consider specific local conditions. For instance, factors with high influence that currently meet residents’ needs may not require additional resources for optimization. Conversely, factors with lower influence but severe existing issues should be prioritized for limited resources and attention during urban renewal practices.

This study contributes to sustainable development by offering targeted strategies for the revitalization of post-industrial communities. By prioritizing renewal efforts based on the impact of objective spatial indicators on resident satisfaction, we can promote the effective preservation and adaptive reuse of historical environments. These findings support ongoing discussions on sustainable urban development and cultural heritage preservation, emphasizing the importance of balancing modern needs with the conservation of historical character, ultimately fostering more resilient and sustainable urban communities.

5. Conclusions

This study uses Shenyang, China, as a case study area and integrates image semantic segmentation technology and multiple regression methods to align residents’ subjective perceptions with objective activity spaces. From the perspective of urban imagery, it analyzes how subjective image factors among residents in post-industrial communities affect satisfaction (see Figure 8 for details). Additionally, it examines how urban objective spatial indicators influence these subjective image factors (see Figure 9 for details). By combining findings from both subjective and objective perspectives, the study further ranks the ability of objective spatial indicators to affect resident satisfaction (see Figure 11 for details).

The study identifies nine subjective perception factors significantly affecting residents’ satisfaction, all positively correlated. The top three factors influencing satisfaction are the overall perception of the area’s appearance, uniqueness of the area, and availability of resting facilities. Additionally, it identifies 27 objective spatial indicators significantly affecting residents’ subjective perception factors. For instance, traffic lights are negatively correlated with the perception of greenery (β = −0.079), while grass is positively correlated (β = 0.626).

By multiplying the absolute values of Beta coefficients (|Beta|_OVS and |Beta|_SVS) corresponding to each objective variable, a ranking of their impact on residents’ satisfaction is obtained. Factors such as pedestrian flow, vehicular flow, open space, sky openness, and perceived level of greenery have a significant impact on residents’ satisfaction, whereas the public nature of industrial heritage, industrial heritage preservation and utilization, and quality of traffic signs have smaller impacts. This study reveals the potential impact of objective environmental indicators on residents’ satisfaction, which helps determine the priorities of various tasks in urban renewal, it can efficiently enhance residents’ quality of life during urban renewal. Additionally, the combination of subjective and objective methods proposed in this study provides a new approach for the preservation of the character of similar historical built environments.