Furniture material detection and risk assessment system

Abstract

The invention provides a furniture material detection and risk assessment system, which relates to the field of intelligent detection, captures a furniture material surface image, adopts a multi-stage image enhancement algorithm, then synchronously captures texture, color, edge and structure four-dimensional features, and performs a feature fusion mechanism. And then, decision dynamic semantic matching is performed on the multi-dimensional joint features extracted in real time and the features extracted by the standard feature database, so that accurate distinguishing of high-similarity materials can be realized. And then, outputting a material identification result for a matching result, and finally associating a risk database to generate a risk assessment report. Therefore, the limitation of traditional destructive detection can be broken through, rapid nondestructive identification of furniture materials is realized, and meanwhile, the problem of misjudgment of high-similarity materials is effectively solved, so that a high-precision and full-dimensional intelligent evaluation scheme is provided for production quality inspection and consumer end verification.

Description

Technical Field

[0001] The present application relates to the field of intelligent detection, and more particularly, in an embodiment of the present application, to a furniture material detection and risk assessment system. Background Art

[0002] As consumers' requirements for furniture environmental protection, safety and quality increase, accurate identification and risk assessment of furniture materials have become key issues that the industry needs to solve urgently. In modern furniture manufacturing, the types of materials are becoming increasingly complex, including natural wood, artificial boards, polymer plastics, metal alloys and composite materials, with significant differences in physical properties, chemical composition and potential risks. The diversity and similarity of appearance of these materials lead to a high misjudgment rate in traditional empirical identification methods, which seriously restricts the efficiency of quality control and safety supervision.

[0003] Specifically, the existing detection technology system still relies on laboratory chemical analysis (such as chromatography, spectroscopy) or physical destructive testing (such as mechanical strength testing). Such methods require sampling and destruction of furniture structures. The process is cumbersome and costly, and cannot meet the needs of real-time detection on the production line or rapid verification on the consumer side. In addition, traditional image recognition technology mostly relies on static analysis of a single feature (such as texture or color). When faced with highly similar materials (such as ring-porous wood and diffuse-porous wood of different tree species), it is difficult to effectively integrate multi-dimensional features, resulting in insufficient recognition accuracy in complex scenarios.

[0004] Therefore, an optimized furniture material testing and risk assessment solution is desired. Summary of the invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a furniture material detection and risk assessment system, which first uses image recognition and analysis methods based on deep learning, and adopts a multi-level image enhancement algorithm to eliminate environmental interference and optimize detail representation after capturing the surface image of the furniture material. Then, the four-dimensional features of texture, color, edge and structure are captured synchronously, and a joint feature vector with material identification is constructed using a feature fusion mechanism. After that, the multi-dimensional joint features extracted in real time are dynamically semantically matched with the features extracted from the standard feature database to achieve accurate distinction of highly similar materials. Subsequently, the material identification result is output for the matching result, and finally the risk database is associated to generate a risk assessment report. The present application breaks through the limitations of traditional destructive detection, and realizes rapid and non-destructive identification of furniture materials through non-contact image acquisition and multi-feature dynamic fusion, while effectively solving the problem of misjudgment of highly similar materials (such as oak and beech, PVC and ABS), thereby providing a high-precision, full-dimensional intelligent evaluation solution for production quality inspection and consumer-end verification.

[0006] According to one aspect of the present application, a furniture material detection and risk assessment system is provided, which includes: A furniture material surface image acquisition module, used to acquire a furniture material surface image acquired by a high-resolution camera; A multi-dimensional feature joint module, used for performing image preprocessing on the furniture material surface image and extracting the furniture material multi-dimensional joint features from the preprocessed furniture material surface image; A material standard coding module, used for extracting a set of furniture material standard coding features from a standard furniture material feature database; A report generation module is used to perform semantic query analysis on the set of multidimensional joint features of the furniture materials and the set of standard coding features of the furniture materials to obtain a material risk assessment report, wherein the report generation module includes: a furniture material query response unit, which is used to perform furniture material semantic query analysis based on a decision point spectrum on the set of multidimensional joint features of the furniture materials and the set of standard coding features of the furniture materials to obtain dynamic query response coding features of furniture materials; and an evaluation unit, which is used to obtain the material risk assessment report based on the dynamic query response coding features of the furniture materials.

[0007] Compared with the prior art, the furniture material detection and risk assessment system provided by the present application first uses image recognition and analysis methods based on deep learning. After capturing the surface image of the furniture material, a multi-level image enhancement algorithm is used to eliminate environmental interference and optimize the detail representation. Then, the four-dimensional features of texture, color, edge and structure are captured synchronously, and a joint feature vector with material identification is constructed using a feature fusion mechanism. After that, the multi-dimensional joint features extracted in real time are dynamically semantically matched with the features extracted from the standard feature database to achieve accurate distinction of highly similar materials. Subsequently, the material identification results are output for the matching results, and finally the risk database is associated to generate a risk assessment report. The present application breaks through the limitations of traditional destructive testing, and realizes rapid and non-destructive identification of furniture materials through non-contact image acquisition and dynamic fusion of multiple features. At the same time, it effectively solves the problem of misjudgment of highly similar materials (such as oak and beech, PVC and ABS), thereby providing a high-precision, full-dimensional intelligent evaluation solution for production quality inspection and consumer-end verification. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 4 is a system block diagram of a furniture material detection and risk assessment system according to an embodiment of the present application.

[0010] Figure 2 Schematic diagram of data flow of a furniture material detection and risk assessment system according to an embodiment of the present application.

[0011] Figure 3 It is a block diagram of a multi-dimensional feature combination module in a furniture material detection and risk assessment system according to an embodiment of the present application.

[0012] Figure 4 It is a block diagram of a report generation module in the furniture material detection and risk assessment system according to an embodiment of the present application.

[0013] Figure 5 4 is a block diagram of a furniture material query response unit in a furniture material detection and risk assessment system according to an embodiment of the present application.

[0014] Figure 6 4 is a block diagram of an evaluation unit in a furniture material detection and risk assessment system according to an embodiment of the present application. DETAILED DESCRIPTION

[0015] Various exemplary embodiments, features and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0016] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0017] In addition, in order to better illustrate the present application, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present application can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present application.

[0018] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0019] As consumers' demands for furniture environmental friendliness, safety and quality increase, accurate identification and risk assessment of furniture materials have become important challenges facing the industry. A variety of materials are used in modern furniture manufacturing, including natural wood, artificial boards, plastics, metal alloys and composite materials, which have significant differences in physical properties and chemical composition, increasing the complexity of identification. Traditional identification methods such as laboratory chemical analysis or physical destructive testing are not only costly and cumbersome, but also often require the destruction of furniture structures, which cannot meet the needs of real-time detection and rapid verification. In addition, traditional image recognition technology based on a single feature performs poorly when processing materials with high appearance similarity and is difficult to provide sufficient recognition accuracy.

[0020] Accordingly, the technical concept of the present application is to use image recognition and analysis methods based on deep learning. After capturing the surface image of the furniture material, a multi-level image enhancement algorithm is used to eliminate environmental interference and optimize the detail representation. Then, the four-dimensional features of texture, color, edge and structure are captured synchronously, and a joint feature vector with material identification is constructed using a feature fusion mechanism. After that, the multi-dimensional joint features extracted in real time are dynamically semantically matched with the features extracted from the standard feature database to achieve accurate distinction of highly similar materials. Subsequently, the material identification results are output for the matching results, and finally the risk database is associated to generate a risk assessment report. The present application breaks through the limitations of traditional destructive testing, and realizes rapid and non-destructive identification of furniture materials through non-contact image acquisition and dynamic fusion of multiple features. At the same time, it effectively solves the problem of misjudgment of highly similar materials (such as oak and beech, PVC and ABS), thereby providing a high-precision, full-dimensional intelligent evaluation solution for production quality inspection and consumer-end verification.

[0021] In response to the above technical problems, this application proposes a furniture material detection and risk assessment system. Figure 1 4 is a system block diagram of a furniture material detection and risk assessment system according to an embodiment of the present application. Figure 2 FIG. 1 is a data flow diagram of a furniture material detection and risk assessment system according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the furniture material detection and risk assessment system 100 of the embodiment of the present application, it includes: a furniture material surface image acquisition module 110, which is used to obtain the furniture material surface image acquired by a high-resolution camera; a multidimensional feature joint module 120, which is used to perform image preprocessing on the furniture material surface image and extract the furniture material multidimensional joint features from the preprocessed furniture material surface image; a material standard coding module 130, which is used to extract a set of furniture material standard coding features from a standard furniture material feature database; a report generation module 140, which is used to perform semantic query analysis on the furniture material multidimensional joint features and the set of furniture material standard coding features to obtain a material risk assessment report.

[0022] In the furniture material detection and risk assessment system 100, the furniture material surface image acquisition module 110 is used to obtain the furniture material surface image collected by the high-resolution camera. It should be understood that in modern furniture manufacturing, there are many kinds of materials, including natural wood, artificial board, polymer plastic, metal alloy and composite material, etc. These materials not only have different physical properties, but also different chemical compositions, which makes the traditional empirical identification method have a high misjudgment rate, which seriously restricts the efficiency of quality control and safety supervision. In the technical solution of the present application, the image captured by the high-resolution camera can capture the extremely subtle texture, color changes, and edge and structural information of the material surface. This fine imaging capability helps to reveal the microstructure of the material surface, such as the arrangement pattern of the ducts in the wood, the gloss difference of the plastic material, or the existence form of the metal oxide layer. In this way, even materials with very similar appearances, such as oak and beech, PVC and ABS, can be effectively distinguished. For example, when distinguishing different types of wood, the high-resolution camera can capture the subtle texture changes on the wood surface, which may be difficult to detect under ordinary cameras, but appear very obvious in high-resolution images. Similarly, for plastic materials, variations in surface gloss can be clearly presented through high-resolution images, which is crucial for identifying different plastic types.

[0023] In the furniture material detection and risk assessment system 100, the multi-dimensional feature joint module 120 is used to perform image preprocessing on the furniture material surface image and extract the furniture material multi-dimensional joint features from the preprocessed furniture material surface image. It should be understood that environmental factors (such as uneven lighting, surface reflection, equipment noise) may cause the material surface image to have blurred details, distorted colors or broken textures, which directly affect the reliability of subsequent feature extraction. For example, the arrangement of ducts on the wood surface may be misjudged as cracks due to shadow occlusion, the gloss difference of plastic materials may cause color feature deviation due to strong light reflection, and the metal oxide layer is easily confused with stains in low-contrast images. If such interference is not fully eliminated, the extraction deviation of a single feature (such as edge continuity) will accumulate, eventually causing material classification errors. Therefore, through a carefully designed multi-level collaborative image preprocessing process, it can be ensured that the quality of the obtained image is high enough for the next step of feature extraction. Moreover, considering that subtle differences in highly similar materials are often hidden in the combination of features in multiple dimensions, it is necessary to comprehensively consider multiple information such as texture features, color features, edge features, and structural features in order to achieve accurate distinction between highly similar materials.

[0024] Figure 3 FIG. 1 is a block diagram of a multi-dimensional feature combination module in a furniture material detection and risk assessment system according to an embodiment of the present application. Figure 3 As shown, in an embodiment of the present application, the multidimensional feature joint module 120 includes: a furniture material surface image preprocessing unit 121, used to perform image preprocessing on the furniture material surface image to obtain a preprocessed furniture material surface image; a furniture material multidimensional feature fusion unit 122, used to extract texture features, color features, edge features and structural features from the preprocessed furniture material surface image to obtain a furniture material texture feature vector, a furniture material color feature vector, a furniture material edge feature vector and a furniture material structural feature vector, and perform feature fusion on the furniture material texture feature vector, the furniture material color feature vector, the furniture material edge feature vector and the furniture material structural feature vector to obtain a furniture material multidimensional joint feature vector as the furniture material multidimensional joint feature.

[0025] Specifically, the furniture material surface image preprocessing unit 121 is used to perform image preprocessing on the furniture material surface image to obtain a preprocessed furniture material surface image. In an embodiment of the present application, the image preprocessing includes image denoising, illumination correction and color enhancement. It should be understood that considering environmental interference (such as uneven illumination, surface reflection, equipment noise) will cause blurred details, color distortion or texture fracture in the material surface image, which directly affects the reliability of subsequent feature extraction. For example, the arrangement of ducts on the wood surface may be misjudged as cracks due to shadow occlusion, the gloss difference of plastic materials may cause color feature deviation due to strong light reflection, and the metal oxide layer is easily confused with stains in low-contrast images. If such interference is not fully eliminated, the extraction deviation of a single feature (such as edge continuity) is accumulated, which eventually leads to material classification errors, especially the misjudgment rate of highly similar materials (such as the spring wood texture of oak and beech, the surface gloss of PVC and ABS) increases significantly. Based on this, the present application performs image preprocessing on the furniture material surface image to obtain a preprocessed furniture material surface image. In particular, the image preprocessing includes image denoising, illumination correction and color enhancement. In particular, the present application designs a multi-level collaborative image preprocessing process: first, an adaptive noise filtering algorithm is used to eliminate the interference of camera sensor noise and tiny scratches on the surface of the material, retaining the continuity of the micro texture; secondly, a multi-scale Retinex illumination correction model is used to separate the ambient light component and the intrinsic reflection characteristics of the material to solve the problem of local overexposure or underexposure caused by strong light in the factory or natural light at the consumer end; further combined with color gamut mapping and dynamic range compression technology, the color shift caused by the light absorption characteristics of the material (such as dark wood) or surface coating (such as metal paint) is corrected to restore the true color distribution of the material. In this way, the preprocessed image not only restores the intrinsic color of the material, but also highlights the microstructure that cannot be captured by traditional imaging equipment, such as the cell wall thickness gradient at the junction of hardwood early wood and late wood, and the streamline pattern produced by plastic injection molding.

[0026] Specifically, the furniture material multi-dimensional feature fusion unit 122 is used to extract texture features, color features, edge features and structural features from the pre-processed furniture material surface image to obtain furniture material texture feature vectors, furniture material color feature vectors, furniture material edge feature vectors and furniture material structural feature vectors, and perform feature fusion on the furniture material texture feature vectors, the furniture material color feature vectors, the furniture material edge feature vectors and the furniture material structural feature vectors to obtain furniture material multi-dimensional joint feature vectors as the furniture material multi-dimensional joint features. It should be understood that the subtle differences of highly similar materials are often hidden in the feature combination of multiple dimensions. For example, the identification of oak and beech requires simultaneous analysis of texture features (annular holes / diffuse holes pattern of duct arrangement), color features (heartwood hue offset), edge features (ray cell distribution density) and structural features (transition morphology between early wood and late wood), while a single feature analysis method (such as relying only on texture grayscale co-occurrence matrix) will cause misjudgment due to local feature similarity (such as both woods have straight grain texture). Similarly, the distinction between PVC and ABS plastics requires comprehensive color gloss (surface reflection spectrum characteristics), edge sharpness (injection molding burr morphology) and structural density (microscopic pore distribution). A single feature dimension cannot capture such complex differences. Therefore, the present application extracts texture features, color features, edge features and structural features from the pre-processed furniture material surface image to obtain furniture material texture feature vectors, furniture material color feature vectors, furniture material edge feature vectors and furniture material structural feature vectors, and performs feature fusion on the furniture material texture feature vectors, the furniture material color feature vectors, the furniture material edge feature vectors and the furniture material structural feature vectors to obtain a multi-dimensional joint feature vector of the furniture material. In this way, the generated joint feature vector containing multi-dimensional identification information can realize nonlinear interaction and semantic enhancement between features. Specifically, in the technical solution of the present application, for texture features, the gray level co-occurrence matrix (GLCM) method is used to analyze the texture pattern of the material surface. For oak, its unique arrangement of annular pore vessels results in specific contrast and angular second-order moment values ​​in the calculated GLCM matrix; in contrast, beech exhibits different statistical properties, such as lower contrast and higher entropy values, due to its diffuse pore vessel arrangement. The texture feature vector generated in this process can effectively capture the subtle differences in texture between the two woods. Color features are obtained through color histograms or Lab color space conversion. For example, in the Lab color space, the color offset of the oak heartwood is large and the color saturation is high, which is related to its unique pigment deposition; while the color distribution of beech is relatively more uniform and the saturation is lower. By quantitatively analyzing the color histograms of oak and beech images, their respective color feature vectors can be constructed, which reflect the color differences between the two materials.The extraction of edge features usually relies on the Sobel operator or the Canny edge detector. Due to its obvious annual ring structure, the edge detection results of oak show clear boundary lines; while the edge transition of beech is smoother due to the more uniform distribution of ducts. The edge feature vector constructed based on these edge information helps to further distinguish the two materials. As for structural features, frequency domain analysis methods such as Fourier transform or wavelet transform can be used to reveal the internal organizational structure of the material. The earlywood and latewood in oak alternate, and this periodic change will appear as a significant frequency component in the Fourier transform results; on the contrary, the structure of beech is relatively consistent, resulting in a lack of obvious periodic signals in its spectrum. The resulting structural feature vector provides additional dimensional information support for identification. After completing the extraction of the above four features, the next key step is to fuse these independent feature vectors into a comprehensive multi-dimensional joint feature vector. This process may involve the application of principal component analysis (PCA), linear discriminant analysis (LDA) or other advanced dimensionality reduction techniques. Through PCA, a new set of basis vectors can be found from the original high-dimensional feature space, so that under this set of basis vectors, the feature vectors of oak and beech have the maximum inter-class distance and the minimum intra-class distance. The multi-dimensional joint feature vector of furniture materials formed in this way not only retains the important information of each single feature, but also enhances the ability to identify subtle differences between materials, thus laying the foundation for the ultimate high-precision material classification.

[0027] In the above-mentioned furniture material detection and risk assessment system 100, the material standard coding module 130 is used to extract a set of furniture material standard coding features from the standard furniture material feature database. It should be understood that considering that the standard furniture material feature database is a multi-dimensional feature knowledge base certified by an authoritative organization, it uses industrial-grade imaging equipment and precision instruments to collect the essential feature data of mainstream furniture materials in the world under controlled environmental conditions. The database not only contains high-resolution image features, but also integrates physical and chemical parameters such as material density, thermal conductivity, and molecular crystallization morphology to form a unique digital fingerprint for each material. For example, for mahogany materials, the database not only records its texture spacing and duct distribution, but also contains biological structural parameters such as early wood rate and cell wall thickness gradient; for polymer materials, it stores microstructural features such as energy distribution and crystallization orientation angle of injection flow marks. These features are encoded by an expert system to form a standard feature vector with industry consensus, which constitutes a legal reference system for material identification. Therefore, in order to provide a global and standardized feature reference benchmark, this application extracts a set of furniture material standard coding feature vectors from the standard furniture material feature database.

[0028] Figure 4It is a block diagram of a report generation module in a furniture material detection and risk assessment system according to an embodiment of the present application. In the above-mentioned furniture material detection and risk assessment system 100, the report generation module 140 is used to perform semantic query analysis on the set of multidimensional joint features of the furniture materials and the set of standard coding features of the furniture materials to obtain a material risk assessment report. It should be understood that the multidimensional joint features of furniture materials not only contain rich material information, but also reflect the subtle differences between materials. However, the multidimensional joint features of furniture materials alone are not enough to directly complete material identification and risk assessment because they lack contextual information and reference. As an authoritative reference system, the standard furniture material feature database stores high-resolution image features and their physical and chemical parameters of various mainstream furniture materials that have been strictly certified, such as density, thermal conductivity, molecular crystal morphology, etc., to form a unique digital fingerprint for each material. By extracting a set of standard coding features of furniture materials from the database and comparing and analyzing them with the multidimensional joint features of the material to be tested, a dynamic semantic space can be constructed in which the deep association between the two is explored. This decision point spectrum-based approach goes beyond traditional linear matching algorithms and can capture the complex interactions between nonlinear features, thereby more accurately identifying materials with extremely high similarity.

[0029] Figure 4 FIG. 1 is a block diagram of a report generation module in a furniture material detection and risk assessment system according to an embodiment of the present application. Figure 4 As shown, in an embodiment of the present application, the report generation module 140 includes: a furniture material query response unit 141, which is used to perform a furniture material semantic query analysis based on a decision point spectrum on the set of the furniture material multidimensional joint features and the furniture material standard coding features to obtain furniture material dynamic query response coding features; an evaluation unit 142, which is used to obtain the material risk assessment report based on the furniture material dynamic query response coding features.

[0030] Specifically, the furniture material query response unit 141 is used to perform furniture material semantic query analysis based on the decision point spectrum on the set of the furniture material multidimensional joint features and the furniture material standard coding features to obtain furniture material dynamic query response coding features. It should be understood that considering that the accurate distinction of highly similar materials depends on the deep semantic interaction between multidimensional features and standard features, traditional feature matching methods (such as cosine similarity calculation or static rule matching) are prone to misjudgment due to the inability to capture nonlinear associations and dynamic contextual relationships. For example, when the multidimensional joint features of the material to be tested have local similarities with multiple materials in the standard feature library (such as the standard coding features of oak and beech), the traditional linear model may mismatch due to ignoring the nonlinear weight distribution between features (such as the contribution of texture features to wood identification is higher than that of color features); at the same time, the composite structural features of artificial boards (such as density board particle distribution and plywood laminated texture) need to be analyzed in the dynamic semantic space to analyze their topological association with standard features, rather than simple numerical comparison. Based on this, the present application performs furniture material semantic query analysis based on the decision point spectrum on the set of the multi-dimensional joint features of the furniture materials and the standard coding features of the furniture materials to obtain the dynamic query response coding features of the furniture materials. In particular, this method parses the topological structure between features by constructing a decision point state neighborhood matrix and a Laplace matrix. Taking metal material detection as an example, the aluminum alloy and stainless steel coding features in the standard feature library are mapped into different clusters in the low-dimensional manifold space in the spectral domain of the Laplace matrix, and the joint features of the materials to be tested are projected into this space through spectral decomposition (such as extracting the eigenvector corresponding to the minimum eigenvalue), and its distance from the standard cluster reflects the semantic similarity of the material category, thereby highly characterizing the response relationship between the two.

[0031] Figure 5 FIG. 1 is a block diagram of a furniture material query response unit in a furniture material detection and risk assessment system according to an embodiment of the present application. Figure 5As shown, in an embodiment of the present application, the furniture material query response unit 141 includes: a furniture material nonlinear decision response subunit 1411, which is used to input the furniture material multidimensional joint feature vector and each furniture material standard coding feature vector in the set of the furniture material standard coding features into a nonlinear decision response unit to obtain a set of furniture material decision point state implicit coding vectors; a furniture material Laplace matrix construction subunit 1412, which is used to construct a furniture material decision point state Laplace matrix of the set of furniture material decision point state implicit coding vectors; a furniture material Laplace matrix spectral decomposition subunit 1413, which is used to perform spectral decomposition on the furniture material decision point state Laplace matrix to obtain a set of furniture material decision point core component coding vectors; a furniture material adaptive fusion subunit 1414, which is used to adaptively fuse the set of furniture material decision point core component coding vectors to obtain a furniture material dynamic query response coding feature vector as the furniture material dynamic query response coding feature.

[0032] In the embodiment of the present application, the furniture material nonlinear decision response subunit 1411 is used to input the furniture material multidimensional joint feature vector and each furniture material standard coding feature vector in the set of furniture material standard coding features into a nonlinear decision response unit to obtain a set of furniture material decision point state implicit coding vectors, which is expressed as a furniture material nonlinear decision response formula:

[0033]

[0034]

[0035] in, is the set of standard encoding feature vectors of the furniture materials, , , and are the first, second, and third in the set of furniture material standard encoding feature vectors. and furniture material standard encoding feature vector, is the multi-dimensional joint feature vector of the furniture materials, is matrix multiplication, and They are The corresponding decision response weight matrix and decision response bias vector, is the activation function, , , , and are the first, second, and third implicit encoding vectors of the furniture material decision point states. , and The implicit encoding vector of the furniture material decision point state, It is a collection of implicit coding vectors of furniture material decision point states. It should be understood that material identification often involves complex interactions of multi-dimensional features such as texture, color, and edge. For example, oak and beech may have similarities in texture patterns, but their structural features such as cell wall thickness gradient or vessel distribution law show nonlinear differences. The subtle differences of these highly similar materials are often implicit in the nonlinear associations between features. Simply relying on linear metrics such as Euclidean distance or cosine similarity will lead to misjudgment due to the inability to capture the deep associations between cross-dimensional features. For example, when the multi-dimensional joint features of the oak to be tested are similar to the encoded features of beech in the standard library in the texture dimension but there are nonlinear differences in the structural dimension, the linear model may misjudge due to the inability to capture the cross-dimensional interaction relationship. To this end, by introducing a nonlinear decision response unit and using an activation function (such as sigmoid) to perform nonlinear transformation on the feature vector, a latent space that can represent the complex semantic relationship between features is constructed. This process is essentially to construct a deep semantic interaction space between real-time detection features and standard features, mapping the linearly inseparable decision boundaries in the original feature space (such as the superposition state of oak and beech in the texture-structure composite feature area) into a high-dimensional latent space, so that it can be transformed into a form that can be effectively distinguished by a linear classifier. Through this nonlinear feature reconstruction, the system can retain the complementarity of multi-dimensional features while strengthening the ability to characterize the essential differences of materials, thereby significantly improving the classification accuracy of highly similar materials (such as wood of different tree species and polymer materials), and providing a reliable feature basis for subsequent dynamic semantic matching and risk assessment.

[0036] In an embodiment of the present application, the furniture material Laplacian matrix construction subunit 1412 is used to: calculate the furniture material decision point state class neighborhood matrix based on the set of furniture material decision point state implicit coding vectors; calculate the furniture material decision point state class degree matrix based on the set of furniture material decision point state implicit coding vectors; calculate the furniture material decision point state Laplacian matrix based on the furniture material decision point state class neighborhood matrix and the furniture material decision point state class degree matrix.

[0037] Specifically, based on the set of implicit coding vectors of the furniture material decision point states, the furniture material decision point state class neighborhood matrix is ​​calculated, which is expressed as the furniture material point state class neighborhood calculation formula:

[0038]

[0039] in, To calculate the Euclidean norm of a vector, is the inverse hyperbolic cosine function, , , , and are the eigenvalues ​​of each position in the neighborhood matrix of the furniture material decision point state class, It is the neighborhood matrix of furniture material decision point states. It should be understandable that subtle differences in highly similar materials are often implicit in the local correlation structure of the feature space. For example, the surface gloss characteristics of PVC and ABS plastics may be similar in the global feature space, but the local connection relationships of micro injection streamlines are significantly different. Among them, PVC is easy to form dense streamline clusters due to the flexibility of the molecular chain, while ABS will produce sparse and tortuous streamline structures due to the rigid segment distribution. Traditional global feature matching methods (such as clustering based on cosine similarity) will ignore such local topological differences due to the convergence of global statistical features, resulting in overlapping distributions of the two types of materials in the feature space, which in turn leads to misjudgment. Therefore, the neighborhood matrix of furniture material decision point states is calculated based on the set of implicit coding vectors of the furniture material decision point states. Among them, the neighborhood matrix of furniture material decision point states quantifies the adjacency relationship between each decision point (i.e., the material code in the standard feature library and the material code to be tested) in the form of graph theory. Its essence is to establish a local connection network in the feature space by calculating the geometric distance between the implicit code vectors of the furniture material decision point states (such as the inverse hyperbolic cosine distance function), thereby enhancing the system's ability to capture subtle differences. For example, when distinguishing between veneer density boards and solid wood composite boards, although the surface textures of the two are similar, the neighborhood matrix will strengthen the local association between the implicit code of the solid wood composite board and the standard code of natural wood (due to the similar internal structure), while weakening its connection with the density board code, thereby achieving precise separation in subsequent spectral aggregation. This explicit modeling of local associations enables the system to break through the interference of global feature similarity and focus on the essential differences in the microstructure of the materials.

[0040] Specifically, based on the set of implicit coding vectors of the furniture material decision point states, the furniture material decision point state class matrix is ​​calculated, which is expressed as the furniture material point state class formula:

[0041]

[0042] in, To calculate the square of the vector-norm, yes The number of vectors in the equation is reduced by one. , and are the eigenvalues ​​of each position on the diagonal of the furniture material decision point state class matrix, It is the furniture material decision point state class degree matrix. It should be understood that the identification of highly similar materials depends on the local structural relationship and global importance difference between the implicit coding vectors. Therefore, the furniture material decision point state class degree matrix is ​​calculated based on the set of furniture material decision point state implicit coding vectors. That is, by calculating the decision point state class degree matrix, the connection strength and global influence of each implicit coding node in the graph structure are quantified. Specifically, the matrix records the degree value of each node (that is, the number of neighboring nodes connected to it) in the form of a diagonal matrix. By embedding the degree value of the node into the Laplace matrix calculation, the system can adaptively adjust the contribution weight of each node in the spectral analysis process - strengthening the topological influence of key feature nodes and weakening the interference of noise nodes.

[0043] Specifically, based on the furniture material decision point state class neighborhood matrix and the furniture material decision point state class degree matrix, the furniture material decision point state Laplace matrix is ​​calculated, which is expressed as the point state Laplace formula:

[0044] in, is the furniture material decision point state Laplace matrix. It should be understood that if the traditional graph structure analysis only relies on the local connection strength of the furniture material decision point state class neighborhood matrix, it may amplify the influence of noise nodes and cause the key features to be submerged; and simply relying on the global weight distribution of the furniture material decision point state class degree matrix, it may weakly determine the role of the core characteristics of the material. By constructing the decision point state Laplace matrix, the topological relationship of the graph structure is converted into a computable algebraic form. The furniture material decision point state Laplace matrix retains the fine information of the local connection strength through the difference operation of the furniture material decision point state class neighborhood matrix and the furniture material decision point state class degree matrix (that is, Laplace matrix = degree matrix-neighborhood matrix), and gives the key nodes a higher topological weight through the degree matrix. In this way, the global and local structural characteristics of the feature space can be revealed through spectral graph theory. Taking metal material detection as an example: the neighborhood matrix of furniture material decision point state quantifies the local connection strength between the characteristic codes of the oxide layer grain boundaries, while the degree matrix of furniture material decision point state gives higher weights to key nodes (such as grain boundary intersection codes); the Laplace matrix of furniture material decision point state fuses the information of the two through algebraic operations to generate a mathematical representation that reflects the overall and local relationship between the metal microstructure.

[0045] In an embodiment of the present application, the furniture material Laplace matrix spectral decomposition subunit 1413 is used to perform spectral decomposition on the furniture material decision point state Laplace matrix to obtain a set of furniture material decision point core component coding vectors, which is expressed as follows using the furniture material Laplace matrix spectral decomposition formula:

[0046] in, For Perform spectral decomposition operation, is the furniture material feature vector matrix, that is, the set of core component encoding vectors of furniture material decision points, , , and are the first, second, and third core component encoding vectors of furniture material decision points. and The core component encoding vector of furniture material decision points, The diagonal elements of the matrix are , , and The furniture material eigenvalue diagonal matrix, , , and They are , , and The corresponding eigenvalues ​​are is the diagonal matrix of eigenvalues ​​of furniture materials. It should be understood that, considering the masking of the identification signals of the essential properties of materials by noise and redundant information in high-dimensional feature space, it is difficult to effectively extract deep structural features due to the limitation of linear assumptions. These identification bases implicit in the differential geometric characteristics of the characteristic manifold need to be dissociated from high-dimensional entanglement through spectral domain transformation. Therefore, in the technical solution of the present application, the Laplace matrix of the decision point state of furniture materials is spectrally decomposed. Taking the detection of oxide layer of metal materials as an example, the microstructural characteristics of the oxide film (such as grain boundary continuity and oxide film density) are often mixed with surface scratches, stains and other noise signals in the spectral data, forming a "dimensional disaster" in the high-dimensional feature space. Traditional linear dimensionality reduction methods (such as principal component analysis) are limited by linear assumptions and cannot effectively separate these mixed features, resulting in the submersion of key identification information (such as the topological structure of grain boundary intersections). Spectral decomposition decouples the manifold structure of the feature space through algebraic topology. The Laplace matrix of the furniture material decision point state is used as a mathematical representation of the graph structure. The spectral decomposition process essentially converts the node connection relationship in the graph into the vibration mode of the core component encoding vector space of the furniture material decision point. Low-frequency feature vectors correspond to the global consistency characteristics of the data manifold (such as the uniformly distributed grain structure of the oxide film), while high-frequency feature vectors reflect local mutation characteristics (such as grain boundary defects or scratches). Spectral decomposition can be used to extract the essential structural characteristics of the material through nonlinear dimensionality reduction, significantly improving the identification accuracy of highly similar materials.

[0047] Preferably, in another example of the present application, the furniture material Laplace matrix spectral decomposition subunit 1413 is used to: perform matrix topology optimization based on cut-cycle space closure on the furniture material decision point state Laplace matrix to obtain an optimized furniture material decision point state Laplace matrix; perform spectral decomposition on the optimized furniture material decision point state Laplace matrix to obtain a set of core component encoding vectors of the furniture material decision points. The process of performing matrix topology optimization based on cut-cycle space closure on the furniture material decision point state Laplace matrix to obtain an optimized furniture material decision point state Laplace matrix is ​​specifically expressed as: It should be understandable that relying only on the spectral decomposition of the Laplacian matrix of the decision point state of the original furniture material may interfere with the global feature extraction due to local noise (such as surface scratches), resulting in feature overlap between the two types of materials when embedded in a low dimension, and causing highly similar materials to be insufficiently characterized due to global structural features.

[0048] Specifically, firstly, a pseudo-inverse operation is performed to extract the connected component information of the furniture material decision point state class neighborhood matrix and the furniture material decision point state class degree matrix, that is, let:

[0049]

[0050] Among them, the matrix and The furniture material decision point state Laplace matrix can be represented respectively The cut space and cycle space of ; Therefore, the decision point state Laplace matrix is ​​optimized through the cut space closure and the loop space closure, which is expressed as:

[0051] in, The natural constant The exponential function value with base , It is added by location point. It is the Laplace matrix of the optimal furniture material decision point state.

[0052] In this way, the global structural invariant in the structural information of the graph is extracted by closure operation based on pseudo-inverse, so that the state Laplacian matrix of the furniture material decision point is The representation of graph topological information is more robust. Specifically, the cut space closure extracts the connected component information of the cut space in the Laplacian matrix of the furniture material decision point state through pseudo-inverse operation, identifies and strengthens the global connectivity pattern of the essential characteristics of the material; the cyclic space closure amplifies the periodic structure in the cyclic space through the nonlinear transformation of the exponential function. By algebraically reconstructing the global structural invariant of the graph, the optimization of the furniture material decision point state Laplacian matrix can not only retain the local feature differences, but also highlight the global topological laws across regions.

[0053] In the embodiment of the present application, the furniture material adaptive fusion subunit 1414 is used to adaptively fuse the set of furniture material decision point core component coding vectors to obtain the furniture material dynamic query response coding feature vector as the furniture material dynamic query response coding feature, which is expressed as the furniture material adaptive fusion formula:

[0054] in, For Perform adaptive fusion operations. and They are The corresponding fusion weight matrix and fusion bias vector, yes function, yes The corresponding furniture material scoring weight vector, yes The corresponding furniture material weight value, is a masking operation, is the preset threshold, yes The corresponding furniture material mask weight value, It is the furniture material dynamic query response encoding feature vector. It should be understood that, considering that the core identification features of different material categories are often distributed in multiple dimensions of the spectral domain space, and their importance changes dynamically with the detection object and the context environment. Therefore, in order to dynamically integrate the spectral domain core components according to the material category and the detection scene, the present application obtains the furniture material dynamic query response encoding feature vector by adaptively fusing the set of the furniture material decision point core component encoding vectors. This fusion strategy combines spectral domain feature analysis with semantic context understanding, so that the dynamic query response encoding has both discriminative power and interpretability, providing high-purity feature input for subsequent classifiers.

[0055] Figure 6 FIG. 1 is a block diagram of an evaluation unit in a furniture material detection and risk evaluation system according to an embodiment of the present application. Figure 6 As shown, in an embodiment of the present application, the evaluation unit 142 includes: a material identification subunit 1421, which is used to obtain a material identification result based on the furniture material dynamic query response encoding feature vector; a material risk assessment report generation subunit 1422, which is used to integrate the material identification result with the risk information pre-stored in the database to output the material risk assessment report.

[0056] In an embodiment of the present application, the material identification subunit 1421 is used to: input the furniture material dynamic query response coding feature vector into the classifier-based material identifier to obtain the material identification result. That is, the furniture material dynamic query response coding feature vector obtained by dynamic query using the set of the furniture material multi-dimensional joint feature vector and the furniture material standard coding feature vector is classified and processed, so that the material identifier based on the classifier, whose core function is to parse the semantic information in the dynamic query response coding feature, establish a mapping relationship from abstract features to specific material categories, and achieve accurate division. The classifier is a machine learning model, and its core function is to parse the semantic information in the dynamic query response coding feature vector and establish a mapping relationship from abstract features to specific material categories. Common classifiers include support vector machines (SVM), random forests, neural networks, etc. These classifiers learn the feature differences between different material categories by training a large amount of known sample data, so as to accurately classify new unknown samples. Specifically, the classifier will assign the input feature vector to the closest material category according to its distribution. For example, when distinguishing between oak and beech, the classifier will determine whether it is closer to the standard feature vector of oak or beech based on the texture features, color features, edge features, and structural features in the dynamic query response encoding feature vector. If the input feature vector is more similar to the standard feature vector of oak, the classifier will output "oak" as the recognition result; conversely, if it is closer to the standard feature vector of beech, it will output "beech".

[0057] Specifically, the material risk assessment report generation subunit 1422 is used to integrate the material identification result with the risk information pre-stored in the database to output the material risk assessment report. It should be understood that simple material type identification cannot directly reflect its potential safety hazards and compliance risks. For example, even if a piece of furniture is accurately identified as an artificial board, it is still necessary to further determine whether its formaldehyde emission is lower than the standard, whether the adhesive contains banned chemicals, or whether the alloy composition of the metal bracket meets the load-bearing standard. Therefore, by integrating the material identification result with the risk information pre-stored in the database to output the material risk assessment report, a more accurate risk assessment can be performed. Specifically, risk information usually includes but is not limited to the following aspects: physical properties (such as strength, hardness), chemical composition (such as harmful substance content), production process (such as the use of environmental protection process) and historical records (such as recall events). This information comes from standard databases certified by authoritative organizations, industry specifications, and historical case analysis. By integrating the material identification results with these rich background data, the potential risks of materials in different aspects can be comprehensively evaluated, and a detailed risk assessment report can be generated. Among them, after obtaining the material identification results, the system automatically retrieves the relevant entries in the standard database and extracts all known risk information of the material. For example, for a certain type of wood, the system not only needs to find its basic physical properties, but also obtain relevant information about its possible health risks (such as whether it contains pests or pathogens) and environmental impacts (such as whether sustainable logging methods are used). Then, the system classifies and organizes the extracted risk information. This step is to ensure that the information structure is clear, easy to understand and apply. For example, for a certain plastic material, the system may classify its risk information into the following categories: chemical composition (such as whether it contains harmful plasticizers such as phthalates), physical properties (such as heat resistance and impact resistance), production process (such as whether environmentally friendly additives are used) and market feedback (such as whether there are consumer complaints or product recall records). Then, this information is deeply integrated with the material identification results. For example, for a certain type of artificial board, the system not only considers its basic material properties (such as density and thickness), but also combines factors such as the type of adhesive used in its production process and its formaldehyde emission to conduct a comprehensive risk assessment. In this way, a comprehensive risk assessment report can be generated, covering the potential risks of the material in multiple dimensions. In this way, the system will generate a detailed material risk assessment report. This report not only contains the basic identification results of the material, but also covers its potential risks in physical properties, chemical composition, production process and other aspects. For example, for a specific composite material, the report may point out that although it has excellent mechanical properties, due to the use of traditional adhesives, the formaldehyde emission exceeds the standard, which poses certain health risks.At the same time, the report will also provide corresponding improvement suggestions, such as recommending the use of more environmentally friendly production processes or selecting alternative materials with low formaldehyde emissions to reduce potential risks.

[0058] Therefore, integrating the material identification results with the risk information pre-stored in the database to output a material risk assessment report is an important step in achieving a comprehensive risk assessment. This process not only relies on accurate material identification results, but also requires the combination of rich background data and professional knowledge to ensure that the final risk assessment report can provide comprehensive and accurate information. Through complex query and matching operations, systematic classification and organization, deep information fusion, and the introduction of expert knowledge bases and external data sources, a detailed and practical risk assessment report can be generated. In addition, the flexible report generation mechanism and strict audit mechanism further improve the reliability and applicability of the system, providing strong support for quality control in the furniture manufacturing industry and consumer safety protection. This process embodies the close integration of modern information technology and professional knowledge, and provides an innovative and effective solution to the challenges faced by traditional detection methods.

[0059] In summary, the furniture material detection and risk assessment system 100 based on the embodiment of the present application is explained, which uses image recognition and analysis methods based on deep learning. After capturing the surface image of the furniture material, a multi-level image enhancement algorithm is used to eliminate environmental interference and optimize the detail representation. Then, the four-dimensional features of texture, color, edge and structure are captured synchronously, and a joint feature vector with material identification is constructed using a feature fusion mechanism. After that, the multi-dimensional joint features extracted in real time are dynamically semantically matched with the features extracted from the standard feature database to achieve accurate distinction of highly similar materials. Subsequently, the material identification result is output for the matching result, and finally the risk database is associated to generate a risk assessment report. The present application breaks through the limitations of traditional destructive testing, and realizes rapid and non-destructive identification of furniture materials through non-contact image acquisition and multi-feature dynamic fusion. At the same time, it effectively solves the problem of misjudgment of highly similar materials (such as oak and beech, PVC and ABS), thereby providing a high-precision, full-dimensional intelligent evaluation solution for production quality inspection and consumer-end verification.

[0060] As described above, the furniture material detection and risk assessment system 100 according to the embodiment of the present application can be implemented in various terminal devices. In one example, the furniture material detection and risk assessment system 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the furniture material detection and risk assessment system 100 can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the furniture material detection and risk assessment system 100 can also be one of the many hardware modules of the terminal device.

[0061] Alternatively, in another example, the furniture material detection and risk assessment system 100 and the terminal device may also be separate devices, and the furniture material detection and risk assessment system 100 may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0062] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0063] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0064] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0065] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0066] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0067] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.

[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit of the technical solution of the present invention.

Claims

1. A furniture material detection and risk assessment system, characterized in that: include: A furniture material surface image acquisition module, used to acquire a furniture material surface image acquired by a high-resolution camera; A multi-dimensional feature joint module, used for performing image preprocessing on the furniture material surface image and extracting the furniture material multi-dimensional joint features from the preprocessed furniture material surface image; A material standard coding module, used for extracting a set of furniture material standard coding features from a standard furniture material feature database; A report generation module is used to perform semantic query analysis on the set of multidimensional joint features of the furniture materials and the set of standard coding features of the furniture materials to obtain a material risk assessment report, wherein the report generation module includes: a furniture material query response unit, which is used to perform furniture material semantic query analysis based on a decision point spectrum on the set of multidimensional joint features of the furniture materials and the set of standard coding features of the furniture materials to obtain dynamic query response coding features of furniture materials; and an evaluation unit, which is used to obtain the material risk assessment report based on the dynamic query response coding features of the furniture materials.

2. The furniture material detection and risk assessment system according to claim 1, characterized in that: The multi-dimensional feature combination module comprises: A furniture material surface image preprocessing unit, used for performing image preprocessing on the furniture material surface image to obtain a preprocessed furniture material surface image; The furniture material multidimensional feature fusion unit is used to extract texture features, color features, edge features and structural features from the preprocessed furniture material surface image to obtain furniture material texture feature vectors, furniture material color feature vectors, furniture material edge feature vectors and furniture material structural feature vectors, and perform feature fusion on the furniture material texture feature vectors, the furniture material color feature vectors, the furniture material edge feature vectors and the furniture material structural feature vectors to obtain a furniture material multidimensional joint feature vector as the furniture material multidimensional joint feature.

3. The furniture material detection and risk assessment system according to claim 2, characterized in that: The image preprocessing includes image denoising, illumination correction and color enhancement.

4. The furniture material detection and risk assessment system according to claim 1, characterized in that: The furniture material query response unit comprises: A furniture material nonlinear decision response subunit, used for inputting the furniture material multidimensional joint feature vector and each furniture material standard coding feature vector in the set of furniture material standard coding features into a nonlinear decision response unit to obtain a set of furniture material decision point state implicit coding vectors; A furniture material Laplace matrix construction subunit, used to construct a furniture material decision point state Laplace matrix of a set of furniture material decision point state implicit coding vectors; A furniture material Laplace matrix spectral decomposition subunit, used for performing spectral decomposition on the furniture material decision point state Laplace matrix to obtain a set of furniture material decision point core component coding vectors; The furniture material adaptive fusion subunit is used to adaptively fuse the set of furniture material decision point core component coding vectors to obtain the furniture material dynamic query response coding feature vector as the furniture material dynamic query response coding feature.

5. The furniture material detection and risk assessment system according to claim 4, characterized in that: The furniture material Laplace matrix construction subunit is used to: Calculating a furniture material decision point state class neighborhood matrix based on the set of furniture material decision point state implicit coding vectors; Calculating the furniture material decision point state class matrix based on the set of furniture material decision point state implicit coding vectors; The furniture material decision point state Laplace matrix is ​​calculated based on the furniture material decision point state class neighborhood matrix and the furniture material decision point state class degree matrix.

6. The furniture material detection and risk assessment system according to claim 5, characterized in that: The furniture material Laplace matrix spectrum decomposition subunit is used for: Performing matrix topology optimization based on cut-cycle space closure on the furniture material decision point state Laplace matrix to obtain an optimized furniture material decision point state Laplace matrix; The optimized furniture material decision point state Laplace matrix is ​​spectrally decomposed to obtain a set of core component coding vectors of the furniture material decision point.

7. The furniture material detection and risk assessment system according to claim 6, characterized in that: The evaluation unit comprises: A material identification subunit, for obtaining a material identification result based on the furniture material dynamic query response coded feature vector; The material risk assessment report generating subunit is used to integrate the material identification result with the risk information pre-stored in the database to output the material risk assessment report.

8. The furniture material detection and risk assessment system according to claim 7, characterized in that: The material identification subunit is used to input the furniture material dynamic query response encoding feature vector into a classifier-based material identifier to obtain the material identification result.

Images