Skull sex identification method based on multiple modes

Abstract

The invention discloses a multi-modality-based skull sex identification method, and relates to the field of skull sex identification, and the method comprises the steps: reconstructing a three-dimensional skull curved surface model based on skull CT scanning data to obtain a three-dimensional skull data set; for each three-dimensional skull curved surface model, adopting Koee iteration to realize conformal mapping from a three-dimensional multi-connected region to a planar circular region to obtain a two-dimensional unit disk, and respectively performing normal calculation and coloring, vertex coordinate calculation and coloring, and curvature calculation and coloring to obtain a normal graph, a spatial position graph and a curvature graph; inputting a normal graph, a spatial position graph, a curvature graph and a gender label corresponding to the three-dimensional skull curved surface model into a multi-modal skull gender identification network for training to obtain a skull gender identification model; the skull gender identification model is used for outputting the corresponding gender when the normal graph, the spatial position graph and the curvature graph of the target three-dimensional skull curved surface model are received. According to the invention, the accuracy, efficiency and generalization ability of identification can be improved.

Description

Technical Field

[0001] The present application relates to the field of skull sex identification, and in particular to a skull sex identification method based on multimodality. Background Art

[0002] Skull sex identification plays a crucial role in many fields, including anthropology, archaeology, and forensic medicine. Accurately determining the sex of a skull can provide key clues and strong evidence for many aspects, including criminal investigation, identification of remains, and research on human evolution. Early methods of skull sex identification relied primarily on manual measurement and observation, with experts determining sex through subjective assessment of skull morphological characteristics. However, these methods are limited by the operator's experience and judgment criteria, and due to the complexity of skull morphological characteristics, it is often difficult to fully and systematically reveal the inherent connections and patterns between different features.

[0003] With the advancement of computer vision and deep learning technologies, automated gender identification methods based on three-dimensional data have gradually become a research hotspot. These methods significantly reduce the tediousness of traditional manual operations through automated feature extraction and can fully exploit the underlying patterns and complex associations in the data. Deep learning models, with their vast parameter space and multi-layered structure, not only improve the accuracy of gender determination but also, to a certain extent, enhance the robustness of the models. However, deep learning methods rely heavily on learning from large amounts of data, while craniofacial data acquisition requires specialized CT or three-dimensional scanning equipment, resulting in limited data availability. Therefore, how to incorporate prior knowledge of the physical differences in male and female skull morphology into learning to reduce reliance on training data remains an unresolved issue for current deep learning-based methods. Furthermore, how to effectively process the high-dimensional features of three-dimensional craniofacial data and accurately retain key gender determination information remains a major challenge in current research. Summary of the Invention

[0004] The purpose of this application is to provide a multimodal skull gender identification method that can improve the accuracy, efficiency and generalization ability of identification.

[0005] To achieve the above objectives, this application provides the following solutions.

[0006] A multimodal skull sex identification method includes the following steps.

[0007] A three-dimensional skull surface model is reconstructed based on the skull CT scan data to obtain a three-dimensional skull dataset.

[0008] For each three-dimensional skull surface model in the three-dimensional skull dataset, Koebe iteration is used to implement conformal mapping from a three-dimensional multiply connected region to a planar circular domain to obtain a two-dimensional unit disk.

[0009] The two-dimensional unit disc is subjected to normal calculation and coloring, vertex coordinate calculation and coloring, and curvature calculation and coloring respectively to obtain a normal map, a spatial position map, and a curvature map.

[0010] A multi-modal skull gender identification network is constructed, which includes a curvature independent main network and a normal-spatial position shared main network arranged in parallel, a cross-modal bidirectional cross attention network connected with the two main networks, and a multi-scale fusion classification network.

[0011] The normal map, the spatial position map, the curvature map and the gender label corresponding to each three-dimensional skull surface model in the three-dimensional skull data set are input into the multi-modal skull gender identification network for training to obtain a skull gender identification model; wherein the skull gender identification model is used to output the corresponding gender when receiving the normal map, the spatial position map and the curvature map of the target three-dimensional skull surface model as input.

[0012] According to the specific embodiments provided in the present application, the following technical effects are disclosed: To solve the problem of high-dimensional feature extraction and analysis difficulty, in the present application, for a three-dimensional skull surface model, Koebe iteration is used to realize conformal mapping from a three-dimensional multi-connected region to a plane circular domain to obtain a two-dimensional unit disc, that is, the three-dimensional skull feature is extracted by using conformal mapping technology. The three-dimensional skull is regarded as a multi-genus open surface, and conformal mapping is used to convert it into a two-dimensional plane feature image, and then normal calculation and coloring, vertex coordinate calculation and coloring, and curvature calculation and coloring are performed respectively. The vertex, normal vector and curvature information of the three-dimensional model are retained and attached to the two-dimensional feature image, which effectively strengthens the key morphological features and reduces the data dimension, simplifies the feature extraction process, reduces the calculation amount, and makes the gender-related features (such as contour and texture) more easily extracted, thereby providing an efficient and representative feature vector for the subsequent model.

[0013] In the present application, a multi-modal skull gender identification network is also constructed, which includes a curvature independent main network and a normal-spatial position shared main network arranged in parallel, a cross-modal bidirectional cross attention network connected with the two main networks, and a multi-scale fusion classification network. Through the construction and training of the above bidirectional guided multi-modal network, the problem of memory consumption in feature processing of high-resolution images can be optimized, and the lack of skull morphological prior knowledge in existing methods can be compensated. Especially the setting of the cross-modal bidirectional cross attention network, after preliminary feature extraction of the two-dimensional skull disc image, the mapped curvature disc is used as morphological prior information to guide the combined feature extraction of the normal map and the curvature map in stages, and the cross attention mechanism is used to realize the final compression of the features, so that the network performs well in the skull gender identification task and has high identification accuracy and generalization ability.

[0014] In practical applications, the trained skull gender identification model outputs the corresponding gender when receiving the normal map, spatial position map and curvature map of the target three-dimensional skull model as input. Compared with the prior art, such application has higher identification efficiency and is more convenient to use. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0016] Figure 1 The flowchart of the skull gender identification method based on multi-modal in an embodiment of the present application.

[0017] Figure 2 The structural diagram of the multi-modal skull gender identification network.

[0018] Figure 3 The structural diagram of the curvature independent backbone network.

[0019] Figure 4 The connection diagram of the first attention module.

[0020] Figure 5 The structural diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments only constitute some embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0022] In order to make the purpose, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0023] In an exemplary embodiment, as shown in Figure 1 A skull gender identification method based on multi-modal is provided, which is executed by a computer device, specifically, can be executed by a terminal or a server, or can be executed by a terminal and a server together. In the embodiments of the present application, the method includes the following steps 101 to 105.

[0024] At step 101, a three-dimensional skull curved surface model is reconstructed based on the skull CT scan data to obtain a three-dimensional skull data set.

[0025] In practical applications, in order to effectively solve the problem of insufficient sample size commonly existing in current skull identity feature analysis, multiple sets of skull CT scan data of multiple regions and multiple different hospitals can be collected, such as 1100 sets. Then, the collected multiple sets of skull CT scan data are screened, and finally 420 sets of complete and available samples can be selected, including 420 sets of data with clear gender information annotation (223 sets of male and 197 sets of female). All the skull CT scan data are saved in DICOM format, and the resolution of each slice scanning image is 512*512 pixels, and the number of slices is between 150 and 250. The richness and diversity of these data provide a solid data foundation for subsequent processing, ensuring the wide applicability of the research results.

[0026] In order to construct a three-dimensional skull data set with high geometric precision and high quality, the original skull CT scan data is processed systematically, including the following steps: (11) For each set of skull CT scan data, a Gaussian filter is used for smoothing processing, and a Sobel edge detection operator is used to accurately extract the outline of the skull and skin.

[0027] (12) Based on the outline of the skull and skin, a Marching Cubes algorithm is used for three-dimensional reconstruction of the skull and skin to obtain a three-dimensional reconstruction model.

[0028] (13) The three-dimensional reconstruction model is smoothed to obtain an initial three-dimensional skull model; specifically, to further improve the model quality, the three-dimensional reconstruction model may have defects such as holes and cracks, which are repaired and smoothed. Among them, the smoothing process uses algorithms such as Laplace smoothing to fine-tune the surface vertex position of the three-dimensional reconstruction model, reduce local irregularities, eliminate residual noise effects, and ensure that the three-dimensional reconstruction model has higher precision to meet the stringent requirements of subsequent research on model quality.

[0029] (14) Adjust all initial three-dimensional skull models corresponding to the skull CT scan data to a preset coordinate system, and then cut to obtain multiple skull frontal surface views. Specifically, due to the significant differences in posture, position and scale of the actually collected CT scan images, these differences will interfere with subsequent data analysis and research. In order to effectively eliminate such effects, all initial three-dimensional skull models are adjusted to the Frankfurt coordinate system in this step. Then, based on the pre-set anatomical landmark points, the initial three-dimensional skull model is cut to accurately obtain the skull frontal surface view.

[0030] (15) The anterior view surface of the skull is aligned and standardized as a three-dimensional skull surface model to obtain a three-dimensional skull data set. Specifically, a Thin Plate Spline registration algorithm is used to align all the anterior view surfaces of the skull, and a standardization process is performed.

[0031] After the above series of processing procedures, the differences between different three-dimensional skull models due to different postures are eliminated, and in terms of scale, all skull samples are consistent in scale due to the unified reference to the Frankfurt coordinate system. Thus, the present application obtains a three-dimensional skull data set with sufficient sample size and high geometric accuracy.

[0032] Step 102, for each three-dimensional skull surface model in the three-dimensional skull data set, a Koebe iteration is used to realize a conformal mapping from a three-dimensional multi-connected region to a planar circular domain to obtain a two-dimensional unit disc. Specifically, to solve the difficulty of high-dimensional three-dimensional skull data feature extraction and analysis, this step gives a fully automated mapping method, which regards the anterior view surface of the skull as a three-dimensional high-genus open surface, converts it into a two-dimensional planar multi-hole disc through Koebe iteration, and analyzes it in combination with a two-dimensional image feature extraction network.

[0033] In one specific application, step 102 includes the following steps.

[0034] (21) Boundary extraction is performed on the three-dimensional skull surface model to obtain bilateral orbital rim boundaries, piriform aperture boundaries, bilateral temporal-sphenoidzygomatic arch gap boundaries, bilateral mandibularzygomatic arch gap boundaries, and curved surface outer contour boundaries.

[0035] Specifically, to ensure that Koebe iteration can be successfully and effectively performed, pre-processing of the skull surface before iteration is an essential step, and the pre-processing mainly includes hole filling and puncture operations. In particular, the supraorbital foramen, infraorbital foramen, mental foramen and other small holes on the skull surface have less impact on the overall topological structure of the skull and the main features of interest in subsequent analysis. Based on this, a fine hole filling operation can be performed to ignore the impact of these small holes on the whole. This operation is based on a morphological image processing algorithm, which converts the skull three-dimensional model into a triangular mesh form through meshing processing. For each triangular patch, calculate its area; and determine an area threshold value based on statistical analysis of a large number of samples; when the area of the triangular patch involved in a certain hole is less than the threshold value, it is determined to be a fine hole and is filled.

[0036] After the fine hole filling is completed, the space high genus open surface of the skull is precisely divided. From the perspective of medical anatomy and topology, the following parts are divided: bilateral orbital rim boundary, piriform aperture boundary, bilateral temporal-sphenoid-zygomatic arch gap boundary, bilateral mandibular-zygomatic arch gap boundary, and curved surface outer contour boundary, a total of eight boundaries.

[0037] In terms of data, the front view surface of the skull in this application is represented in the form of a triangular mesh, and the front surface of the skull is denoted as , the vertex set of all triangular facets is denoted as , the normal vector of each vertex in the three-dimensional space is denoted as , the edge set is denoted as , the triangular facet set is denoted as , and each triangular facet can be specifically represented as , where .

[0038] The bilateral orbital rim boundary is a continuous curve around the eye socket, which defines the spatial range of the eye socket, and can be defined by determining the coordinates of a series of feature points around the eye socket, and there are left and right two. For any one, the feature point set around the eye socket is denoted as , and the coordinates of each point in the three-dimensional space are denoted as , . The bilateral orbital rim boundary can be represented as the ordered connection of these feature points, that is: .

[0039] The piriform aperture boundary is the edge of the piriform bone opening where the nostrils are located, and is one of the important anatomical landmarks of the face. The feature point set around the piriform aperture is denoted as , and the coordinates are denoted as , . The piriform aperture boundary can be represented as: .

[0040] The bilateral temporal-sphenoid-zygomatic arch gap boundary is located in the gap between the temporal bone, sphenoid bone and zygomatic arch, and its shape and position are of great significance for studying the lateral structure of the skull. There are two of them, and the feature point sets around the bilateral temporal-sphenoid-zygomatic arch gap are denoted as and , and the coordinates are denoted as and , and . Then the bilateral temporal-sphenoid-zygomatic arch gap boundary can be represented as: . Wherein, , .

[0041] The bilateral mandibular-zygomatic arch gap boundary is located at the gap between the mandible and the zygomatic arch, and is closely related to the mastication function and appearance of the face. There are two of them, and the feature point set around the bilateral mandibular-zygomatic arch gap is respectively and , the coordinates are respectively and , and . Then the bilateral mandibular-zygomatic arch gap boundary can be expressed as: ; wherein, , .

[0042] The curved surface outer contour boundary defines the overall shape range of the frontal view of the skull. Let the feature point set around the outer contour be , and the coordinates be , . The outer contour boundary can be expressed as: .

[0043] (22) Seven holes are determined according to the bilateral orbital rim boundary, the piriform aperture boundary, the bilateral temporal-sphenoid-zygomatic arch gap boundary, and the bilateral mandibular-zygomatic arch gap boundary.

[0044] (23) Select any one of the seven holes as the current hole, and fill all the holes except the current hole to obtain a topological ring surface. In a specific application, this step includes the following steps.

[0045] 1) Select any hole other than the current hole as the to-be-filled hole.

[0046] 2) The center point of the to-be-filled hole is determined by a centroid calculation-based method; let the vertex set on the hole boundary be , and each vertex has coordinates in the three-dimensional space. The coordinates of the center point are calculated by the following formula: ; ; .

[0047] 3) Connect the center point to each vertex of the to-be-filled hole in turn to obtain a set of line segments.

[0048] 4) According to the line segment set and the boundary of the hole to be filled, a new triangular patch is determined in combination with the constraint of the triangle inequality. The line segment in the line segment set and the edge on the boundary of the hole to be filled jointly constitute a new triangular patch. In order to ensure that the newly generated triangular patch meets the geometric constraint condition, the triangle inequality (the sum of two edges is greater than the third edge) needs to be met. In this way, the effective filling of the hole is realized.

[0049] 5) The new triangular patch is used to fill the hole to be filled.

[0050] 6) Until all holes except the current hole are filled to obtain a topological torus.

[0051] Based on the above automatic filling operation, not only the complex hole structure on the surface of the skull can be efficiently processed, but also the overall topological features of the skull can be accurately preserved.

[0052] (24) A conformal mapping of the topological torus to a standard planar circular domain is calculated by using a dynamic Yamabe flow to map the current hole as a circular hole. Specifically, for the calculation process in one Koebe iteration of one current hole, the following steps are included: 1) for the topological torus, the cotangent edge weight is calculated. When all edge weights are non-negative, the triangular partition at this time is Delaunay. This triangular partition can guarantee good geometric properties in subsequent calculations; 2) the conformal factor is initialized, and the initial conformal factor is 0; 3) the dynamic Yamabe flow iteration is solved, the current Gaussian curvature is calculated, the Yamabe energy gradient is calculated, the Hessian matrix is calculated, the linear system is solved to update the conformal factor, and the inner boundary is converged to the standard circle until the iteration is ended; 4) the conformal mapping is applied; 5) the updated result surface is obtained.

[0053] wherein the function formula of the dynamic Yamabe flow is: .

[0054] wherein, is a discrete conformal factor, which is used to determine the deformation degree of the surface in the conformal transformation; V is a set of triangular patch vertices of the three-dimensional skull surface model, is a real number set; is the target curvature of the i-th triangular patch vertex, which refers to the vertex curvature value expected to be reached; is the current discrete Gaussian curvature of the i-th triangular patch vertex at time The formula describes the rate of change of the discrete conformal factor over time, and by continuously adjusting the discrete conformal factor, the curvature of the surface is close to the target curvature.

[0055] (25) until seven hole mapping is completed, get with the three-dimensional skull surface model conformal equivalent with multiple circular hole two-dimensional unit disc.

[0056] In each round of iteration, the present application selects a hole, fills the remaining holes to obtain a new topological torus, repeats the calculation process based on dynamic Yamabe flow until the convergence condition is met, maps the hole processed in this round to a circle, and until all holes are mapped to a standard circle. The above processing process introduces a method based on the discrete homogenization theory, which realizes the conversion from a three-dimensional multi-connected skull surface to a two-dimensional unit disc with the help of Koebe iteration, and can effectively solve the problem of high computational complexity caused by the complex topological structure and irregular geometry of the skull. The data of the obtained two-dimensional unit disc can be represented as: the vertex set is , the triangular patch set is , and for each vertex , its normal vector in three-dimensional space is still the original .

[0057] Step 103, normal calculation and coloring, vertex coordinate calculation and coloring, and curvature calculation and coloring are performed on the two-dimensional unit disc respectively to obtain a normal map, a spatial position map, and a curvature map; specifically, after successfully obtaining a two-dimensional unit multi-hole disc conformally equivalent to a three-dimensional skull surface model, in order to more intuitively display the characteristic information of the skull surface, a skull geometric feature image is further generated, that is, the image of the skull conformal mapping to the plane domain stores the coordinates, normal vector, curvature and other information of the vertices of the original three-dimensional space surface.

[0058] In one specific application, normal calculation and coloring, vertex coordinate calculation and coloring are performed on the two-dimensional unit disc to obtain a normal map and a spatial position map, including the following steps.

[0059] 31) Based on the two-dimensional unit disc, determine the triangular patch set of the corresponding three-dimensional skull surface model.

[0060] 32) For each triangular patch in the triangular patch set, based on any vertex, calculate two edge vectors, and then perform cross multiplication operation on the two edge vectors to obtain a normal vector; for each triangular patch , starting from the vertex , calculate two edge vectors and ; the direction of the normal vector is obtained through cross multiplication operation, that is , the vector is perpendicular to the plane formed by the edge vectors, and the direction follows the right-hand rule (if the vertices are arranged in counterclockwise order, the normal vector points to the outside).

[0061] 33) For each triangle patch in the set of triangle patches, the patch barycenter coordinates are calculated by the arithmetic mean of the coordinates of the three vertices; for each triangle patch , the corresponding vertex of the original three-dimensional discrete surface after mapping contains three coordinate components x, y, z. For the coordinates of the triangle patch center point (barycenter) , the arithmetic mean of the coordinates of the three vertices can be calculated: .

[0062] 34) The normal vector and the patch barycenter coordinates are normalized respectively; to ensure that the normal vector is of unit length, normalization is required, that is, , , where ||n|| represents the length of the normal vector. The normalization of the patch barycenter coordinates is the same.

[0063] 35) The normalized normal vector is mapped to the RGB color value in the form of interval conversion, and is assigned to the two-dimensional unit disc to obtain the normal map; the normalized normal vector can be directly mapped to the RGB color value, and each component is converted from to interval, that is: , , . According to the needs, the floating-point value can be quantized to an 8-bit unsigned integer, that is , , , and finally the color of each patch can be represented as a four-tuple . In this way, the corresponding relationship between the normal vector and the color is established, such as the positive X-axis direction corresponding to the pure red color , the positive Y-axis corresponding to the pure green color , and the positive Z-axis corresponding to the pure blue color .

[0064] 36) The normalized patch barycenter coordinates are mapped to the RGB color value in the form of interval conversion, and are assigned to the two-dimensional unit disc to obtain the spatial position map. The normalization of the patch barycenter coordinates is the same.

[0065] In addition, curvature calculation and coloring are performed on the two-dimensional unit disc to obtain the curvature map, including the following steps.

[0066] 41) Based on the two-dimensional unit disc, the corresponding set of triangle patch vertices of the three-dimensional skull surface model is determined.

[0067] 42) For each vertex in the set of triangle patch vertices, the geodesic neighborhood is determined; specifically, the neighborhood is defined as the geodesic distance is the geodesic neighborhood. Traverse the grid by breadth-first search, starting from the ith vertex, accumulate the shortest path length (i.e. geodesic distance) between vertices, stop the traversal when the path length exceeds , the vertices and edges obtained by the traversal constitute the neighborhood .

[0068] 43) Based on the vertices in the geodesic neighborhood, calculate the neighborhood area of the geodesic neighborhood; wherein the neighborhood area is the sum of the areas of all triangles in the neighborhood, the calculation formula is: .

[0069] wherein, can be calculated by the vector cross product from the coordinates of the three vertices of the triangle , , . .

[0070] 44) Based on the edges in the geodesic neighborhood, calculate the geometric properties of each edge; specifically, the edge is determined by the endpoints and , the edge vector is defined as: ; the unit vector is: ; is the Euclidean length of the edge . represents the effective length of the edge cut by the neighborhood , if the edge is completely contained in the neighborhood , then , if it is partially contained, it is simplified to (approximate processing).

[0071] Let the normal vectors of the two triangles on both sides of the edge be and , first calculate the cosine value of the included angle of the normal vectors , limit it to to avoid numerical problems, get , wherein clip is a clipping function that limits the dot product result to [-1, 1] to meet the value requirements of the cosine function; then determine the sign by the right-hand rule, wherein is the cross product of the two normal vectors, is the edge vector. If , it indicates that the edge is located in the convex region. , indicating that it is located in the concave area, and finally the signed angle is obtained ,in The range is ,symbol Decide positive and negative.

[0072] 45) Based on the neighborhood area of ​​the geodesic neighborhood and the geometric properties of each edge, calculate the curvature matrix; specifically, for each edge , calculate the outer product matrix and multiply it by , get the contribution of each edge to the matrix, and then add the contributions of all edges to the matrix ,Right now: ; The matrix Divide by the neighborhood area , and obtain the curvature matrix This matrix is ​​transformed by the normal vector of the edge in the neighborhood (given by Reflection) and direction (by Representation) depicting the surface at the vertex The local curvature characteristics at the vertex i. for: .

[0073] in, , indicating the edge The angle between the normal vectors of the triangle patches on both sides; is a unit vector The outer product matrix ( matrix), whose elements are ; Unit vector Calculated based on edge e.

[0074] 46) Perform eigenvalue decomposition on the curvature matrix to obtain the principal curvature; specifically, perform eigenvalue decomposition on the curvature matrix Perform eigenvalue decomposition and solve the characteristic equation , and get the eigenvalue and the corresponding eigenvector .in, and are the minimum and maximum eigenvalues ​​respectively. Since they are locally developable, , corresponding to the principal curvature 、 , the principal curvature direction is given by the eigenvector and given.

[0075] 47) Determine the average curvature of the vertex based on the principal curvature, and the calculation formula is as follows: .

[0076] average curvature reflects the overall bending degree of the surface at the vertex, the sign of which is determined by the sign of represents a convex region, represents a concave region.

[0077] 48) Based on the average curvature of all vertices in the set of triangular patch vertices, the average curvature of the vertices is extended to the triangular patches by the barycentric coordinate interpolation method to obtain the curvature information of each triangular patch.

[0078] 49) Map all the triangular patch curvature information to RGB color values and assign the two-dimensional unit disc to obtain a curvature map, wherein the convex region and the concave region can be visualized by color difference.

[0079] Step 104, constructing a multi-modal skull gender identification network; the multi-modal skull gender identification network includes a curvature independent main network and a normal-space position shared main network arranged in parallel, a cross-modal bidirectional cross-attention network and a multi-scale fusion classification network connected with the two main networks. Through the cooperative work of the above networks, the comprehensive capture and efficient use of skull morphological features are ensured, providing strong feature support for gender identification.

[0080] To fully tap the uniqueness and relevance of different modal features, a double-main feature extraction network is provided in the present application: a curvature independent main network and a normal-space position shared main network. Among them, the curvature independent main network is used to process the curvature map. This design is to introduce the prior knowledge of skull morphology - curvature as the core index to describe the concave-convex changes of the skull surface, which has clear morphological differences in the gender binary significant anatomical regions; the normal-space position shared main network is used to share weights to process the normal map and the spatial position map, and to strengthen the inherent correlation between the two in geometric topology description (both reflect the spatial structure attribute of the skull surface).

[0081] The normal-space position shared main network includes a normal subnetwork and a spatial position subnetwork, and the weights of the normal subnetwork and the spatial position subnetwork are shared; as Figure 2 shown, the structure of the curvature independent main network, the normal subnetwork and the spatial position subnetwork is the same, all of which adopt a hierarchical Transformer architecture based on a moving window, and build a pyramid feature level through multi-stage feature extraction.

[0082] Taking the curvature independent main network as an example, as Figure 3 shown, the curvature independent main network includes an input preprocessing module ( Figure 3The input preprocessing module, the Patch division module, the linear embedding module, the first attention module, the second attention module, the third attention module, and the fourth attention module are not shown in the figure, and a normalization module can be additionally arranged as required. The four attention modules each include two moving window multi-head self-attention sub-modules arranged in sequence.

[0083] The input preprocessing module is configured to map the geometric and texture information of the curvature map to a regular pixel grid by a uniform sampling algorithm to obtain a curvature pixel map. The curvature map is a two-dimensional unit disc vector map with color texture, and its digital representation is as follows: a vertex set is , a triangular patch set is , each patch represents a triangular topological structure composed of vertices , and contains color texture value information .

[0084] To adapt to the input requirements of the network, the input preprocessing module can be regarded as a vector map down-sampling module to realize the conversion from a vector map to a pixel map and size normalization. The core idea of the module is to map the geometric and texture information of the vector map to a regular pixel grid based on the topological relationship between the vertices and the patches by a uniform sampling algorithm. The specific steps are as follows: based on the triangular patch set , the sampling region is divided, and each patch is taken as an independent sampling unit; sampling points are uniformly generated in each patch, and the texture value of the sampling point is calculated by the barycentric coordinate interpolation according to the three vertex coordinates and the texture value of the patch. Finally, the inputs of the three modalities are standardized to 512x512 pixel specifications to form three groups of input tensors: the curvature map , the spatial position map (recording the three-dimensional spatial coordinates of each point on the skull surface), and the normal map (describing the normal vector direction of the skull surface). These standardized inputs lay a foundation for subsequent feature extraction and cross-modal interaction.

[0085] The Patch division module is configured to divide the curvature pixel map into a plurality of image blocks. Specifically, the input image is divided into 4x4 non-overlapping image blocks (Patches). Especially, for an input image of 512x512, a total of Patches can be generated.

[0086] The linear embedding module is configured to perform mapping conversion from a pixel space to a feature space on the image blocks to obtain sequence feature vectors. Specifically, the pixel value of each Patch is converted into a 128-dimensional feature vector by linear projection to realize the mapping from the pixel space to the feature space. Let the pixel value of the i-th Patch be ​, its linear embedding process can be expressed as: ;in, Indicates flattening the 4×4×3 Patch into a 48-dimensional vector. is the projection weight matrix, is the bias term, That is the The initial feature vector of each patch. The feature vectors of all patches constitute the initial feature map Through the above-mentioned patch division and linear embedding, the pixel image is converted into sequence features suitable for Transformer processing.

[0087] like Figure 4 Figure 2 shows the connection diagram of the first attention module. Each moving window multi-head self-attention submodule includes a first normalization layer, a multi-layer perceptron, a second normalization layer, and an attention mechanism layer, arranged in sequence. The moving window multi-head self-attention submodule is used to introduce a multi-head self-attention mechanism to the sequence feature vector and, combined with the window movement strategy, determine the curvature feature block.

[0088] To efficiently capture local spatial correlations in feature maps, this application uses a multi-head self-attention mechanism based on a moving window. This mechanism divides the feature map into 7×7 local windows and only calculates self-attention within the window, significantly reducing computational complexity. At the same time, a window shifting strategy (window positions in adjacent stages are offset by half the window size) enhances information interaction between windows. The calculation process is as follows: .

[0089] in, They are query, key, and value matrices (obtained by linearly transforming the feature map), which are used to capture the correlation of features at different positions; is the dimension of the key vector (here ), used to normalize the attention score to avoid numerical overflow; It is a logarithmically spaced continuous position deviation, and the network's ability to model spatial structure is enhanced by dynamically adjusting the weights of different relative positions within the window.

[0090] To construct a pyramid feature hierarchy, each feature extraction stage (except the last stage) includes a patch fusion operation, which means setting up a patch fusion submodule to halve the resolution and double the number of channels by merging the features of four adjacent patches. In other words, the curvature feature blocks output by the moving window multi-head self-attention submodule are fused. The resolution of the stage feature map is , the number of channels is , then after fusion The characteristic parameters of the stage are: This design enables the network to capture features at different scales, from local details at low stages (e.g., eyebrow arch edges) to global structures at high stages.

[0091] In addition, the normal-spatial position shared backbone network adopts a 4-stage architecture, and shared weights are used to process the normal map and the spatial position map For the normal map, initial features are generated through patch embedding After that, high semantic features are gradually generated through 4 stages of feature extraction: wherein, represents the feature processing layer of the stage, which includes 2 mobile window self-attention blocks and 1 patch fusion submodule (no patch fusion for the last stage). The feature extraction process of the spatial position map is the same as that of the normal Figure 1 : ; since both the normal map and the spatial position map describe the geometric topological properties of the skull surface (the normal vector reflects the surface orientation, and the spatial position reflects the three-dimensional coordinates), weight sharing can strengthen the feature correlation between the two, improving the overall modeling ability of the skull spatial structure. The feature parameters of each stage are as follows: the first attention module Stage 1 (128x128, 128 channels), the second attention module Stage 2 (64x64, 256 channels), the third attention module Stage 3 (32x32, 512 channels), and the fourth attention module Stage 4 (16x16, 1024 channels).

[0092] The curvature independent backbone network adopts the same 4-stage architecture as the normal-spatial position shared backbone network, but the parameters are independently initialized and are specifically used to process the curvature map Curvature, as a key indicator describing the concave-convex changes of the skull surface (e.g., there is a significant difference in the curvature of the male eyebrow arch and the female eyebrow arch), needs to be learned independently to learn its unique patterns. Its feature extraction process is: wherein, is the initial embedding feature of the curvature map, and the output features of each stage are the same as those of the normal-spatial position shared backbone network , maintain the same resolution and channel number, ensuring the feasibility of subsequent cross-modal interaction.

[0093] The core advantage of the dual backbone design set in this application is that through the collaborative strategy of "independent modeling-shared learning", both the unique discriminability of the curvature feature and the inherent correlation between the normal and spatial position features are strengthened, solving the problem that a single backbone network cannot balance the differences between multiple modalities.

[0094] To realize the deep interaction and information complementarity of the two backbone features, a cross-modal bidirectional cross-attention guiding mechanism is introduced at each stage (Stage 1 to Stage 4) of feature extraction, so that the curvature independent backbone network and the normal-space position shared backbone network can guide each other during feature learning. The shared backbone provides spatial structure constraints for curvature features, and the curvature backbone supplements surface morphology details for shared features, breaking the static limitations of traditional multi-modal fusion such as "early splicing" or "late fusion". Specifically, the cross-modal bidirectional cross-attention network includes four bidirectional guiding modules.

[0095] The bidirectional guiding module is used to: receive curvature features, normal features and spatial position features corresponding to the feature extraction stage; based on the normal features and the spatial position features, perform feature learning with the curvature features as a reference to obtain a first learning query result; through a residual connection, the first learning query result is fused with the normal features and the spatial position features respectively, and then fed back to the normal subnetwork and the spatial position subnetwork; based on the curvature features, perform feature learning with the normal features and the spatial position features as a reference to obtain a second learning query result; through a residual connection, the second learning query result is fused with the curvature features, and then fed back to the curvature independent backbone network.

[0096] Wherein, the normal features and the spatial position features are referenced to the curvature features, and the process of absorbing the surface morphology information described by the curvature includes: let the i-th normal feature be The normal feature of the stage shared backbone is , the spatial position feature is , the feature of the curvature backbone is , and the query process is: .

[0097] .

[0098] Wherein, is a cross-attention function, which is similar to self-attention, but the keys (K) and values (V) come from heterogeneous modal features, specifically: .

[0099] Through this process, the normal features can integrate the surface concave-convex patterns described by the curvature (such as the influence of the cheekbone curvature on the direction of the normal vector), and the spatial position features can associate the spatial distribution of curvature changes (such as the correspondence between the mandibular angle position and the curvature mutation).

[0100] The process of curvature features absorbing geometric topological information to optimize their own representation with normal features and spatial position features as reference is: .

[0101] .

[0102] Through this process, the curvature feature can combine the direction information of the normal vector (such as the region with positive curvature corresponding to the convergence mode of the normal vector) with the spatial coordinate constraint (such as the curvature variation range of the glabella bone region being limited by its spatial position), to improve the discrimination ability of the key anatomical parts.

[0103] To ensure that the original feature information is not covered, the cross-attention output is fused with the original feature through a residual connection to realize dynamic updating of the feature: ; ; ; This updating mechanism enables the feature to dynamically absorb cross-modal complementary information while retaining its core information.

[0104] In summary, the progressive optimization of the cross-modal bidirectional cross-attention guidance in the bidirectional guidance module (from Stage 1 to Stage 4) enables the feature interaction to gradually upgrade from low-level local detail association to high-level global semantic collaboration, ultimately forming a comprehensive feature representation with multi-modal information.

[0105] The multi-scale fusion classification network is used for: feature fusion and classification output of the final curvature feature output by the curvature independent backbone network, the final normal feature output by the normal sub-network, and the final spatial position feature output by the spatial position sub-network. Specifically, after processing by the above-mentioned double backbone network, three groups of high-level features can be obtained: normal feature , spatial position feature (resolution 16x16, channel 1024) and curvature backbone (same specification). To make full use of these features for gender determination, the present application designs a multi-scale feature fusion classification network, which realizes optimal integration of features through high-level feature refinement, global integration and channel weight learning, and improves the discrimination ability of the model. Specifically, the processing steps realized in the multi-scale fusion classification network corresponding to the integrated decision fusion module described in Figure 2 are as follows: Fine fusion of normal feature and spatial position feature: taking normal feature as the query (Q) and value (V), and spatial position feature as the key (K), the cross-attention mechanism is used to strengthen the weight of the normal feature in the key anatomical parts: ; wherein the residual connection ensures that the original discriminability of the normal feature is not lost, and the cross-attention uses the coordinate information of the spatial position feature to guide the focus of the normal vector to the regions with significant gender differences, so that the fused feature At the same time, it has direction information and spatial positioning ability.

[0106] Then, the fine fusion features and the curvature backbone high-level features are integrated to form global feature representation: The integration is based on the complementarity of feature space: Focus on the spatial structure and normal distribution of the skull, Highlight the surface curvature change pattern, and the superposition of the two realizes the complete description of "morphological structure + surface characteristics", providing comprehensive feature support for gender identification. That is, after feature integration, the final features obtained by integration are sent to the classifier for classification, so as to obtain the gender and estimated gender probability of three-dimensional skull determination.

[0107] To further improve the discriminability of the features, a channel attention mechanism (SE-Net) is introduced to dynamically learn the importance weight of each channel, and the feature channels that contribute significantly to gender identification are strengthened, which includes global average pooling, channel weight learning, weighted feature generation, and classification output.

[0108] Global average pooling is: global average pooling is performed on the global feature (size ) to compress the spatial dimension information into channel dimension statistics: .

[0109] Wherein, is the global average response of the channel, reflecting the average contribution of the channel feature in the whole space.

[0110] Channel weight learning is: through two fully connected layers, the dependence between channels is learned to generate normalized channel weights: ; wherein, and are the fully connected layer weights ( is the compression ratio), and are used to introduce nonlinearity and weight normalization, respectively, is the importance weight of the channel.

[0111] Weighted feature generation is: multiply the learned channel weight with the global feature channel by channel to strengthen the feature contribution of the key channel: .

[0112] Classification output is: after LayerNorm and adaptive average pooling on the weighted feature, input the linear classification head to generate gender prediction results: .

[0113] In addition, to improve the robustness of the model, the application additionally designs an auxiliary classification branch to learn the high-level features of the curvature backbone classify alone, and the learning quality of the curvature features is enhanced through auxiliary loss: Based on the auxiliary classification branch, the total loss in the training process is the weighted sum of the cross-entropy main loss and the auxiliary loss: Wherein, is the main loss ( is the true label), is the auxiliary loss, is the balance coefficient. The auxiliary loss forces the curvature backbone to learn more discriminative features through deep supervision, while regularizing the main branch to reduce the risk of overfitting.

[0114] To achieve high-precision skull gender identification, the application designs a multi-modal skull gender identification network based on morphological feature guidance. The network aims to deeply mine skull morphological features and effectively fuse multi-modal information through the fusion of curvature maps, spatial position maps (XYZ) and normal Figure 3 maps of the skull, an end-to-end learning framework that breaks through the limitations of traditional single-modal feature learning, providing strong technical support for high-precision skull gender identification.

[0115] Step 105, input the normal map, spatial position map, curvature map and gender label corresponding to each three-dimensional skull model in the three-dimensional skull data set into the multi-modal skull gender identification network for training to obtain a skull gender identification model; wherein the skull gender identification model is used to output the corresponding gender when receiving the normal map, spatial position map and curvature map of the target three-dimensional skull model as input.

[0116] During training, the main parameter configuration of the network is as follows: the input image size is 512x512x3, the patch size is 4x4, and the initial embedding dimension is 128; the number of channels in the four feature extraction stages is 128, 256, 512, and 1024 respectively, and each stage contains 2 moving window self-attention blocks (Stage 3 is 16 to enhance feature learning ability); the number of cross-attention heads matches the number of channels in the corresponding stage (3, 6, 12, 24), and the channel attention compression ratio , the auxiliary loss weight .

[0117] In the implementation process, the application uses the PyTorch framework to build the network and uses the AdamW optimizer (initial learning rate , weight decay ), the training batch size is 4, the pre-trained weights based on the single-main version trained on the same single small-scale dataset are loaded to expect the model to achieve fast convergence, the total number of iterations is 200 epochs, and the cosine annealing strategy is used to adjust the learning rate. Experiments show that the network can be efficiently trained on a single NVIDIA RTX4070 graphics card, with an inference time of about 83 ms / sample, balancing the accuracy and efficiency requirements.

[0118] To fully verify the advancement of the method of the present application, it is compared with mainstream skull gender identification methods and traditional machine learning algorithms. The comparison experiment uses the same performance evaluation indicators (accuracy, AUC value) to ensure the comparability and objectivity of the results.

[0119] In comparison with existing skull gender identification research, the present method shows significant advantages. As shown in Table 1, on a dataset containing 420 samples, the gender identification accuracy of the present method reaches 97.6%, and the AUC value is 0.98, which is better than most of the comparison methods. Among them, the improved back propagation neural network achieves an accuracy of 96.8% on 267 samples, but relies on manual feature extraction; the deep feature selection and fusion strategy achieves an accuracy of 96.4% on 421 samples, while the present method further improves the performance through conformal mapping and multi-modal fusion. It is worth noting that although some studies have achieved an accuracy of 94.4% on a small sample dataset (90 samples), their generalization ability is questionable due to the limited sample size. The present method achieves high accuracy and high AUC value on a medium-sized dataset, proving its advantages in balancing sample efficiency and identification performance.

[0120] Table 1

[0121] The comparison experiment with traditional machine learning methods further verifies the superiority of the deep learning architecture. As shown in Table 2, on the same dataset, the present method significantly outperforms SVM (72.9% / 0.828), KNN (63.0% / 0.630), random forest (78.5% / 0.722), and other traditional algorithms with an accuracy of 97.6% and an AUC value of 0.984. Among them, the KNN algorithm performs the worst, possibly because it cannot capture the complex nonlinear relationship of skull features; although SVM can handle nonlinear problems through kernel functions, it still performs worse than deep learning methods in high-dimensional feature space. Traditional machine learning methods rely on manual feature engineering, while the present method effectively extracts potential discriminant patterns of skull morphology through automated feature extraction and multi-modal fusion.

[0122] Table 2

[0123] The present application proposes a solution based on conformal mapping and multi-modal fusion learning to address the technical difficulties of skull gender identification. The effectiveness and innovation of the method are verified through systematic experiments. The large-scale three-dimensional skull dataset constructed by the present application covers multiple regions and multiple nationalities, providing a reliable data foundation for research. The automatic conformal mapping method successfully converts the three-dimensional skull surface into a two-dimensional vector map that retains topological and geometric features, solving the problem of high-dimensional data processing. The generation of normal maps, curvature maps, and XYZ space maps effectively retains gender binary characteristics. The dual-main architecture of the bidirectional guided multi-modal learning network realizes the synergy of independent modeling of curvature features and shared learning of normal-space features. The cross-modal bidirectional cross-attention mechanism promotes dynamic feature interaction, and the multi-scale fusion module enhances key feature discriminability.

[0124] The experimental results show that the method has an accuracy of 97.6% and an average AUC of 0.98, and performs stably across national and regional samples. The gradient heat map shows that the present application accurately captures the differences in key areas such as the eye socket and the brow. The present application not only improves the accuracy and generalization ability of skull gender identification, but also provides an important reference for the field of three-dimensional biological feature recognition.

[0125] In summary, the present application has the following advantages compared to existing technologies: (1) A large-scale three-dimensional skull and two-dimensional vector mapping dataset is constructed: To effectively address the small sample size problem of existing identification methods, head CT scan data from multiple regions, different age groups, and different genders is collected and standardized for registration processing, resulting in a high-quality, widely representative three-dimensional skull dataset. On this basis, an additional two-dimensional vector mapping dataset of the cranial anterior surface with significant morphological features is constructed. This provides rich morphological information and a reliable analysis foundation for subsequent research.

[0126] (2) An automatic three-dimensional skull feature image processing method is constructed: To address the difficulty of high-dimensional feature extraction and analysis, the present application uses conformal mapping technology to conformally map the three-dimensional skull surface to a two-dimensional unit disc. Based on the vertex coordinates, normal vectors, and curvature information of the three-dimensional model, the cranial two-dimensional feature image is constructed, effectively strengthening the key morphological features of the skull and reducing the data dimension. This simplifies the feature extraction process, reduces the computational load, and makes gender-related features more easily trained by deep learning networks, thereby providing an efficient and representative feature vector for subsequent models.

[0127] (3) Set up a double-main feature extraction architecture: In view of the characteristics of multi-modal features, a curvature independent main trunk and a normal-space position shared main trunk are designed, the feature uniqueness and relevance are balanced through the "independent modeling-shared learning" strategy, and the problem that a single main network is difficult to balance multi-modal differences is solved. Cross-modal bidirectional cross-attention guidance is set up: a bidirectional interaction mechanism is introduced at each feature extraction stage, so that the double-main features can dynamically absorb cross-modal complementary information, progressive information fusion from local details to global structure is realized, and the static limitation of traditional fusion methods is broken through. Multi-scale adaptive fusion and deep supervision are set up: through high-level feature refinement, channel attention weighting and auxiliary loss design, the optimal integration and key feature enhancement of multi-modal features are realized, and the capture ability and generalization performance of the model to gender difference features are improved. These designs jointly build an end-to-end learning framework that can deeply mine cranium morphological features and effectively fuse multi-modal information, providing strong technical support for high-precision cranium gender identification.

[0128] In an exemplary embodiment, a computer device, which can be a server or a terminal, is provided, and an internal structure diagram thereof can be as shown in Figure 5 The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through a network connection. The computer program is executed by the processor to implement a multi-modal based cranium gender identification method.

[0129] Those skilled in the art can understand that Figure 5 the structure shown in the above

[0130] In an exemplary embodiment, a computer device is also provided, which includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps in each of the above method embodiments.

[0131] In an exemplary embodiment, a computer readable storage medium storing a computer program is provided, the computer program, when executed by a processor, implements the steps of any of the above method embodiments.

[0132] In an exemplary embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps of any of the above method embodiments.

[0133] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0134] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. Any reference to memory, database or other medium used in the embodiments provided by the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0135] The database involved in each of the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, and the like, without being limited thereto. The processor involved in each of the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, and the like, without being limited thereto.

[0136] The technical features of the above embodiments can be combined in any manner. To make the description concise, all possible combinations of the technical features in the above embodiments are not described, but it should be considered that any combination of the technical features is within the scope of the present disclosure, as long as there is no contradiction.

[0137] The principles and implementation manners of the present application are described by using specific examples herein, and the above embodiments are only used to help understand the method of the present application and its core idea. Meanwhile, for those skilled in the art, the specific implementation manners and application ranges can be changed according to the idea of the present application. In summary, the content of the present description should not be understood as a limitation of the present application.

Claims

1. A method for skull gender identification based on multi-modal, characterized in that, The method comprises: reconstructing a three-dimensional skull surface model based on skull CT scan data to obtain a three-dimensional skull data set; for each three-dimensional skull surface model in the three-dimensional skull data set, achieving conformal mapping from a three-dimensional multi-connected region to a planar circular region by Koebe iteration to obtain a two-dimensional unit disc; respectively performing normal calculation and coloring, vertex coordinate calculation and coloring, and curvature calculation and coloring on the two-dimensional unit disc to obtain a normal map, a spatial position map, and a curvature map; constructing a multi-modal skull gender identification network; the multi-modal skull gender identification network comprises a curvature independent main network and a normal-spatial position shared main network arranged in parallel, a cross-modal bidirectional cross attention network connected with the two main networks, and a multi-scale fusion classification network; inputting the normal map, the spatial position map, the curvature map, and the gender label corresponding to each three-dimensional skull surface model in the three-dimensional skull data set into the multi-modal skull gender identification network for training to obtain a skull gender identification model; wherein the skull gender identification model is used to output a corresponding gender when receiving a normal map, a spatial position map, and a curvature map of a target three-dimensional skull surface model as input.

2. The multi-modal based skull gender identification method of claim 1, wherein, reconstructing a three-dimensional skull surface model based on skull CT scan data to obtain a three-dimensional skull data set, comprising: for each set of skull CT scan data, performing smoothing processing by using a Gaussian filter, and extracting the outline of the skull and skin by using a Sobel edge detection operator; based on the outline of the skull and skin, performing three-dimensional reconstruction of the skull and skin by using a Marching Cubes algorithm to obtain a three-dimensional reconstruction model; performing smoothing processing on the three-dimensional reconstruction model to obtain an initial three-dimensional skull model; adjusting all initial three-dimensional skull models corresponding to the skull CT scan data to a preset coordinate system, and then cutting to obtain a plurality of skull frontal view surfaces; aligning and standardizing the skull frontal view surfaces as three-dimensional skull surface models to obtain a three-dimensional skull data set. 3.The multi-modal based skull gender identification method of claim 1, wherein, for each three-dimensional skull surface model in the three-dimensional skull data set, achieving conformal mapping from a three-dimensional multi-connected region to a planar circular region by Koebe iteration to obtain a two-dimensional unit disc, comprising: extracting the boundary of the three-dimensional skull surface model to obtain bilateral orbital rim boundaries, piriform aperture boundaries, bilateral temporal-sphenoidzygomatic arch gap boundaries, bilateral mandibularzygomatic arch gap boundaries, and curved surface outer contour boundaries; determining seven holes according to the bilateral orbital rim boundaries, the piriform aperture boundaries, the bilateral temporal-sphenoidzygomatic arch gap boundaries, and the bilateral mandibularzygomatic arch gap boundaries; selecting any one of the seven holes as a current hole, and filling all holes except the current hole to obtain a topological ring surface; calculating the conformal mapping of the topological ring surface to a standard planar circular region by a dynamic Yamabe flow to map the current hole to a circular hole; until the mapping of the seven holes is completed, a two-dimensional unit disc with multiple circular holes conformally equivalent to the three-dimensional skull surface model is obtained.

4. The multi-modal based skull gender identification method of claim 3, wherein, The function formula of the dynamic Yamabe flow is: ； wherein, is a discrete conformal factor, for determining the degree of deformation of the curved surface in the conformal transformation; V is a set of triangle mesh vertices of a three-dimensional skull curved surface model, is a set of real numbers; is the target curvature of the i-th triangle mesh vertex, which refers to the vertex curvature value expected to be achieved; is the current discrete Gaussian curvature of the i-th triangle mesh vertex at time . 5.The multi-modal based skull gender identification method of claim 3, wherein, Filling all holes except the current hole to get a topological torus, comprising: Selecting any hole except the current hole as a hole to be filled; Determining the center point of the hole to be filled by a method based on the center of mass calculation; Connecting the center point to each vertex of the hole to be filled in turn to get a set of line segments; Determining new triangular patches according to the set of line segments and the boundary of the hole to be filled, combined with the constraint of the triangle inequality; Filling the hole to be filled according to the new triangular patches; Until all holes except the current hole are filled to get a topological torus. 6.The multi-modal based skull gender identification method of claim 1, wherein, Respectively performing normal calculation and coloring, vertex coordinate calculation and coloring on the two-dimensional unit disc to get a normal map and a spatial position map, comprising: Based on the two-dimensional unit disc, determining a set of triangular patches of the corresponding three-dimensional skull surface model; For each triangular patch in the set of triangular patches, based on any vertex, calculating two edge vectors, and then performing cross multiplication operation on the two edge vectors to get a normal vector; For each triangular patch in the set of triangular patches, calculating the patch barycentric coordinates through the arithmetic average of the coordinates of the three vertices; Respectively normalizing the normal vector and the patch barycentric coordinates; Mapping the normalized normal vector to RGB color values by interval conversion, and assigning the two-dimensional unit disc to get a normal map; Mapping the normalized patch barycentric coordinates to RGB color values by interval conversion, and assigning the two-dimensional unit disc to get a spatial position map.

7. The multi-modal based skull gender identification method of claim 1, wherein, Performing curvature calculation and coloring on the two-dimensional unit disc to get a curvature map, comprising: Based on the two-dimensional unit disc, determining a set of triangular patch vertices of the corresponding three-dimensional skull surface model; For each vertex in the set of triangular patch vertices, determining a geodesic neighborhood; Based on the vertices in the geodesic neighborhood, calculating the neighborhood area of the geodesic neighborhood; Based on the edges in the geodesic neighborhood, calculating the geometric properties of each edge; Based on the neighborhood area of the geodesic neighborhood and the geometric properties of each edge, calculating a curvature matrix; Performing eigenvalue decomposition on the curvature matrix to get principal curvatures; Determining the average curvature of the vertex according to the principal curvatures; Based on the average curvature of all vertices in the set of triangular patch vertices, extending the average curvature of the vertex to the triangular patch by the barycentric coordinate interpolation method to get triangular patch curvature information; Mapping all the triangular patch curvature information to RGB color values and assigning the two-dimensional unit disc to get a curvature map.

8. The multi-modal based skull gender identification method of claim 7, wherein, the curvature matrix of the vertex i is: ； wherein, is the area of the neighborhood, B is the geodesic neighborhood, denotes the edge measured geodesic neighborhood the effective length of the intercept; denotes the edge the angle between the normals of the two triangular patches, ranging from the sign decides the positive or negative; is the outer product matrix of the unit vectors with elements ; the unit vectors are calculated from the edge e as follows: ； ； where edge is determined by the end points and . is the Euclidean length of edge . 9.The multi-modal based skull gender identification method of claim 1, wherein, The normal-spatial position shared backbone network comprises a normal sub-network and a spatial position sub-network, and the weights of the normal sub-network and the spatial position sub-network are shared; The structure of the curvature independent backbone network, the normal sub-network and the spatial position sub-network is the same; The curvature independent backbone network comprises, in sequence, an input preprocessing module, a Patch division module, a first attention module, a second attention module, a third attention module and a fourth attention module; wherein the four attention modules each comprise, in sequence, a linear embedding submodule, two mobile window multi-head self-attention submodules; The input preprocessing module is configured to map the geometric and texture information of the curvature map to a regular pixel grid by a uniform sampling algorithm to obtain a curvature pixel map; the Patch division module is configured to divide the curvature pixel map into a plurality of image blocks; the linear embedding submodule is configured to perform mapping conversion of the image blocks from a pixel space to a feature space to obtain a sequence feature vector; and the mobile window multi-head self-attention submodule is configured to introduce a multi-head self-attention mechanism for the sequence feature vector, and determine a curvature feature block in combination with a window moving strategy.

10. The multi-modal based skull gender identification method of claim 9, wherein, The cross-modal bidirectional cross-attention network comprises four bidirectional guide modules; The bidirectional guide module is configured to: correspondingly extract the curvature feature, the normal feature and the spatial position feature in a feature extraction stage; Based on the normal feature and the spatial position feature, perform feature learning with the curvature feature as a reference to obtain a first learning query result; Through a residual connection, the first learning query result is fused with the normal feature and the spatial position feature respectively, and then fed back to the normal subnetwork and the spatial position subnetwork; based on the curvature feature, perform feature learning with the normal feature and the spatial position feature as references to obtain a second learning query result; Through a residual connection, the second learning query result is fused with the curvature feature, and then fed back to the curvature independent backbone network; The multi-scale fusion classification network is configured to: perform feature fusion and classification output on the final curvature feature output by the curvature independent backbone network, the final normal feature output by the normal subnetwork and the final spatial position feature output by the spatial position subnetwork.

Images