A method and system for analyzing architectural decor images

Abstract

The application relates to the technical field of image analysis, and discloses a building decoration image analysis method and system, which extracts and quantifies an aesthetic feature vector from overall visual performance by performing visual analysis and processing on an input building decoration image. Subsequently, an association database of the aesthetic feature vector and decoration element labels is constructed, and a nearest neighbor matching is performed in the database according to the extracted aesthetic feature vector to obtain the aesthetic feature vector of the closest professional gallery image. Finally, a decoration element combination rule is inversely solved according to the nearest neighbor matching result, and a corresponding decoration element combination suggestion is generated. The application aims to break through the limitations of traditional methods on discrete element recognition and co-occurrence frequency statistics by deeper analysis of aesthetic features, so as to better capture the redefinition of the "relationship" between elements in the evolution of design styles and the changes in the "overall atmosphere" created thereby, and provide designers with more forward-looking and personalized design references.

Description

Technical Field

[0001] This invention relates to the field of image analysis technology, and in particular to a method and system for analyzing architectural decoration images. Background Technology

[0002] In the field of interior design, auxiliary systems analyze a large number of architectural and decorative images to extract patterns in the combination of decorative elements, providing designers with design references. These systems typically collect and process professional interior design images, using image recognition technology to convert visual information into data, and then statistically analyze the co-occurrence frequency of different decorative elements to summarize universally applicable combination rules. For example, the system might discover that "Nordic style sofas" are often paired with "light-colored wood-grain flooring" and "green plants." When designers input their requirements, the system generates preliminary design schemes or material lists based on these rules, providing efficient data support in the form of renderings.

[0003] However, this analytical method, which relies on past data and fixed patterns, increasingly reveals its limitations when faced with ever-changing design trends. Design styles are not static but evolve with time, culture, and aesthetic concepts. For example, today's "modern minimalism" may incorporate more warmth from natural materials and environmental considerations, while systems based on old data often only repeat classic combinations from the past, failing to capture these subtle changes in design trends. When designers attempt to explore emerging styles, such as "wabi-sabi," the system-generated solutions may appear out of place, lacking a sense of fashion and understanding of new aesthetic concepts. They may simply be a patchwork of elements, failing to reflect the style's unique use of material texture, spatial white space, and the interplay of light and shadow, as well as its deep pursuit of the beauty of imperfection. Summary of the Invention

[0004] This invention provides a method and system for analyzing architectural decoration images, addressing the technical problems of existing interior design support systems in capturing subtle design trend changes and understanding deeper design intentions when faced with constantly evolving design trends. Furthermore, these systems suffer from low accuracy in processing non-standard data, leading to chaotic and unpredictable analysis results. Firstly, in order to solve the above-mentioned technical problems, the present invention provides a method for analyzing architectural decoration images, comprising: Visual analysis processing is performed on the input architectural decoration image to extract and quantify aesthetic feature vectors from the overall visual performance of the architectural decoration image. The aesthetic feature vectors include color saturation distribution, texture complexity, light softness and contrast, spatial white space ratio, and line and shape features. Construct an association database between the aesthetic feature vectors and decorative element annotations, wherein the association database stores the aesthetic feature vectors of professional image library images and their corresponding decorative element annotations; Based on the aesthetic feature vector, nearest neighbor matching is performed in the association database to obtain the aesthetic feature vector of the professional image library image that is closest to the aesthetic feature vector; Based on the nearest neighbor matching results, the rules for combining decorative elements are deduced in reverse. The deduction of the rules for combining decorative elements includes analyzing the decorative elements and their combination methods that are common to images in the nearest professional image library; and generating decorative element combination suggestions corresponding to the rules for combining decorative elements.

[0005] Preferably, the step of performing visual analysis processing on the input architectural decoration image, extracting and quantifying aesthetic feature vectors from the overall visual representation of the architectural decoration image, includes: Adaptive brightness and contrast adjustment is performed on the architectural decoration image; Semantic segmentation is performed on the adjusted architectural decoration image to separate the main physical structure regions in the adjusted architectural decoration image; Extract the central skeleton line for each of the main physical structure regions; Within the main physical structure region segmented by the semantic segmentation, the color values ​​of pixels are extracted, and the relative color relationships between the main color clusters are calculated to generate the aesthetic feature vector. The aesthetic feature vectors are normalized.

[0006] The step of extracting pixel color values ​​within the main physical structure region segmented by semantic segmentation and calculating the relative color relationships between main color clusters includes: The distribution characteristics of pixel color values ​​within the main physical structure region segmented by the semantic segmentation are evaluated. The distribution characteristics evaluation includes calculating the number of histogram peaks of the pixel color values ​​and the distance between adjacent peaks. The number of main color clusters is dynamically determined based on the distribution characteristics assessment results; The primary color clusters are identified based on the number of primary color clusters; Among the identified primary color clusters, the relative color relationships between the primary color clusters are calculated.

[0007] Preferably, the adaptive brightness and contrast adjustment of the architectural decoration image includes: The architectural decoration image is separated into illumination and reflection components to obtain the illumination component and the reflection component. The illumination component is dynamically compressed to obtain the dynamically compressed illumination component. The reflection component is enhanced with detail to obtain the enhanced reflection component. The dynamic range compressed illumination component is fused with the detail enhanced reflection component to obtain the adjusted architectural decoration image.

[0008] Preferably, the step of performing semantic segmentation on the adjusted architectural decoration image to separate the main physical structure regions in the adjusted architectural decoration image includes: Multi-scale feature extraction is performed on the adjusted architectural decoration image to obtain visual information of the adjusted architectural decoration image at different resolutions; The multi-scale features are fused to generate a feature representation that includes contextual information; Based on the fused feature representation, the pixels of the adjusted architectural decoration image are classified to identify and separate the main physical structure regions; Based on the classification results, a segmentation mask for the main physical structure region is generated.

[0009] Preferably, after classifying the pixels of the adjusted architectural decoration image based on the fused feature representation, the method further includes: In the pixel classification process, structural boundary smoothing constraints are introduced to suppress responses to structural surface textures, patterns, or digital enhancement effects; The pixel classification process introduces structural boundary smoothing constraints to suppress responses to structural surface textures, patterns, or digital enhancement effects, including: The pixel classification results are divided into local regions to identify physical structure regions with different boundary smoothness requirements; For each of the physical structure regions, the local gradient distribution characteristics of the boundary are calculated, whereby the local gradient distribution characteristics represent the boundary sharpness or boundary ambiguity of the physical structure region. Based on the local gradient distribution characteristics, the constraint strength for the smoothness of the structural boundary of each physical structural region is dynamically determined. The constraint strength is negatively correlated with the boundary sharpness and positively correlated with the boundary ambiguity.

[0010] Preferably, extracting the central skeleton line for each of the main physical structure regions includes: Geometric analysis based on the shape characteristics of the main physical structure region is performed to identify irregular boundary segments and complex geometric features in the main physical structure region; Based on the irregular boundary segments and the complex geometric features, the local smoothness parameters extracted from the skeleton lines are dynamically adjusted. These local smoothness parameters are used to control the sensitivity of the skeleton lines to local details. During the skeleton line extraction process, a local smoothness parameter is applied to the irregular boundary segment to suppress interference with surface details; The distance transformation is performed on the main physical structure region, and the central skeleton line is obtained based on the distance transformation result and the dynamically adjusted local smoothness parameter.

[0011] Preferably, the step of performing geometric analysis based on the shape features of the main physical structure region to identify irregular boundary segments and complex geometric features in the main physical structure region includes: Multi-scale curvature analysis is performed on the boundary of the main physical structure region to obtain information on the curvature variation of the boundary of the main physical structure region at different scales. The curvature variation information at different scales is fused to generate a curvature feature representation that includes local details and global trends; Based on the fused curvature feature representation, an adaptive threshold is set to distinguish geometric noise from the irregular boundary segment or the complex geometric feature; Based on the differentiation results, the irregular boundary segments and the complex geometric features are identified.

[0012] Preferably, the step of performing a distance transformation on the main physical structure region and obtaining the central skeleton line based on the distance transformation result and the dynamically adjusted local smoothness parameter includes: The distance transformation is performed on the main physical structure region to obtain the distance transformation result; Based on the distance transformation result and the dynamically adjusted local smoothness parameter, the central skeleton line is obtained, which includes the connectivity and hole structure of the main physical structure region.

[0013] Secondly, the present invention provides an architectural decoration image analysis system, comprising: The input end is used to perform visual analysis processing on the input architectural decoration image, extract and quantify the aesthetic feature vector from the overall visual performance of the architectural decoration image, and the aesthetic feature vector includes color saturation distribution, texture complexity, light softness and contrast, spatial white space ratio, and line and shape features. The construction end is used to construct an association database of the aesthetic feature vectors and decorative element annotations. The association database stores the aesthetic feature vectors of professional image libraries and their corresponding decorative element annotations. Based on the aesthetic feature vectors, nearest neighbor matching is performed in the association database to obtain the aesthetic feature vectors of the professional image libraries that are closest to the aesthetic feature vectors. The matching end is used to reverse-solve the decorative element combination rules based on the nearest neighbor matching results. The reverse-solved decorative element combination rules include analyzing the decorative elements and their combination methods common to the images in the nearest professional image library; and generating decorative element combination suggestions corresponding to the decorative element combination rules.

[0014] This application proposes a method and system for analyzing architectural decoration images. By performing visual analysis on input architectural decoration images, aesthetic feature vectors are extracted and quantified from the overall visual presentation. These aesthetic feature vectors include color saturation distribution, texture complexity, light softness and contrast, spatial white space ratio, and line and shape features. Subsequently, a database linking aesthetic feature vectors with decorative element annotations is constructed. Based on the extracted aesthetic feature vectors, nearest neighbor matching is performed in the database to obtain the aesthetic feature vectors of the closest professional image library images. Finally, the rules governing the combination of decorative elements are deduced based on the nearest neighbor matching results, and corresponding suggestions for combining decorative elements are generated. This application aims to overcome the limitations of traditional methods in identifying discrete elements and calculating co-occurrence frequencies through deeper aesthetic feature analysis. This allows for a better capture of the redefinition of the "relationships" between elements in the evolution of design styles and the resulting changes in the "overall atmosphere," providing designers with more forward-looking and personalized design references. Attached Figure Description

[0015] Figure 1 This is a flowchart of an architectural decoration image analysis method provided by an embodiment of the present invention; Figure 2 This is a flowchart of another architectural decoration image analysis method provided in an embodiment of the present invention; Figure 3 This is a flowchart of another architectural decoration image analysis method provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an architectural decoration image analysis system provided in an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Reference Figure 1 The present invention provides a flowchart of a method for analyzing architectural decoration images, comprising the following steps: S1, Perform visual analysis processing on the input architectural decoration image, extract and quantify the aesthetic feature vector from the overall visual performance of the architectural decoration image, the aesthetic feature vector includes color saturation distribution, texture complexity, light softness and contrast, spatial white space ratio, and line and shape features. S2, construct the association database of the aesthetic feature vector and the decorative element annotation, the association database stores the aesthetic feature vector of professional image library images and their corresponding decorative element annotations; S3, based on the aesthetic feature vector, perform nearest neighbor matching in the association database to obtain the aesthetic feature vector of the professional image library image that is closest to the aesthetic feature vector; S4. Based on the nearest neighbor matching results, reverse the rules of decorative element combination. The reverse rules of decorative element combination include analyzing the decorative elements and their combination methods that are common to the images in the nearest professional image library; and generating decorative element combination suggestions corresponding to the rules of decorative element combination.

[0018] The "aesthetic feature vector" mentioned in this application is a quantitative representation used to describe the visual aesthetic attributes of architectural decorative images. It includes multiple dimensions, such as "color saturation distribution," which describes the vibrancy of colors in the image and their distribution; "texture complexity," which quantifies the fineness and repetition patterns of textures in the image; "light softness and contrast," which assesses the smoothness of light and shadow transitions and the degree of difference between bright and dark areas in the image; "white space ratio," which measures the proportion of blank areas not occupied by major decorative elements in the image, which is usually related to the sense of breathing room and balance in the design; and "line and shape features," which describe the direction, thickness, and geometric composition of lines in the image. These features together constitute a multi-dimensional vector that can comprehensively and objectively reflect the overall aesthetic style of the image.

[0019] The "Associated Database" is a storage structure that contains aesthetic feature vectors of a large number of professional image libraries and their corresponding "decorative element annotations." "Decorative element annotations" refer to the classification and description of specific decorative elements (such as sofas, coffee tables, lamps, wall materials, etc.) appearing in the images. By associating aesthetic feature vectors with decorative element annotations, the database can establish a mapping relationship between visual aesthetics and specific design elements.

[0020] "Nearest neighbor matching" is a data retrieval technique that aims to find one or more vectors in a database that are closest to the query vector (i.e., the aesthetic feature vector extracted from the input architectural and decorative image) in the feature space. Distance can be measured using various methods such as Euclidean distance and cosine similarity. Through nearest neighbor matching, the system can identify professional image libraries that are most similar to the input image in aesthetic style.

[0021] "Inverse analysis of decorative element combination patterns" refers to in-depth analysis of similar images after obtaining nearest-neighbor matching results to reveal their common decorative elements and their combination methods. This is not merely about simply counting the co-occurrence frequency of elements, but more importantly, understanding how these elements interact to create a specific aesthetic style. Ultimately, "decorative element combination suggestions" provide users with specific design guidance based on the inverse analysis of patterns, such as recommending specific styles of furniture, color schemes, or material selections.

[0022] In practical implementation, the architectural decoration image analysis method of this application can be carried out according to the following steps: First, the input architectural decoration images undergo visual analysis to extract and quantify aesthetic feature vectors from the overall visual presentation of the images. For example, deep learning models, such as convolutional neural networks (CNNs), can be used for feature extraction. These models can be pre-trained on a large dataset of labeled architectural decoration images, enabling them to learn and recognize visual aesthetic features within the images. During processing, the image can be input into a pre-trained feature extraction network, which outputs a high-dimensional vector—the aesthetic feature vector. This aesthetic feature vector can encompass multiple dimensions, including color saturation distribution, texture complexity, light softness and contrast, spatial white space ratio, and line and shape features. For example, color saturation distribution can be obtained by calculating the mean and variance of saturation in different color regions of an image; texture complexity can be quantified using methods such as Gabor filter banks or Local Binary Patterns (LBP); light softness and contrast can be evaluated by analyzing the image's brightness histogram and edge information; the proportion of white space can be obtained by identifying the main object region using image segmentation techniques and then calculating the pixel ratio of non-object regions; and line and shape features can be extracted from the image using Hough transform or edge detection algorithms and then quantified.

[0023] Secondly, an association database of aesthetic feature vectors and decorative element annotations is constructed. This database can be a structured database, such as a relational database or a NoSQL database. During construction, a large number of professional image libraries can be collected, and each image can be manually or semi-automatically annotated with decorative elements. For example, for a picture of a Nordic-style living room, decorative elements such as "light wood grain flooring," "gray fabric sofa," "simple chandelier," and "green plants" can be annotated. Simultaneously, visual analysis processing is performed on these professional image libraries to extract and quantify their aesthetic feature vectors. Then, the aesthetic feature vector of each professional image library image and its corresponding decorative element annotation are stored as a record in the association database. For example, a vector database can be used to store the aesthetic feature vectors, and a relational database can be used to store the decorative element annotations, linking the two through a unique identifier.

[0024] Next, based on the extracted aesthetic feature vector, nearest neighbor matching is performed in the associated database to obtain the aesthetic feature vectors of the professional image library images that are closest to the extracted aesthetic feature vector. Various similarity metrics can be used for nearest neighbor matching, such as Euclidean distance, cosine similarity, or Manhattan distance. For example, when a new architectural decoration image is input, its aesthetic feature vector is first extracted. Then, this vector is compared with the aesthetic feature vectors of all professional image library images in the associated database, and the distance or similarity between them is calculated. Finally, the aesthetic feature vectors of the K professional image library images with the smallest distance (or highest similarity) are selected as the nearest neighbor matching results. To improve matching efficiency, Approximate Nearest Neighbor (ANN) algorithms can be used, such as KD-trees, Locality Sensitive Hashing (LSH), or the Faiss library.

[0025] Finally, based on the nearest neighbor matching results, the rules governing the combination of decorative elements are deduced, and corresponding suggestions for combining decorative elements are generated. After obtaining the K most similar professional image library images, these images can be further analyzed. Deducing the rules governing the combination of decorative elements involves analyzing the decorative elements and their combinations common to the neighboring professional image library images. For example, the most frequently occurring decorative elements in these K images can be statistically analyzed, along with the common pairing relationships between these elements. For instance, if it is found that "wooden dining tables" and "wicker dining chairs" frequently appear together in these K images, and are usually paired with "warm-toned pendant lights," then this can be considered a rule governing the combination of decorative elements. Based on this, suggestions for combining decorative elements corresponding to the rules governing the combination of decorative elements are generated. For example, based on the deduced rules, suggestions can be provided to users such as "It is recommended to use a wooden dining table paired with wicker dining chairs, supplemented by warm-toned pendant lights, to create a natural and warm dining atmosphere." These suggestions can be presented in the form of text descriptions, image examples, or 3D models.

[0026] This application effectively solves the problem that traditional architectural decoration image analysis methods struggle to capture subtle design trend changes and understand deeper design intentions and aesthetic principles when faced with constantly evolving design trends. Traditional methods rely too heavily on the identification of discrete "elements" and the statistics of co-occurrence frequencies, resulting in solutions that often lag behind trends and lack personalization and foresight.

[0027] Compared to existing technologies, the core innovation of this application lies in its extraction and quantification of "aesthetic feature vectors" from the "overall visual representation" of architectural decorative images, and using this as a basis for matching and pattern deduction. Traditional methods may only be able to identify "L-shaped sofas" and "round coffee tables" and statistically analyze their frequent co-occurrence. However, this application, by extracting aesthetic dimensions such as color saturation distribution, texture complexity, light softness and contrast, spatial white space ratio, and line and shape characteristics, can capture "deep design intentions" or "aesthetic principles" that transcend individual element categories and simple co-occurrence frequencies. For example, when designers attempt to explore "wabi-sabi style," traditional systems may only be able to simply piece together some "old" element combinations, failing to understand the unique use of material texture, spatial white space, and light and shadow variations in "wabi-sabi style," as well as the deep pursuit of imperfect beauty.

[0028] This application quantifies these abstract aesthetic features, enabling the system to understand the redefinition of the "relationships" between elements and the resulting changes in the "overall atmosphere." For example, by analyzing the softness and contrast of light, the system can distinguish between the bright and transparent modern style and the soft and tranquil wabi-sabi style; by analyzing the proportion of white space, the system can understand the use of "emptiness" in minimalism or wabi-sabi styles. This nearest-neighbor matching based on aesthetic feature vectors can more accurately find professional image libraries that are similar to the input image in "spirit," thereby deciphering the rules governing the combination of decorative elements that truly conform to new trends.

[0029] Therefore, this application can generate more forward-looking and personalized suggestions for combinations of decorative elements. These suggestions not only include specific decorative elements, but more importantly, provide guidance on how to combine these elements to achieve specific aesthetic effects. This makes the system more adept at handling cutting-edge design needs, significantly improving designers' work efficiency and satisfaction, and providing the interior design field with a smarter and more contemporary auxiliary tool.

[0030] In some embodiments of this application, reference is made to Figure 2 S1 includes: S11, Adaptive brightness and contrast adjustment is performed on the architectural decoration image; S12, perform semantic segmentation on the adjusted architectural decoration image to separate the main physical structure regions in the adjusted architectural decoration image; S13, extract the central skeleton line for each of the main physical structure regions; S14, within the main physical structure region segmented by semantic segmentation, extract the color values ​​of pixels and calculate the relative color relationships between the main color clusters to generate the aesthetic feature vector; S15, normalize the aesthetic feature vector.

[0031] Specifically, adaptive brightness and contrast adjustments are performed on architectural and decorative images to optimize their visual quality and make them more suitable for subsequent analysis and processing. This adjustment process dynamically adjusts the image based on its overall brightness distribution and contrast to avoid overexposure or underexposure and enhance the image's detail.

[0032] Semantic segmentation of the adjusted architectural decoration image to separate the main physical structure regions involves classifying each pixel in the image into a predefined semantic category, such as walls, floors, ceilings, and furniture, thereby identifying the main physical structures in the image. Semantic segmentation allows for precise definition of the boundaries between different decorative elements, providing structured regional information for subsequent feature extraction.

[0033] Furthermore, extracting the central skeleton line for each major physical structure region refers to using image processing algorithms to extract skeleton lines representing the shape and topology of each separated physical structure region. The central skeleton line simplifies complex shapes, captures their core geometric features, and aids in the analysis of line and shape characteristics.

[0034] Extracting pixel color values ​​and calculating the relative color relationships between major color clusters within the main physical structural regions identified through semantic segmentation refers to analyzing the color information of pixels within each identified physical structural region. This involves identifying major color clusters in the image using methods such as clustering, and further analyzing the relative relationships of hue, saturation, and brightness among these color clusters to quantify the aesthetic feature of color saturation distribution.

[0035] Finally, normalization of the aesthetic feature vectors refers to unifying the extracted aesthetic feature vectors (such as color saturation distribution, texture complexity, light softness and contrast, spatial white space ratio, and line and shape features) to the same numerical range or scale. Normalization can eliminate differences in units and numerical ranges between different features, ensuring that each feature has the same weight in subsequent matching and analysis, thereby improving the accuracy and stability of the analysis.

[0036] This application's solution refines the original architectural decoration image analysis process into a series of ordered sub-steps, achieving precise extraction and quantification of aesthetic feature vectors. First, the adaptive brightness and contrast adjustment step ensures optimal visual analyzability of the image, laying the foundation for subsequent feature extraction. Second, the semantic segmentation step decomposes the image into regions with clear physical meaning, allowing subsequent feature extraction to be targeted at specific structures, avoiding potential confusion from global analysis. By extracting the central skeleton line, the line and shape features in the architectural decoration image can be effectively captured; these features are crucial for evaluating design style. Extracting pixel color values ​​within specific physical structural regions and calculating the relative color relationships between major color clusters allows for precise quantification of color saturation distribution, a key component of aesthetic features. Finally, normalization of all aesthetic feature vectors ensures fair comparison and utilization of different features in subsequent association database matching and inverse derivation of decorative element combination patterns, thereby improving the robustness and accuracy of the entire analysis method.

[0037] The above technical solutions enable more refined and comprehensive visual analysis of architectural decoration images. Adaptive brightness and contrast adjustment effectively improves image quality, making subsequent feature extraction more accurate. The application of semantic segmentation allows aesthetic feature extraction to focus on key physical structures in the image, avoiding interference from irrelevant background information, thereby improving the representativeness and effectiveness of features. The extraction of the central skeleton line provides a structured basis for the quantification of line and shape features. The calculation of major color clusters and their relative color relationships makes the quantification of color saturation distribution more detailed and accurate. The normalization of aesthetic feature vectors ensures the comparability of features in different dimensions, significantly improving the accuracy and stability of subsequent nearest neighbor matching. Overall, these refined steps work together to enable the aesthetic feature vectors extracted from architectural decoration images to more accurately and comprehensively reflect the overall visual performance of the image, providing a high-quality data foundation for subsequent inverse deciphering of the combination rules of decorative elements.

[0038] For details, please refer to Figure 3 S14 includes: S141, evaluate the distribution characteristics of pixel color values ​​within the main physical structure region segmented by the semantic segmentation. The distribution characteristics evaluation includes calculating the number of histogram peaks of the pixel color values ​​and the distance between adjacent peaks. S142, Based on the distribution characteristic evaluation results, dynamically determine the number of main color clusters; S143, Identify the main color clusters based on the number of main color clusters; S144, Calculate the relative color relationships between the identified primary color clusters.

[0039] Specifically, the evaluation of pixel color value distribution characteristics refers to analyzing the distribution of pixel color values ​​in the color space to reveal their inherent structure and patterns. Calculating the number of peaks in the histogram of pixel color values ​​aims to identify the types or quantities of primary colors in the image; each peak typically corresponds to one primary color. The distance between adjacent peaks measures the degree of difference or smoothness of transition between these primary colors; a larger distance indicates a more significant color difference, while a smaller distance indicates a smoother color transition. The purpose is to provide a quantitative basis for subsequent color cluster identification, ensuring an accurate preliminary judgment of the image's color composition.

[0040] Furthermore, the number of primary color clusters is dynamically determined based on the evaluation results of distribution characteristics. This means that the system does not use a preset fixed number of color clusters, but rather adaptively adjusts the number of color clusters to be identified based on the color complexity and distribution characteristics of the image itself. For example, if the evaluation results show that the histogram has a large number of peaks and a large distance between peaks, more color clusters may need to be identified; conversely, if the number of peaks is small and the distance between peaks is small, fewer color clusters can be identified. This dynamic determination method can better adapt to the color styles of different architectural and decorative images, avoiding undersegmentation or oversegmentation problems caused by a fixed number.

[0041] Based on this, the primary color clusters are identified according to their number. Once the number of color clusters is determined, various clustering algorithms, such as K-means clustering, Mean-Shift clustering, or Gaussian Mixture Model (GMM), can be used to group the pixel color values ​​within the main physical structure regions. Each group is a primary color cluster, representing a representative color in the image. The identified color clusters effectively simplify complex color information into a few core color groups, facilitating subsequent analysis.

[0042] Finally, among the identified primary color clusters, the relative color relationships between these clusters are calculated. Relative color relationships can be understood as the relative positions, distances, or pairing rules of different color clusters in a color space. For example, the hue difference, saturation difference, brightness difference, or Euclidean distance between two color clusters can be calculated. The aim is to capture the harmony, contrast, or hierarchy of color combinations, which is crucial for evaluating the aesthetic characteristics of architectural and decorative images. By analyzing these relative relationships, a deeper understanding of the image's color composition and the designer's color intentions can be achieved.

[0043] This application's scheme accurately captures the types of main colors in an image and the degree of difference between them by evaluating the distribution characteristics of pixel color values ​​within the main physical structural regions identified through semantic segmentation. This meticulous evaluation allows the system to dynamically determine the number of main color clusters, avoiding the limitations of fixed color cluster numbers in traditional methods and ensuring adaptability to images with different color styles. Based on this, by identifying these dynamically determined main color clusters and further calculating their relative color relationships, this scheme can quantify the color aesthetic features of an image from multiple dimensions, such as color harmony, contrast, and primary / secondary relationships, providing more refined and accurate color information for subsequent aesthetic feature vector extraction.

[0044] The above technical solution enables more refined and accurate quantification of the color characteristics of the main physical structural areas in architectural decoration images. Specifically, dynamically determining the number of main color clusters allows color analysis to better adapt to the complexity and diversity of the images themselves, avoiding biases caused by fixed parameters. Simultaneously, calculating the relative color relationships between main color clusters allows for a more comprehensive capture of the aesthetic attributes of color combinations, such as color contrast and harmony, thereby improving the accuracy and representativeness of the aesthetic feature vector. Therefore, this solution provides more insightful color information for the analysis of architectural decoration images, thereby enhancing the accuracy of the inverse deduction of decorative element combination rules and the practicality of the recommendations.

[0045] In some embodiments described above, adaptive brightness and contrast adjustment for architectural decoration images is proposed. However, in practical applications, architectural decoration images often face complex and variable lighting environments, such as strong local shadows, overexposed highlights, or overall insufficient light. Traditional adaptive adjustment methods may struggle to simultaneously optimize brightness and contrast under these complex lighting conditions and effectively preserve image details, potentially leading to loss of detail, color distortion, or an unnatural overall appearance in the adjusted image, thus affecting the accuracy of subsequent aesthetic feature vector extraction. Therefore, this application further proposes an optimized scheme for adaptive brightness and contrast adjustment of architectural decoration images, aiming to improve the adjustment effect through a refined image processing workflow to better adapt to complex lighting environments.

[0046] The above-mentioned adaptive brightness and contrast adjustment of architectural decoration images specifically includes: The architectural decoration image is separated into illumination and reflection components to obtain the illumination component and the reflection component. The illumination component is dynamically compressed to obtain the dynamically compressed illumination component. The reflection component is enhanced with detail to obtain the enhanced reflection component. The dynamic range compressed illumination component is fused with the detail enhanced reflection component to obtain the adjusted architectural decoration image.

[0047] Specifically, the illumination component can be understood as the brightness information in an image produced by direct or indirect illumination from a light source; it mainly reflects the overall light and dark distribution of the image. The reflection component, on the other hand, refers to the inherent color and texture information of an object's surface in the image; it is unrelated to illumination and mainly reflects the object's details and material properties.

[0048] Separating architectural and decorative images into illumination and reflection components aims to decouple brightness and detail information for independent optimization. Dynamic range compression of the illumination component involves processing the separated component to reduce its brightness variation, thus preventing overly bright or dark areas from appearing in the image. Dynamic range compression results in a more uniform overall brightness distribution, improving visual comfort.

[0049] In practical applications, detail enhancement of the reflection component refers to processing the separated reflection component to highlight the texture, edges, and fine structures of objects in the image. The enhanced reflection component makes the visual details of the image clearer and sharper, improving its expressiveness. Finally, the dynamic range-compressed illumination component is fused with the enhanced reflection component. The purpose is to recombine the independently optimized brightness and detail information to generate an adjusted architectural and decorative image that possesses both good brightness contrast and rich detail.

[0050] This application's solution decomposes architectural decoration images into illumination and reflection components, enabling independent processing of image brightness and detail information. The illumination component carries the overall brightness and darkness information of the image; dynamic range compression of this component effectively solves overexposure or underexposure problems caused by uneven illumination, resulting in a more balanced overall brightness distribution. Simultaneously, the reflection component contains the image's inherent texture and structural details; enhancing its detail effectively improves image clarity and sharpness, preventing detail loss during brightness adjustments. It is precisely this divide-and-conquer strategy that allows for the synergistic optimization of image brightness, contrast, and detail, overcoming the limitations of traditional methods that struggle to balance brightness and detail under complex lighting conditions.

[0051] In some preferred embodiments, a specific example is given below. Suppose the input is an image containing interior architectural decorations, in which the window areas are overexposed due to strong light, while the corners of the room appear dark due to insufficient light, and the fine textures on the walls are difficult to discern.

[0052] First, the illumination and reflection components of the architectural decoration image are separated. This can be achieved using algorithms based on Retinex theory, such as single-scale Retinex (SSR) or multi-scale Retinex (MSR), which decompose the image into low-frequency components representing illumination and high-frequency components representing reflection.

[0053] Secondly, dynamic range compression is applied to the separated illumination components. For example, methods such as logarithmic transformation, gamma correction, or histogram equalization can be used to bring the brightness values ​​of overexposed and underexposed areas back to a suitable range, making the overall brightness distribution of the image more uniform, eliminating overexposure in window areas, and making corner areas clearly visible.

[0054] Next, the separated reflection components are enhanced for detail. This can be achieved through techniques such as unsharp masking, difference of Gaussians (DoG), or wavelet transform to highlight details such as wall textures and furniture edges, making them visually clearer and sharper.

[0055] Finally, the dynamic range compressed illumination component is fused with the detail-enhanced reflection component. This fusion process is typically achieved through simple multiplication or weighted averaging, recombining the two into a new image. The resulting adjusted architectural image possesses both uniform brightness distribution and rich detail; window details are restored, corner objects are clearly visible, and wall textures become more vivid, providing high-quality visual input for subsequent aesthetic feature extraction.

[0056] In some embodiments described above in this application, semantic segmentation is performed on the adjusted architectural decoration image to separate the main physical structure regions. However, in practical applications, architectural decoration images often contain complex structures, diverse textures, and different scale information. If a segmentation method with single-scale features or lacking contextual information is used, the segmentation accuracy may be insufficient, making it difficult to accurately identify and separate the main physical structure regions in the image, thereby affecting the accuracy of subsequent aesthetic feature vector extraction and analysis. If the above problems are not addressed, the quantization of aesthetic feature vectors may be inaccurate, which in turn affects the quality of the inverse solution of the decorative element combination rules and the generation of suggestions.

[0057] In response, this application further proposes the aforementioned semantic segmentation of the adjusted architectural decoration image to separate the main physical structure regions, including: Multi-scale feature extraction is performed on the adjusted architectural decoration image to obtain visual information of the adjusted architectural decoration image at different resolutions; The multi-scale features are fused to generate a feature representation that includes contextual information; Based on the fused feature representation, the pixels of the adjusted architectural decoration image are classified to identify and separate the main physical structure regions; Based on the classification results, a segmentation mask for the main physical structure region is generated.

[0058] Specifically, multi-scale feature extraction refers to capturing visual information of an image at different spatial scales by using convolutional kernels of different sizes or images with different downsampling rates. Its purpose is to ensure that both minute decorative details and grand architectural structures can be effectively perceived and encoded. For example, pyramidal network structures (such as FPN or PSPNet) can be used to obtain feature maps of an image at multiple resolution levels.

[0059] Multi-scale feature fusion can be understood as integrating features extracted from different scales to generate a more comprehensive and discriminative feature representation. This fusion is typically achieved through methods such as concatenation, summation, or attention mechanisms, aiming to combine local detail information with global contextual information, thereby enhancing the understanding of image content. Through fusion, the limitations of single-scale features can be effectively overcome, allowing subsequent pixel classification to better utilize the overall structure and local texture information of the image.

[0060] In practical applications, the classification of pixels in adjusted architectural decoration images based on the fused feature representation involves using the decoder part of a deep learning model (such as U-Net, DeepLab, etc.) to map the fused features back to the original image resolution and predict the category of each pixel (e.g., main physical structural areas such as walls, floors, ceilings, windows, and furniture). The goal is to achieve accurate pixel-level classification, laying the foundation for subsequent separation of the main physical structural areas.

[0061] Generating segmentation masks for the main physical structure regions based on the classification results refers to converting the pixel classification output into binary or multi-valued image masks, where each mask region corresponds to an identified main physical structure region. For example, for an identified "wall" region, a corresponding mask can be generated, where the wall pixels have a value of 1 and other pixels have a value of 0. The purpose is to provide clear and accurate region boundaries, facilitating further analysis and processing of each independent structural region.

[0062] Through the above technical solutions, this application can significantly improve the accuracy and robustness of semantic segmentation of architectural decoration images. Multi-scale feature extraction ensures effective perception of physical structures of different sizes and levels of detail in the image; feature fusion enhances the understanding of image contextual information, making the segmentation results more consistent with the semantics of the actual scene; and pixel classification and segmentation mask generation based on fused features achieve accurate identification and separation of major physical structure regions. This not only solves the problems of inaccurate segmentation and blurred boundaries that may occur in traditional single-scale segmentation methods in complex architectural decoration images, but also provides a high-quality regional foundation for the subsequent accurate extraction and quantification of aesthetic feature vectors, thereby improving the overall performance and practical value of the entire architectural decoration image analysis method.

[0063] In some preferred embodiments, a specific example is given below. Suppose the input is an image of architectural decorations including a living room, sofa, windows, and walls. First, multi-scale feature extraction is performed on this adjusted image. This can be done using a convolutional neural network (CNN) based feature extractor, such as ResNet or VGG networks. By extracting feature maps at different levels (e.g., from shallow to deep), visual information of the image at different resolutions is obtained. For example, shallow features may contain local details such as edges and textures, while deep features contain higher-level semantic information and global structure.

[0064] Next, these multi-scale features are fused. This can be achieved using a Feature Pyramid Network (FPN), which fuses high-level semantic information with low-level detail information through upsampling, thereby generating a feature representation with rich semantics and detail across all scales. For example, high-resolution low-level features can be element-wise added or concatenated with upsampled low-resolution high-level features to generate a feature representation that includes contextual information.

[0065] Then, based on the fused feature representation, the pixels of the adjusted architectural decoration image are classified. This can be achieved using a semantic segmentation model, such as U-Net or DeepLabV3+, whose decoder part receives the fused feature representation and classifies each pixel in the image, predicting the probability that it belongs to a major physical structure region such as "wall," "sofa," "window," or "floor."

[0066] Finally, based on the pixel classification results, segmentation masks for the main physical structure regions are generated. For example, each pixel is assigned to the category with the highest probability, resulting in a pixel-level classification map. This classification map is then converted into a series of binary masks, each corresponding to an identified main physical structure region. For example, a "wall" mask can be generated, where all pixels classified as walls have a value of 1, and the rest have a value of 0; similarly, "sofa" masks, "window" masks, etc., can be generated. These precise segmentation masks will serve as the basis for subsequent extraction of aesthetic feature vectors, ensuring accurate quantification of the aesthetic attributes of each individual structural region.

[0067] In some embodiments described above in this application, semantic segmentation is proposed to separate the main physical structural regions in the adjusted architectural decoration image. However, during pixel classification, if the structural boundaries are not effectively processed, the segmentation results may be overly sensitive to subtle features such as surface texture, patterns, or numbers, resulting in uneven and inaccurate structural boundaries that affect the accuracy of subsequent aesthetic feature extraction.

[0068] In response, this application further proposes to introduce structural boundary smoothing constraints in the above pixel classification process to suppress the response to structural surface textures, patterns, or digital enhancement effects.

[0069] Specifically, during the pixel classification process, structural boundary smoothing constraints are introduced to suppress responses to structural surface textures, patterns, or digital enhancement effects. The pixel classification process introduces structural boundary smoothing constraints to suppress responses to structural surface textures, patterns, or digital enhancement effects, including: The pixel classification results are divided into local regions to identify physical structure regions with different boundary smoothness requirements; For each of the physical structure regions, the local gradient distribution characteristics of the boundary are calculated, whereby the local gradient distribution characteristics represent the boundary sharpness or boundary ambiguity of the physical structure region. Based on the local gradient distribution characteristics, the constraint strength for the smoothness of the structural boundary of each physical structural region is dynamically determined. The constraint strength is negatively correlated with the boundary sharpness and positively correlated with the boundary ambiguity.

[0070] The introduction of structural boundary smoothing constraint refers to applying an additional regularization mechanism when classifying pixels of the adjusted architectural decoration image. This aims to ensure that the boundaries of the identified main physical structure regions have a certain degree of smoothness, and to avoid the segmentation results being broken or inaccurate due to the misidentification of surface details such as fine textures, patterns or numbers in the image as structural boundaries.

[0071] The purpose of dividing the pixel classification results into local regions is to divide the initially classified pixel set into multiple independent local regions so that the characteristics of each region can be refined, thereby identifying which regions' boundaries need stronger smoothing and which regions' boundaries need to maintain their sharpness.

[0072] The purpose of calculating the local gradient distribution characteristics of the boundary for each physical structure region is to quantify the sharpness of the boundary of that region. The local gradient distribution characteristics of the boundary can be understood as the degree and range of pixel intensity change along the boundary direction. High gradient values ​​usually indicate high boundary sharpness, while low gradient values ​​or a wide gradient distribution range indicate high boundary blur.

[0073] The constraint strength for structural boundary smoothing is dynamically determined based on the characteristics of local gradient distribution, with the aim of achieving adaptive boundary smoothing. The constraint strength is negatively correlated with boundary sharpness, meaning that for clear and sharp boundaries, the smoothing constraint is weaker to preserve their original geometric features; while it is positively correlated with boundary ambiguity, meaning that for fuzzy and unclear boundaries, the smoothing constraint is stronger to effectively regulate and smooth them.

[0074] This application's solution effectively addresses the problem of traditional semantic segmentation methods misidentifying non-structural details such as textures, patterns, or numbers as structural boundaries when processing architectural and decorative images with rich surface details. By introducing structural boundary smoothing constraints during pixel classification, the system can accurately assess the sharpness or blurriness of boundaries. Based on this, the strength of the structural boundary smoothing constraint is dynamically adjusted, ensuring that clear structural boundaries are preserved while effectively smoothing noisy boundaries caused by surface details, thus generating more accurate and robust semantic segmentation results.

[0075] Through the above technical solution, this application can effectively suppress responses to surface textures, patterns, or digital enhancements, thereby avoiding misidentification of subtle surface details as structural boundaries during pixel classification. Consequently, the resulting semantic segmentation has smoother and more accurate physical structural region boundaries, significantly improving the recognition accuracy and robustness of the main physical structural regions. This provides more reliable foundational data for subsequent aesthetic feature vector extraction (such as color saturation distribution and texture complexity), ensuring the accuracy and practicality of the final decorative element combination suggestions.

[0076] In some preferred embodiments, it is assumed that the input is an image of architectural decoration, which includes a wall with a fine pattern and a window with a sharp edge.

[0077] First, multi-scale feature extraction and fusion are performed on the adjusted architectural decoration image, and the pixels are initially classified based on the fused feature representation to identify the wall area and window area.

[0078] Next, the preliminary classification results are divided into local areas, with the wall area and window area treated as two independent physical structure areas.

[0079] Next, the local gradient distribution characteristics of the wall region's boundary are calculated. Because the wall has a fine texture, its boundary may exhibit some ambiguity or irregularity locally. For the window region's boundary, the local gradient distribution characteristics are calculated; since window edges are typically very sharp, their boundary will exhibit high sharpness.

[0080] Finally, based on the calculated local gradient distribution characteristics, the constraint strength for smoothing the structural boundaries is dynamically determined. For the wall region, due to its higher boundary ambiguity, a stronger smoothing constraint is applied to eliminate the subtle irregularities caused by the patterns, resulting in a smooth and continuous overall boundary. For the window region, due to its higher boundary sharpness, a weaker smoothing constraint is applied to maximize the preservation of the window edge clarity and geometric accuracy. Through this adaptive smoothing process, the final semantic segmentation mask can accurately separate the smooth wall region from the window region with sharp edges, avoiding interference from surface textures in structural boundary recognition.

[0081] In some embodiments described above in this application, a central skeleton line is extracted for each major physical structural region. However, in actual architectural and decorative images, major physical structural regions often contain irregular boundary segments and complex geometric features, such as intricate carvings, curved structures, or decorative patterns. If a uniform skeleton line extraction strategy is directly adopted, the extracted skeleton lines may be overly sensitive to these local details, resulting in redundant, inaccurate, or broken skeleton lines, which affects the accurate inverse deduction of the combination rules of decorative elements.

[0082] In this regard, this application further proposes that the steps for extracting the central skeleton line for each major physical structural region include: Geometric analysis based on the shape characteristics of the main physical structure region is performed to identify irregular boundary segments and complex geometric features in the main physical structure region; Based on the irregular boundary segments and the complex geometric features, the local smoothness parameters extracted from the skeleton lines are dynamically adjusted. These local smoothness parameters are used to control the sensitivity of the skeleton lines to local details. During the skeleton line extraction process, a local smoothness parameter is applied to the irregular boundary segment to suppress interference with surface details; The distance transformation is performed on the main physical structure region, and the central skeleton line is obtained based on the distance transformation result and the dynamically adjusted local smoothness parameter.

[0083] Specifically, geometric analysis of key physical structural regions based on region shape features refers to accurately identifying non-standard or complex geometric shapes by analyzing geometric attributes such as boundary contours, curvature variations, and topological structures. For example, multi-scale curvature analysis can be used to detect sharp angles, smooth curves, or concave-convex variations at boundaries, thereby distinguishing subtle geometric noise from actual irregular boundary segments or complex geometric features. The aim is to provide refined regional feature information for subsequent skeleton line extraction.

[0084] The local smoothness parameters extracted from the skeleton line are dynamically adjusted based on the identified irregular boundary segments and complex geometric features. This can be understood as adaptively setting the skeleton line algorithm's response to details based on the local geometric characteristics of the region.

[0085] Local smoothness parameters control the smoothness of skeleton lines in different areas. For example, a lower smoothness can be used in straight or regular areas to preserve structural details, while a higher smoothness can be used in irregular or complex areas to filter out noise and suppress interference from surface details. The goal is to ensure that skeleton lines accurately reflect the main structure while avoiding being misled by secondary details.

[0086] In practical applications, during skeleton line extraction, local smoothness parameters are applied to irregular boundary segments to suppress interference from surface details. Specifically, when executing the skeleton line generation algorithm, dynamically adjusted local smoothness parameters are applied to irregular boundary regions identified by geometric analysis. For example, local weighting functions can be introduced in distance transformations or morphological operations, or local convergence conditions can be adjusted during iterative refinement. This imposes stronger smoothness constraints on skeleton line generation in these regions, reducing skeleton line branches caused by local textures or minor bumps. The aim is to ensure that the extracted skeleton lines better represent the macroscopic morphology of the main physical structure, rather than its surface fine textures. Furthermore, a distance transformation is performed on the main physical structure region, and the central skeleton line is obtained based on the distance transformation result and the dynamically adjusted local smoothness parameters.

[0087] Distance transform is an image processing technique that encodes the distance from each pixel in a binary image to its nearest background pixel. By combining the distance transform result with dynamically adjusted local smoothness parameters, the position and shape of skeleton lines can be determined more accurately. For example, based on the distance transform result, a thinning algorithm based on local maxima can be used, and adjusted smoothness parameters can be applied at irregular boundary segments to guide the skeleton lines to extend along the center of the region while avoiding over-response to local details. The goal is to generate a central skeleton line that reflects the region's topology and has good smoothness.

[0088] This application's solution introduces geometric analysis based on region shape features to first achieve a refined understanding of the main physical structure regions, identifying irregular boundary segments and complex geometric features that may be difficult to handle using traditional methods. It is precisely this pre-analysis that allows the subsequent skeleton line extraction process to dynamically adjust local smoothness parameters according to the actual complexity of the region. During skeleton line extraction, appropriate local smoothness parameters are applied to the identified irregular boundary segments, effectively suppressing interference from surface details and preventing unnecessary bending or branching of the skeleton line due to local noise or fine textures. Finally, by combining the distance transformation results and the dynamically adjusted local smoothness parameters, a central skeleton line that accurately reflects the topological connectivity of the main physical structure regions while possessing good smoothness and robustness can be generated.

[0089] By employing the aforementioned technical solution, this application effectively addresses the problem that traditional skeleton line extraction methods often produce redundant, inaccurate, or overly detail-sensitive skeleton lines when processing architectural and decorative images with irregular boundary segments and complex geometric features. Through geometric analysis of regional shape features and dynamic adjustment of local smoothness parameters, the extracted central skeleton line better represents the macroscopic form of the main physical structure, reducing errors caused by interference from local details. This significantly improves the accuracy and robustness of skeleton line extraction, providing more reliable foundational data for subsequent aesthetic feature vector extraction and inverse derivation of decorative element combination rules, thereby enhancing the overall performance and practicality of the entire architectural and decorative image analysis method.

[0090] In some preferred embodiments, it is assumed that an image of architectural decoration containing intricately carved window frames needs to be analyzed.

[0091] First, semantic segmentation is performed on the image to separate the window frame, which is the main physical structure region.

[0092] Next, a geometric analysis based on the shape features of the window frame area is performed. For example, by performing multi-scale curvature analysis on the window frame boundary, subtle curves, sharp corners, and irregular concave-convex structures in the carved parts can be identified. These are identified as irregular boundary segments and complex geometric features. Based on these identification results, the system dynamically adjusts the local smoothness parameters extracted from the skeleton lines. For example, a higher smoothness parameter is set at the carved details to avoid the skeleton lines over-tracking every carved texture, while a lower smoothness parameter is set at the straight parts of the window frame to maintain its structural integrity.

[0093] Subsequently, a distance transformation is performed on the window frame area, and a skeleton line extraction algorithm is executed based on these dynamically adjusted local smoothness parameters. The resulting central skeleton line accurately reflects the overall structure of the window frame while smoothly traversing complex carved areas, avoiding redundant branches caused by detail interference. This provides a clear and concise structural representation for subsequent decorative element analysis.

[0094] Specifically, the above-mentioned geometric analysis based on the shape characteristics of the main physical structure region to identify irregular boundary segments and complex geometric features in the main physical structure region can be achieved in the following way.

[0095] Multi-scale curvature analysis is performed on the boundary of the main physical structure region to obtain information on the curvature variation of the boundary of the main physical structure region at different scales. The curvature variation information at different scales is fused to generate a curvature feature representation that includes local details and global trends; Based on the fused curvature feature representation, an adaptive threshold is set to distinguish the subtle geometric noise from the irregular boundary segments or the complex geometric features. Based on the differentiation results, the irregular boundary segments and the complex geometric features are identified.

[0096] Multi-scale curvature analysis refers to the calculation and analysis of the curvature of the boundaries of major physical structural regions at different spatial scales. For example, a Gaussian smoothing filter can be used to smooth the boundary at different standard deviations, and then the curvature at each smoothing scale can be calculated. The aim is to capture the geometric characteristics of the boundary at different levels of elaboration, thereby gaining a comprehensive understanding of its shape. The curvature variation information at different scales can be understood as the boundary curvature values ​​and their distribution calculated at different smoothing scales. This information reflects the degree and direction of the boundary's bending in both local and global contexts.

[0097] The curvature variation information at different scales is fused, specifically by feature vector concatenation, weighted averaging, or machine learning-based methods, to generate a unified curvature feature representation containing rich contextual information. This fused feature representation can simultaneously reflect subtle local changes at the boundary and the overall trend, avoiding the loss of detail or noise sensitivity issues that may occur in single-scale analysis.

[0098] Based on the fused curvature feature representation, setting an adaptive threshold refers to dynamically determining one or more thresholds according to the statistical characteristics or preset rules of the fused curvature feature representation. For example, a suitable percentile can be selected as the threshold based on the histogram distribution of the curvature features. The purpose is to flexibly adapt to the geometric characteristics of different images and different physical structure regions, improving the accuracy of differentiation. The subtle geometric noise typically refers to small, random boundary fluctuations introduced during image acquisition or processing that do not belong to the actual structural features. Irregular boundary segments or complex geometric features refer to boundary portions with significant bends, sharp corners, depressions, or convexities—non-smooth, irregular shapes. Through adaptive thresholding, these noises can be effectively distinguished from the true geometric features.

[0099] Identifying the irregular boundary segment and the complex geometric feature based on the distinction results means marking the boundary segment whose curvature feature value exceeds or falls below a specific threshold as an irregular boundary segment or a complex geometric feature.

[0100] This application's approach employs multi-scale curvature analysis of the boundaries of key physical structural regions, comprehensively capturing the geometric characteristics of the boundaries at different levels of detail. By fusing curvature variation information at different scales, a curvature feature representation that includes both local details and reflects global trends can be generated, overcoming the limitations of single-scale analysis, such as sensitivity to noise or neglect of important details. Furthermore, by setting an adaptive threshold, the discrimination criteria can be dynamically adjusted according to the actual curvature feature distribution, effectively filtering out subtle geometric noise while accurately identifying truly irregular boundary segments and complex geometric features. This method ensures the robustness and accuracy of the geometric analysis, providing a reliable basis for the dynamic adjustment of local smoothness parameters during subsequent skeleton line extraction.

[0101] The above technical solution enables accurate identification of irregular boundary segments and complex geometric features in the main physical structure regions. Compared with traditional single-scale or fixed-threshold methods, this application employs multi-scale analysis and adaptive thresholding, significantly improving the accuracy and robustness of boundary geometric feature identification, effectively avoiding misjudging subtle noise as structural features or missing important irregular structures. This provides a more refined and reliable input for the dynamic adjustment of local smoothness parameters during subsequent skeleton line extraction, thereby improving the quality of central skeleton line extraction and the ability to preserve details in architectural and decorative images.

[0102] In some embodiments of this application, when extracting the central skeleton line from the main physical structure region, a distance transformation is performed on the main physical structure region, and the central skeleton line is obtained based on the distance transformation result and dynamically adjusted local smoothness parameters. However, in practical applications, relying solely on distance transformation and local smoothness parameters may not fully guarantee that the extracted central skeleton line accurately reflects the topology of the original main physical structure region, such as its connectivity or internal hole structure. If the skeleton line is topologically inaccurate, it may lead to deviations in subsequent aesthetic feature analysis and inverse derivation of decorative element combination rules.

[0103] In response, this application further proposes the aforementioned method of performing distance transformation on the main physical structure regions, and obtaining the central skeleton line based on the distance transformation results and dynamically adjusted local smoothness parameters, including: The distance transformation is performed on the main physical structure region to obtain the distance transformation result; Based on the distance transformation result and the dynamically adjusted local smoothness parameter, the central skeleton line is obtained, and the topological constraints ensure that the central skeleton line includes the connectivity and hole structure of the main physical structure region.

[0104] Specifically, distance transform is an image processing technique used to calculate the distance of each pixel in an image to the nearest zero pixel (typically the background). The result is usually a distance map, where the value of each pixel represents its distance to a boundary. Dynamically adjusted local smoothness parameters, as described in the above embodiments, are used during skeleton line extraction to control the sensitivity of the skeleton lines to local details based on irregular boundary segments and complex geometric features of the main physical structure regions, thereby suppressing interference from surface details.

[0105] Topological constraints can be understood as a series of rules or conditions applied during skeleton line extraction to preserve the connectivity and hole structure of the original main physical structure regions. For example, in skeletonization algorithms (such as thinning algorithms or distance-transform-based skeleton extraction algorithms), topology preservation operations can be introduced to ensure that the connectivity of the region is not destroyed or holes are accidentally filled or created when iteratively removing pixels. Specifically, this can be done by checking the connectivity of the pixel's neighborhood before and after removal, for example, using Euler number or connected component analysis. The goal is to ensure that the extracted central skeleton line not only reflects the geometric center of the region, but more importantly, accurately preserves the topological features of the original region, thus providing a reliable basis for subsequent structural analysis.

[0106] This application's solution effectively addresses the potential topological distortion problem inherent in traditional methods by introducing topological constraints during the process of obtaining the central skeleton line based on distance transformation results and dynamically adjusted local smoothness parameters. The application of these topological constraints allows for strict monitoring and maintenance of the connectivity and hole structure of the main physical structural regions during skeleton line extraction. For example, when the skeletonization algorithm attempts to remove a pixel, the topological constraints assess whether the removal operation would cause region breakage or hole disappearance; if so, the removal of the pixel is prevented. This mechanism ensures that the generated central skeleton line is geometrically simplified but topologically consistent with the original region, thus avoiding structural misjudgments due to the loss of topological information.

[0107] Through the aforementioned technical solution, the extracted central skeleton line can more accurately and robustly represent the intrinsic structure of the main physical structural areas in architectural decoration images. This topology-preserving skeleton line has a significant positive impact on subsequent aesthetic feature vector extraction, especially the quantification of line and shape features. It ensures accurate identification of structural connectivity and holes, thereby improving the understanding of complex geometric forms and spatial layouts in architectural decoration images and providing a solid data foundation for generating more accurate suggestions for decorative element combinations.

[0108] In some preferred embodiments, a specific example is given below. Suppose there is a U-shaped architectural decoration area containing a noticeable hole. If skeletonization is performed using only distance transformation and local smoothness parameters, in some cases, due to noise or improper algorithm parameter settings, the internal hole may be "filled in" during skeletonization, or the lines connecting to the outside may break at some minor points.

[0109] Specifically, after performing distance transformation on the main "square-in-square" physical structure region, a distance map can be obtained. When performing skeleton extraction based on the distance map, topological structure constraints are introduced. For example, a topology-preserving thinning algorithm can be adopted. When removing boundary pixels in each iteration, the algorithm checks whether the removal operation will change the Euler number of the image (the Euler number is related to the number of connected components and holes). If removing a pixel causes a change in the Euler number, that is, it damages the connectivity or the hole structure, the pixel will not be removed. In this way, even when there are irregularities or subtle structures on the regional boundary, the obtained central skeleton line can still accurately include the external connectivity and the internal hole structure of the "square-in-square" region. Accordingly, the subsequent line and shape feature analysis performed on the skeleton line can accurately identify the complete "square-in-square" structure, avoiding misjudgment caused by the loss of topological information.

[0110] With reference Figure 2 , a structural diagram of an architectural decoration image analysis system provided by an embodiment of the present invention includes: an input end, configured to perform visual analysis processing on an input architectural decoration image, extract and quantify an aesthetic feature vector from the overall visual performance of the architectural decoration image, wherein the aesthetic feature vector includes color saturation distribution, texture complexity, light softness and contrast, spatial blank ratio, and line and shape features; a construction end, configured to construct an association database of the aesthetic feature vector and decoration element labels, wherein the association database stores aesthetic feature vectors of professional gallery images and their corresponding decoration element labels; perform neighbor matching in the association database according to the aesthetic feature vector, so as to obtain the aesthetic feature vector of the professional gallery image with the closest distance to the aesthetic feature vector; a matching end, configured to inversely solve a decoration element combination law according to a neighbor matching result, wherein the inversely solving of the decoration element combination law includes analyzing decoration elements shared by neighbor professional gallery images and their combination modes; and generating a decoration element combination suggestion corresponding to the decoration element combination law.

[0111] It should be noted that the architectural decoration image analysis system provided by the embodiment of the present invention is configured to perform all process steps of the architectural decoration image analysis method in the above embodiments, and the working principles and beneficial effects of the two correspond one-to-one, thus they will not be repeated herein.

[0112] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0113] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for analyzing architectural decoration images, characterized in that, The method includes: The input architectural decoration image undergoes visual analysis processing. An aesthetic feature vector is extracted and quantified from the overall visual performance of the image. This aesthetic feature vector includes color saturation distribution, texture complexity, light softness and contrast, spatial white space ratio, and line and shape features. Adaptive brightness and contrast adjustments are then applied to the architectural decoration image. Semantic segmentation is performed on the adjusted image to separate the main physical structure regions. A central skeleton line is extracted for each main physical structure region to aid in the analysis of line and shape features. Within the semantically segmented main physical structure regions, pixel color values ​​are extracted, and the relative color relationships between main color clusters are calculated. Finally, the aesthetic feature vector is normalized. Construct an association database between the aesthetic feature vectors and decorative element annotations, wherein the association database stores the aesthetic feature vectors of professional image library images and their corresponding decorative element annotations; Based on the aesthetic feature vector, nearest neighbor matching is performed in the association database to obtain the aesthetic feature vector of the professional image library image that is closest to the aesthetic feature vector; Based on the nearest neighbor matching results, the rules for combining decorative elements are deduced in reverse. The deduction of the rules for combining decorative elements includes analyzing the decorative elements and their combination methods that are common to images in the nearest professional image library; and generating decorative element combination suggestions corresponding to the rules for combining decorative elements.

2. The method for analyzing architectural decoration images according to claim 1, characterized in that, The step of extracting pixel color values ​​within the main physical structure region segmented by semantic segmentation and calculating the relative color relationships between main color clusters includes: The distribution characteristics of pixel color values ​​within the main physical structure region segmented by the semantic segmentation are evaluated. The distribution characteristics evaluation includes calculating the number of histogram peaks of the pixel color values ​​and the distance between adjacent peaks. The number of main color clusters is dynamically determined based on the distribution characteristics assessment results; The primary color clusters are identified based on the number of primary color clusters; Among the identified primary color clusters, the relative color relationships between the primary color clusters are calculated.

3. The method for analyzing architectural decoration images according to claim 1, characterized in that, Adaptive brightness and contrast adjustment of the architectural decoration image includes: The architectural decoration image is separated into illumination and reflection components to obtain the illumination component and the reflection component. The illumination component is dynamically compressed to obtain the dynamically compressed illumination component. The reflection component is enhanced with detail to obtain the enhanced reflection component. The dynamic range compressed illumination component is fused with the detail enhanced reflection component to obtain the adjusted architectural decoration image.

4. The method for analyzing architectural decoration images according to claim 1, characterized in that, The step of performing semantic segmentation on the adjusted architectural decoration image to separate the main physical structure regions in the adjusted architectural decoration image includes: Multi-scale feature extraction is performed on the adjusted architectural decoration image to obtain visual information of the adjusted architectural decoration image at different resolutions; The multi-scale features are fused to generate a feature representation that includes contextual information; Based on the fused feature representation, the pixels of the adjusted architectural decoration image are classified to identify and separate the main physical structure regions; Based on the classification results, a segmentation mask for the main physical structure region is generated.

5. The method for analyzing architectural decoration images according to claim 4, characterized in that, After classifying the pixels of the adjusted architectural decoration image based on the fused feature representation, the method further includes: During pixel classification, structural boundary smoothing constraints are introduced to suppress responses to structural surface textures, patterns, or digital enhancements. The process of pixel classification introduces structural boundary smoothing constraints to suppress responses to structural surface textures, patterns, or digital enhancement effects, including: The pixel classification results are divided into local regions to identify physical structure regions with different boundary smoothness requirements; For each of the physical structure regions, the local gradient distribution characteristics of the boundary are calculated, whereby the local gradient distribution characteristics represent the boundary sharpness or boundary ambiguity of the physical structure region. Based on the local gradient distribution characteristics, the constraint strength for the smoothness of the structural boundary of each physical structural region is dynamically determined. The constraint strength is negatively correlated with the boundary sharpness and positively correlated with the boundary ambiguity.

6. The method for analyzing architectural decoration images according to claim 1, characterized in that, The step of extracting the central skeleton line for each of the main physical structure regions includes: Geometric analysis based on the shape characteristics of the main physical structure region is performed to identify irregular boundary segments and complex geometric features in the main physical structure region; Based on the irregular boundary segments and the complex geometric features, the local smoothness parameters extracted from the skeleton lines are dynamically adjusted. These local smoothness parameters are used to control the sensitivity of the skeleton lines to local details. During the skeleton line extraction process, a local smoothness parameter is applied to the irregular boundary segment to suppress interference with surface details; The distance transformation is performed on the main physical structure region, and the central skeleton line is obtained based on the distance transformation result and the dynamically adjusted local smoothness parameter.

7. The method for analyzing architectural decoration images according to claim 6, characterized in that, The geometric analysis of the main physical structure region based on the region shape features to identify irregular boundary segments and complex geometric features in the main physical structure region includes: Multi-scale curvature analysis is performed on the boundary of the main physical structure region to obtain information on the curvature variation of the boundary of the main physical structure region at different scales. The curvature variation information at different scales is fused to generate a curvature feature representation that includes local details and global trends; Based on the fused curvature feature representation, an adaptive threshold is set to distinguish geometric noise from the irregular boundary segment or the complex geometric feature; Based on the differentiation results, the irregular boundary segments and the complex geometric features are identified.

8. The method for analyzing architectural decoration images according to claim 6, characterized in that, The step of performing a distance transformation on the main physical structure region and obtaining the central skeleton line based on the distance transformation result and the dynamically adjusted local smoothness parameter includes: The distance transformation is performed on the main physical structure region to obtain the distance transformation result; Based on the distance transformation result and the dynamically adjusted local smoothness parameter, the central skeleton line is obtained, which includes the connectivity and hole structure of the main physical structure region.

9. A system for analyzing architectural decoration images, characterized in that, The system includes: The input end is used to perform visual analysis processing on the input architectural decoration image. It extracts and quantifies aesthetic feature vectors from the overall visual performance of the architectural decoration image. These aesthetic feature vectors include color saturation distribution, texture complexity, light softness and contrast, spatial white space ratio, and line and shape features. The input end then performs adaptive brightness and contrast adjustment on the architectural decoration image. Semantic segmentation is performed on the adjusted architectural decoration image to separate the main physical structure regions. A central skeleton line is extracted for each main physical structure region to aid in the analysis of line and shape features. Within the semantically segmented main physical structure regions, pixel color values ​​are extracted, and the relative color relationships between main color clusters are calculated. Finally, the aesthetic feature vectors are normalized. The construction end is used to construct an association database of the aesthetic feature vectors and decorative element annotations. The association database stores the aesthetic feature vectors of professional image libraries and their corresponding decorative element annotations. Based on the aesthetic feature vectors, nearest neighbor matching is performed in the association database to obtain the aesthetic feature vectors of the professional image libraries that are closest to the aesthetic feature vectors. The matching end is used to reverse-solve the decorative element combination rules based on the nearest neighbor matching results. The reverse-solved decorative element combination rules include analyzing the decorative elements and their combination methods common to the images in the nearest professional image library; and generating decorative element combination suggestions corresponding to the decorative element combination rules.

Images