Method and system for obtaining three-dimensional ear mold model and storage medium

Abstract

The invention relates to the technical field of image processing, in particular to an ear mold three-dimensional model obtaining method and system and a storage medium. The method comprises the following steps: if acquisition parameters of an acquisition probe are lower than a preset acquisition parameter threshold value, controlling the acquisition probe to be inserted into an auditory meatus inlet, and acquiring an auricular conchae image; utilizing the auricular conchae image to identify the contour boundary of the auricular conchae, generating an entrance reference surface, and determining a auricular return fitting point; identifying an external auditory canal area according to the conchae contour boundary, collecting an external auditory canal image, extracting an auditory canal center line, measuring the sectional dimension, and constructing an eardrum curved surface; and performing hole repairing on the eardrum curved surface, performing smoothing processing, verifying the rationality of the eardrum structure, and exporting a standard three-dimensional eardrum file. According to the invention, the automatic generation of the standard three-dimensional ear mold is realized based on the image processing technology, and the ear mold reconstruction precision and the matching rate are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a method and system for acquiring a three-dimensional ear mold model and a storage medium. BACKGROUND

[0002] Traditional ear mold acquisition methods mostly rely on manual mold taking or single two-dimensional image measurement to construct ear molds. However, in complex ear canal structures, the morphology of the concha cavity and the external auditory canal is quite different, and manual or two-dimensional measurement is difficult to provide high-precision three-dimensional geometric information. Existing ear mold three-dimensional modeling methods mostly use fixed cross-sections or simple interpolation strategies, lack adaptive processing capabilities for longitudinal changes and surface continuity of the ear canal, and thus the model is prone to have holes, sharp edges or geometric distortion in complex eardrum and external auditory canal structures. In the boundary recognition process, traditional methods mostly rely on manual annotation or global thresholding, which cannot accurately capture the details and texture changes of the concha cavity profile, and thus the profile extraction is incomplete, affecting the accuracy of subsequent three-dimensional surface construction. The surface generation link is usually based on static cross-section fitting, which lacks joint constraints on continuity between cross-sections and local curvature changes, and is prone to local non-smoothness or error accumulation. Existing systems lack automated hole filling, smoothing optimization and morphological rationality detection mechanisms in eardrum surface repair and structure verification, and it is difficult to balance model integrity and anatomical accuracy. The overall three-dimensional ear mold acquisition accuracy and reliability need to be improved. SUMMARY

[0003] Therefore, it is necessary to provide a method and system for acquiring a three-dimensional ear mold model and a storage medium to solve at least one of the above technical problems.

[0004] To achieve the above-mentioned purpose, a method for acquiring a three-dimensional ear mold model comprises the following steps: Step S1: If the acquisition parameter of the acquisition probe is lower than the preset acquisition parameter threshold, the acquisition probe is inserted into the entrance of the ear canal to acquire the concha cavity image. Step S2: The concha cavity profile boundary is identified using the concha cavity image, an entrance reference surface is generated, and the antihelix fitting point is determined. Step S3: The external auditory canal region is identified according to the concha cavity profile boundary, the external auditory canal image is acquired, the ear canal centerline is extracted and the cross-sectional size is measured, and the eardrum surface is constructed. Step S4: The eardrum surface is repaired for holes, smoothed, verified for eardrum structure rationality, and a standard three-dimensional ear mold file is exported.

[0005] Preferably, the present specification also provides a computer-readable storage medium storing a computer program, which is executed by a processor to implement any one of the methods for acquiring a three-dimensional ear mold model.

[0006] Preferably, the present specification also provides an acquisition system for an ear mold three-dimensional model for performing the acquisition method for an ear mold three-dimensional model as described above, the acquisition system for an ear mold three-dimensional model comprising: An image acquisition module is configured to control the acquisition probe to be inserted into the entrance of the ear canal to acquire the concha cavity image if the acquisition parameter of the acquisition probe is lower than the preset acquisition parameter threshold; A fitting point determination module is configured to identify the contour boundary of the concha cavity by using the concha cavity image, generate an entrance reference surface, and determine the antihelix fitting point; An eardrum curved surface construction module is configured to identify the external ear canal region according to the contour boundary of the concha cavity, acquire the external ear canal image, extract the ear canal center line and measure the cross-sectional dimension, and construct the eardrum curved surface; An ear mold file export module is configured to perform hole repair and smoothing processing on the eardrum curved surface, verify the rationality of the eardrum structure, and export a standard three-dimensional ear mold file.

[0007] The present application has the following advantages: (1) By using the multi-stage three-dimensional reconstruction method based on the concha cavity image and the external ear canal image, the accurate spatial positioning of the ear canal and the eardrum is realized, the continuity and anatomical structure alignment of the ear mold three-dimensional model in the depth direction are ensured, and the accuracy and reliability of the three-dimensional ear mold construction are improved; (2) The contour recognition strategy of fusing edge features and texture features is adopted to perform weighted scoring and noise removal on the contour of the concha cavity, so that the accurate boundary extraction of the complex ear canal structure is realized, the contour recognition accuracy is improved, and the error caused by image noise is reduced; (3) In the process of constructing the external ear canal cross-sectional sequence and generating the eardrum curved surface, the cross-sectional continuity and curved surface smoothness are effectively maintained by combining the depth reference line and the curvature change constraint, the accurate restoration of the eardrum micro-geometric features is realized, and the limitations of the traditional cross-sectional fitting method that is prone to holes and sharp distortion are broken through; (4) By performing automatic hole repair, smoothing processing and structure rationality verification on the eardrum curved surface, it is ensured that the final three-dimensional ear mold model is complete and conforms to the anatomical specification, high-quality data basis is provided for personalized ear mold design, hearing aid devices or ear medical research, and the applicability and repeatability of the ear mold acquisition method are effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0008] Other features, objects and advantages of the present application will become more apparent from the following detailed description of non-limiting embodiments made with reference to the accompanying drawings: Fig. 1 The present application is a step flowchart for an acquisition method for an ear mold three-dimensional model; Fig. 2 The present application is a module schematic diagram of an acquisition system for an ear mold three-dimensional model; Fig. 3 Fig. 1 is a schematic diagram of a three-dimensional model of an ear mold according to the present application; The implementation, functional features and advantages of the present application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0009] The technical method of the present application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all the other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0010] In addition, the accompanying drawings are only schematic diagrams of the present application, and are not necessarily drawn to scale. The same reference signs in the drawings represent the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities, which do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0011] It should be understood that although the terms “first”, “second” and the like can be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, without departing from the scope of the exemplary embodiments, a first element can be referred to as a second element, and similarly a second element can be referred to as a first element. The term “and / or” as used herein includes any and all combinations of one or more of the associated listed items.

[0012] To achieve the above-mentioned object, please refer to Figs. 1 to 3 The present application provides an acquisition method for a three-dimensional model of an ear mold, comprising the following steps: Step S1: If the acquisition parameter of the acquisition probe is lower than the preset acquisition parameter threshold, control the acquisition probe to be inserted into the entrance of the ear canal to acquire the image of the concha cavity; In an embodiment, when the acquisition parameter of the acquisition probe is lower than the preset threshold, the probe is controlled to be inserted into the entrance of the ear canal, to maintain a stable pushing posture, and to acquire the image of the concha cavity frame by frame through an endoscope or a high-definition optical camera, to ensure that each frame of image is consistent in time domain and clear, to provide basic data for subsequent contour boundary identification. The probe is kept axially aligned with the ear canal during the acquisition process to avoid image distortion or obstruction.

[0013] In another embodiment, assuming that the acquisition probe resolution is 1920x1080, the frame rate is 30Hz, 60 frames are continuously acquired, the acquisition time of each frame is about 33ms, the acquisition depth is not more than 25mm, the acquisition distance is about 15mm, and high-quality original image data is generated, which is used for subsequent fine boundary analysis.

[0014] Step S2: recognizing the concha cavity contour boundary by using the concha cavity image, generating an entrance reference plane, and determining the antihelix fitting point; In an embodiment, the acquired concha cavity image is converted into a grayscale image, and the gray scale gradient amplitude is extracted as an edge feature; the gray scale difference is calculated in each pixel neighborhood, and the direction of the line connecting the maximum and minimum points of the gray scale difference is taken as a texture feature; the edge feature and the texture feature are fused to generate a weighted score, high-score pixel points are screened and connected into a continuous curve, and isolated noise points are removed at the same time to obtain a complete concha cavity contour boundary. The contour boundary is equidistantly sampled, an initial plane is fitted and generated, a normal vector is calculated and rotated and corrected with the ear canal depth direction to form an entrance reference plane; the contour is projected onto the reference plane, and the point with the smallest curvature and facing the ear canal opening is recognized as the antihelix fitting point.

[0015] In another embodiment, assuming that there are about 1500 contour candidate points, 1200 points are retained after scoring and noise removal; the initial reference plane normal deviation is about 3°, and the deviation with the ear canal depth direction is less than 1° after rotation correction; finally, 3 antihelix fitting points are recognized, which are located at the front, middle and rear segments of the concha cavity entrance respectively, and the three-dimensional coordinate accuracy of each point is about 0.2mm.

[0016] Step S3: identifying the external ear canal region according to the concha cavity contour boundary, acquiring the external ear canal image, extracting the ear canal centerline and measuring the cross-sectional size, and constructing the eardrum curved surface; In an embodiment, the concha cavity contour boundary is subjected to spatial extension analysis, and the depth image slices are extracted layer by layer along the boundary normal, the cross-sectional gray scale and the closedness of each layer are calculated, and the extension direction of the external ear canal is judged; the cross-sectional contour is tracked frame by frame along the depth direction, the cross-sectional size and the centroid position are measured, and the continuous and smooth eardrum curved surface is constructed by connecting the consecutive cross sections through spline interpolation and sealing the closed cross-sectional contour at the eardrum end.

[0017] In another embodiment, assuming that 20 layers of slices are extracted in the depth direction, the cross-sectional diameter of each layer is 6-8mm, and the centroid offset is not more than 0.5mm; there are 4 closed cross sections at the eardrum end, and after spline interpolation, the total number of points of the curved surface is about 12000 points, and the maximum cross-sectional curvature change is about 0.15mm, which ensures that the eardrum curved surface is continuous and consistent with the anatomical morphology. -1

[0018] Step S4: hole repair and smoothing processing are performed on the eardrum curved surface, the rationality of the eardrum structure is verified, and a standard three-dimensional ear mold file is derived. ​

[0019] In one embodiment, the missing area of ​​the eardrum surface is identified, the hole is locally filled with triangles, and then Laplacian smoothing is used to make the local curvature continuous. The eardrum thickness, surface continuity and normal direction are checked by the structural rationality verification module. After confirming that the structure is reasonable, a standard three-dimensional ear mold file (such as STL format) is exported for ear mold production or product design.

[0020] In another embodiment, six holes were detected on the curved surface, with the largest hole diameter being approximately 2 mm; the smoothing process was iterated three times, with a window radius of 1.5 mm each time, resulting in a maximum normal deviation of less than 0.3 mm after smoothing; the exported 3D ear mold file contained approximately 15,000 triangular facets, with a total volume of approximately 1.2 cm³. 3 It can be used directly for 3D printing or digital analysis.

[0021] Of particular importance, step S4, which involves repairing the hole in the curved surface of the eardrum, includes: Identify unclosed regions in the tympanic membrane surface; construct boundary loops at the boundary points of the unclosed regions, and generate triangular mesh fragments based on the boundary loops to fill the holes; In one embodiment, boundary detection is performed on the constructed eardrum surface to mark the set of boundary points of the unclosed regions. For sets Closed boundary loops are constructed sequentially to generate triangular mesh fragments to fill the unclosed hole areas. After filling, the triangular mesh fragments are merged with the original eardrum surface to form a continuous and complete eardrum surface, providing a foundation for subsequent smoothing and export of the 3D model.

[0022] In another embodiment, it is assumed that five unclosed boundary regions are detected, with the number of boundary points at each region being [18, 24, 15, 20, 22]. For each boundary loop, a corresponding triangular mesh segment is generated, with the number of segments being [12, 16, 10, 14, 13]. Simultaneously with the generation of the triangular mesh segments, the normals of the segment vertices are interpolated to obtain a set of interpolated normal vectors. The interpolated triangular mesh segments are then subjected to topological consistency checks, including boundary continuity, vertex connection integrity, and face normal consistency. The check results show that all five filled regions pass the topological consistency check, successfully constructing a complete eardrum surface.

[0023] During the filling process, normal interpolation is performed on the triangular mesh segments, and the topological consistency of the triangular mesh segments after normal interpolation is verified.

[0024] In one embodiment, linear weighted interpolation is used to calculate the normal of each newly generated vertex, and a weighted average is performed with the normals of neighboring vertices to ensure a smooth transition between the normal of the filled region and the original surface. The topology of the generated triangular mesh is then checked, including vertex repetition, boundary closure, and triangular face orientation consistency, to confirm the integrity of the topological structure.

[0025] In another embodiment, assuming that the filled triangular mesh segment generates 75 triangular faces in total, the average deviation of the vertex normal interpolation result in x, y, z three directions is not more than 0.02, and after topological verification, it is found that there are 2 vertices with slight abnormalities. After adjusting the position of the vertices locally, all segments meet the topological consistency requirements, thereby completing the eardrum hole repair and smoothing processing.

[0026] Especially important is that the verification of the rationality of the eardrum structure in step S4 includes: Based on the smoothed eardrum surface, the geometric continuity index of the eardrum surface is calculated to detect whether there is an abnormal sharp angle; In an embodiment, the smoothed eardrum surface is divided into a plurality of small facets, the included angle between the vertices of each facet and the normal change rate of the facet are calculated, and the continuity index is generated If the continuity index is greater than a preset threshold (such as 30°), it is determined that there is an abnormal sharp angle. By traversing all the facets, the abnormal sharp region is marked to provide a basis for subsequent morphology verification and hole repair.

[0027] In another embodiment, assuming that the eardrum surface is composed of about 1200 triangular facets, the continuity index of each facet is calculated , and the range is [5°, 28°], wherein only 8 facets are slightly higher than the threshold 30°, all of which are located in the edge region of the eardrum, indicating that the overall surface continuity is good and there is no obvious sharp abnormality.

[0028] The consistency of the eardrum surface with the longitudinal direction of the ear canal is judged by calculating the included angle between the center line of the ear canal and the normal vector of the eardrum surface. The curvature distribution of the eardrum surface is globally analyzed. If the curvature changes smoothly and is within a preset curvature threshold range, the morphology is determined to be reasonable. Otherwise, return to hole repair.

[0029] In an embodiment, the sampling points are uniformly sampled along the center line of the ear canal, and the normal vectors are calculated on the corresponding eardrum surface. The included angle between the normal vector and the direction of the center line is calculated for each sampling point , if (such as 15°), it is determined that the direction of the point is reasonable. Then the curvature distribution is calculated, if the global curvature changes smoothly and is within a preset curvature range , it is determined that the eardrum morphology is reasonable; otherwise, return to step S1 or perform hole repair.

[0030] In another embodiment, assuming that 100 points are sampled along the center line, the included angle The range of the global average included angle is [2°, 14°], and the global average included angle is about 8°, which meets the requirement of direction consistency; the curvature distribution range is about , and the average is about , which is smooth and within the preset threshold range, so the eardrum curvature form is determined to be reasonable, and further repair is not needed.

[0031] Preferably, the ear cavity image is converted into a gray-scale image, and the gray-scale gradient amplitude is extracted as an edge feature. The ear cavity image is converted into a gray-scale image; the gray-scale gradient amplitude of each pixel point in the gray-scale image is calculated as an edge feature, which is used for preliminary contour recognition. In an embodiment, the collected ear cavity image is converted into a gray-scale image; the gray-scale gradient amplitude of each pixel point in the gray-scale image is calculated as an edge feature, which is used for preliminary contour recognition. The edge feature can reflect the high gray-scale change area of the ear cavity contour in the image, thereby providing candidate points for texture feature calculation.

[0032] In another embodiment, assuming that the resolution of the collected image is 1024x1024, the gray-scale gradient amplitude of about 1048576 pixels in each frame is calculated. The gradient threshold is set to 0.2, and only about 250,000 high-gradient pixel points are selected as edge candidate points to reduce noise interference; further, about 50,000 isolated points are removed according to connectivity analysis, and about 200,000 continuous edge pixels are obtained, which provide a basis for subsequent texture feature calculation.

[0033] For any pixel point in the gray-scale image, the gray-scale difference between each pixel point in the neighborhood and the center pixel point is calculated; the direction of the line connecting the pixel point with the maximum gray-scale difference to the pixel point with the minimum gray-scale difference in the neighborhood is taken as the texture feature of the center pixel point. In an embodiment, any pixel point in the gray-scale image is taken as the center, a neighborhood (for example, 5x5 pixels) is selected, and the gray-scale difference between each pixel point in the neighborhood and the center pixel point is calculated; the direction of the line connecting the pixel point with the maximum gray-scale difference to the pixel point with the minimum gray-scale difference in the neighborhood is taken as the texture feature of the center pixel point. The texture feature can enhance the reliability of the edge contour and provide auxiliary information for the sensitive points of the curved surface change.

[0034] In another embodiment, assuming that the neighborhood is 7x7 pixels, the texture feature vector of about 800,000 pixel points in each frame is calculated. The texture direction is normalized, and about 500,000 pixel points with a deviation of less than 15° from the average direction of the neighborhood are selected. In combination with the edge points in step S1, only about 300,000 pixel points that have both edge features and consistent texture directions are retained, which are used for subsequent contour fusion and spatial mapping to improve the three-dimensional contour accuracy.

[0035] The edge feature and the texture feature are used to determine the ear cavity contour boundary.

[0036] In an embodiment, the edge feature and the texture feature are fused by weighting with a ratio of 60% for the edge feature and 40% for the texture feature, and the contour confidence of each pixel point is evaluated; then the pixel points are screened to remove isolated or low-confidence points, and a complete contour boundary is generated through connectivity analysis. In another embodiment, assuming that the number of pixel points of the edge feature and the texture feature is about 350,000 and 300,000 respectively, about 500,000 pixel points are determined as the contour points with high confidence after weighting, about 100,000 isolated points are removed, and finally a contour of about 400,000 continuous pixel points is formed, and the spatial accuracy can reach 0.15 mm.

[0037] In another embodiment, assuming that the resolution of the concha cavity image is 1024x1024, and the spatial size of each pixel is about 0.1 mm x 0.1 mm; after weighting the edge feature and the texture feature with a ratio of 60% and 40%, in the contour confidence map generated, the high-confidence pixel points account for about 4.8% of the total pixel points; after removing isolated points and noise points through connectivity analysis, a contour of about 400,000 continuous pixel points is finally formed. Refinement can be performed on the key points of the contour, such as adding neighborhood interpolation and smoothing operation to the corners and curved areas, so that the continuity and smoothness of the contour curve in the tortuous area are improved, and thus the overall spatial accuracy is maintained within 0.15 mm.

[0038] Preferably, the edge feature and the texture feature are used to determine the contour boundary of the concha cavity, and the method comprises the following steps: The weighted results of the edge feature and the texture feature are fused to generate an edge intensity score; in the edge intensity score, the pixel points satisfying a preset edge intensity threshold value are selected as the candidate points of the concha cavity contour, and are connected into a continuous edge curve, and noise points are removed, so as to serve as the contour boundary of the concha cavity.

[0039] In an embodiment, the edge feature and the texture feature are fused by weighting with a ratio of 60% and 40%, and the contour confidence score of each pixel point is calculated; then the pixel points with a score higher than a threshold value of 0.7 are selected as candidate points. The candidate points are subjected to connectivity analysis to remove isolated points and noise points with less than 5 pixel points, and then a complete continuous contour is formed through curve connection. The final contour is composed of about 350,000 continuous pixel points, the spatial accuracy can reach 0.15 mm, and the main curve and concave-convex structure of the concha cavity can be clearly displayed.

[0040] In another embodiment, assuming that the original concha gray image resolution is 1024x1024, the edge feature pixel points are about 350,000, and the texture feature pixel points are about 300,000. In the edge intensity score generated after weighted fusion, high-confidence pixel points account for about 15% of the total pixels, and about 500,000 points are determined as candidate contour points. After removing about 100,000 isolated points and small cluster points, a contour curve of about 400,000 continuous pixel points is finally formed. Further local smoothing processing is performed on the contour key areas (such as the tip of the concha and the curved inner corner), so that the contour continuity and smoothness are improved, and the overall spatial accuracy is still maintained at about 0.15 mm, which can be used for subsequent three-dimensional modeling or dissection analysis.

[0041] Preferably, in step S2, the entrance reference surface is generated, and determining the concha attachment point comprises: After obtaining the concha contour boundary, equidistant sampling is performed to obtain discrete boundary coordinates; an initial reference surface is fitted using the discrete boundary coordinates, and a normal vector of the plane is calculated; In an embodiment, every 0.5 mm of the extracted concha contour curve is sampled to obtain about 500 discrete boundary coordinate points. The least squares plane fitting method is used to fit these discrete points into a plane to obtain an initial reference surface. By calculating the normal vector N0 of the fitted plane, the overall direction of the concha contour can be represented. The initial reference surface is used for subsequent entrance correction and concha attachment point positioning, and provides a stable reference for maintaining spatial accuracy.

[0042] In another embodiment, assuming that the total length of the concha contour is about 250 mm, equidistant sampling is performed at an interval of 0.5 mm to obtain 500 discrete points. The initial reference surface obtained by fitting has a normal vector that forms an angle of about 12° with the ear canal axial reference line, which exceeds the preset threshold of 10°, and needs to be corrected by rotation. During the fitting process, the distribution of local residual points can be recorded for subsequent optimization of the smoothness of the reference surface.

[0043] The normal vector is compared with the ear canal axial reference line. If the deviation between the two exceeds the preset angle threshold, the initial reference surface is corrected by rotation to make it orthogonal to the ear canal longitudinal direction to form an entrance reference surface; In an embodiment, the angle θ between the normal vector N0 of the initial reference surface and the ear canal longitudinal direction vector A is calculated. When θ is greater than 10°, the initial reference surface is rotated by θ degrees around the normal vector projection axis to make the normal vector orthogonal to the ear canal longitudinal direction, and a corrected entrance reference surface is obtained. After rotation correction, local smoothing processing is performed on the sampling points near the contour on the reference surface to ensure that the entrance plane is continuous and has no protrusions.

[0044] In another embodiment, it is assumed that the initial fitting reference plane normal vector is deviated from the ear canal axis by 12°, and the deviation is reduced to about 1° after rotation correction. During the correction process, about 50 sampling points on the entrance contour are finely adjusted locally to ensure that the plane closely fits the contour while maintaining continuity and smoothness. The corrected entrance reference plane can be used to determine the position of the ear return fitting point, with a spatial accuracy of ≤0.2 mm.

[0045] Determine the ear return fitting point using the entrance reference plane.

[0046] In an embodiment, the contour points projected on the entrance reference plane are used to analyze the local smooth area, and the smooth area close to the contour surface is selected as the fitting point to ensure the stability of the fitting and maximize the coverage area. A final set of fitting points is determined for ear return installation or further analysis.

[0047] In another embodiment, it is assumed that there are about 400 contour points projected on the entrance reference plane, and the 50 points with the smallest local convexity are selected as the fitting point candidates through curvature analysis. Finally, about 30 ear return fitting points are determined through uniformity screening of spatial distribution. This set of fitting points covers about 85%-90% of the area of the concha cavity entrance, ensuring the stability and repeatability of the fitting, and facilitating subsequent ear return customization or assembly.

[0048] Preferably, determining the ear return fitting point using the entrance reference plane comprises: Projecting the concha cavity contour boundary to the entrance reference plane to form a closed projection curve; In an embodiment, first, a set of three-dimensional boundary points (about 500 sampling points) of the concha cavity contour is obtained, and these points are projected onto the entrance reference plane along the normal vector direction to obtain a closed projection curve. Then, the curve is smoothed to eliminate local sharp peaks or noise points, ensuring that the closed curve is continuous and smooth in shape. This closed curve is used to judge the opening area and convex-concave structure of the concha cavity, providing a stable two-dimensional reference for fitting point selection.

[0049] In another embodiment, it is assumed that the total length of the concha cavity contour is about 250 mm, and after projection, 500 points of the closed projection curve are obtained. The closed curve is elliptical, with a major axis of about 20 mm and a minor axis of about 15 mm. To improve accuracy, the closed curve is smoothed using a cubic spline, so that the local maximum curvature of the curve does not change by more than 0.5 mm. This closed projection curve serves as the basis for subsequent fitting point determination.

[0050] Identify the point in the closed projection curve with the smallest curvature radius and oriented towards the ear canal opening direction, and determine it as the ear return fitting point. Output the position coordinates of the point in the three-dimensional coordinate system.

[0051] In an embodiment, the local radius of curvature is calculated for each point of the closed projection curve, and in combination with the projection direction of the curve point to the ear canal axis, the point with the minimum radius of curvature and pointing to the opening direction of the ear canal is screened out as the ear return fitting point. Then the point is inversely mapped back to the three-dimensional space along the normal vector to obtain the final position coordinates in the three-dimensional coordinate system for the ear return installation or further analysis.

[0052] In another embodiment, assuming there are 500 points on the closed projection curve, about 5 points with the minimum curvature are obtained by calculating the local curvature. In combination with the screening condition of pointing to the opening direction of the ear canal, the point with the minimum curvature is finally selected as the ear return fitting point. The three-dimensional coordinates of the point are (X, Y, Z) ≈ (12.4 mm, -8.7 mm, 6.3 mm). In order to enhance the stability, the average position of the neighboring points (±2 mm) with the minimum curvature can also be calculated to obtain the smoothed fitting point coordinates (12.5 mm, -8.6 mm, 6.2 mm) for the ear return design or custom fitting device.

[0053] Preferably, the step S3 of identifying the outer ear canal region according to the concha cavity profile boundary comprises: The spatial extension analysis is performed on the concha cavity profile boundary to determine the boundary contraction direction; in the direction, continuous image slices are extracted to the longitudinal region, and the cross-sectional gray scale of the continuous image slices is calculated; if the cross-sectional gray scale continuously decreases in the longitudinal direction, it is determined that the direction is the extension direction of the outer ear canal; In an embodiment, the three-dimensional boundary point set (about 500 sampling points) of the concha cavity profile is first obtained, and the local contraction / protrusion trend thereof in the three-dimensional space is analyzed. Continuous image slices are extracted to the longitudinal region along the contraction direction, and the cross-sectional gray scale value is obtained with a resolution of about 0.1 mm x 0.1 mm per slice. By analyzing the change of each cross-sectional gray scale in the longitudinal direction, if the gray scale value continuously decreases, it is determined that the direction is the extension direction of the outer ear canal. This method can be used to automatically identify the spatial relationship between the ear canal entrance and the deep part, and provide a directional reference for subsequent cross-sectional sequence construction.

[0054] In another embodiment, assuming that the concha cavity profile length is about 25 mm, 50 continuous image slices are extracted along the longitudinal direction, and the average cross-sectional gray scale of each slice is: [200, 195, 189, 183, 178, 172, 168, 162, 158, …, 40] (unit gray scale value 0-255). It is found that the gray scale value shows a monotonic decreasing trend, so it is determined that the direction is the extension direction of the outer ear canal. Through this direction, the longitudinal positioning can be determined to provide an accurate reference for cross-sectional sequence construction.

[0055] The contour boundary of the concha cavity is taken as a starting cross-section, and the cross-section contour is tracked frame by frame along the extension direction of the external auditory canal to establish a cross-section sequence; in the cross-section sequence, the centroid position of each cross-section is calculated and connected to form a longitudinal reference line, and non-continuous boundary points are removed by monitoring the cross-section closure of the cross-section periphery; the cross-section sequence and the longitudinal reference line are jointly defined as the external auditory canal region.

[0056] In an embodiment, the contour boundary of the concha cavity is taken as a starting cross-section, and the cross-section contour is tracked frame by frame along the extension direction of the external auditory canal to obtain a continuous cross-section sequence. For each cross-section, the centroid coordinates are calculated, and the centroids are connected to form a longitudinal reference line. At the same time, the cross-section contour closure is monitored, and regions with incomplete cross-section boundaries or isolated points are removed to ensure the continuity and accuracy of the sequence. Finally, the cross-section sequence and the longitudinal reference line are jointly defined as the external auditory canal region, which is used for subsequent analysis of the ear return fitting point or the ear canal shape.

[0057] In another embodiment, it is assumed that the cross-section sequence is collected along the extension direction of the external auditory canal for 50 frames, and each frame of cross-section contour contains about 100-120 sampling points. The centroid position of each cross-section is calculated and connected in turn to form a longitudinal reference line. The cross-section closure is monitored to find that about 5 frames have isolated points or broken boundaries, and the sequence is left with 45 frames after removing these points. The three-dimensional length of the external auditory canal region formed by the sequence is about 25 mm, and the average cross-section diameter is about 6-8 mm, which provides a clear spatial definition for subsequent ear return positioning or simulation modeling.

[0058] Preferably, the contour boundary of the concha cavity is analyzed for spatial extension, and the shrinkage direction of the boundary is determined, including: Sampling points are selected on the contour boundary of the concha cavity, and the extension vector of each sampling point in the normal direction is calculated; during the advancement of the extension vector, image slices are extracted layer by layer and the cross-sectional area of each slice is calculated; In an embodiment, first, sampling points are uniformly selected on the contour boundary of the concha cavity (about 200), and the local normal direction of each sampling point is calculated as the initial direction of the extension vector. Along the extension vector, image slices are extracted layer by layer in the longitudinal direction (each layer is about 0.1 mm thick), and the cross-sectional area of each slice is calculated. By analyzing the trend of the cross-sectional area with the extension depth, the spatial extension characteristics of each sampling point can be obtained, which provides a basis for subsequent determination of the shrinkage direction of the boundary.

[0059] In another embodiment, assuming that the 10 layers of image slices extend along the normal direction of a certain sampling point, each layer has a thickness of 0.1 mm, and the corresponding cross-sectional areas are: [35, 33, 30, 28, 25, 23, 21, 19, 17, 15]. It is observed that the cross-sectional area gradually decreases in the longitudinal direction, and the boundary direction of the extension direction of the sampling point is determined as the preliminary shrinkage direction. Repeat the process for 200 sampling points, and if the preliminary shrinkage directions of most sampling points (about 160) are consistent, then the direction is determined as the boundary shrinkage direction of the overall concha cavity profile.

[0060] If the cross-sectional area gradually decreases with the extension depth, then the extension direction is determined as the boundary preliminary shrinkage direction. In the results of the extension vectors of multiple sampling points, if the boundary preliminary shrinkage directions of most vectors are consistent, then the direction is determined as the boundary shrinkage direction.

[0061] In an embodiment, the extension vectors of all sampling points and their corresponding cross-sectional area change trends are statistically analyzed, and the area reduction rate of each vector in the normal direction is calculated. If the area continuously decreases with the depth in most sampling points, then the direction of these vectors is determined as the boundary preliminary shrinkage direction. Through statistical analysis and clustering of the preliminary shrinkage direction, the vector direction with the highest direction consistency is selected as the overall boundary shrinkage direction of the concha cavity, providing a clear direction reference for subsequent external auditory canal extension tracking or cross-sectional sequence construction.

[0062] In another embodiment, assuming that the extension vectors of 160 sampling points out of 200 sampling points have a continuous decrease in cross-sectional area along the normal direction, and the average area reduction rate is / layer; the remaining 40 sampling points have intermediate local increases or noise fluctuations. After direction clustering analysis, it is found that the directions of the 160 sampling points differ by less than ±5°, and the direction is determined as the overall boundary shrinkage direction of the concha cavity. This shrinkage direction can be used to guide the subsequent extraction of cross sections along the external auditory canal direction and the construction of the ear canal region.

[0063] Preferably, in step S3, the external auditory canal image is collected, the ear canal centerline is extracted, and the cross-sectional size is measured to construct the eardrum surface, which includes: The external auditory canal image is collected, the ear canal centerline is extracted, and the cross-sectional fitting of the external auditory canal image is performed. The cross-sectional profile is extracted and the cross-sectional size is calculated. The cross-sectional profiles are connected in longitudinal order by spline interpolation, and the closed profile with high curvature change at the eardrum end is identified. The closed profile is sealed to construct the eardrum surface.

[0064] In an embodiment, the external ear canal is continuously collected by using a high-resolution ear canal endoscope or CT image (resolution about 0.1 mm) to generate a sequence of continuous images in the longitudinal direction. The center line of the ear canal is extracted by an image processing algorithm (such as gray centroid tracking or pipeline fitting), and a cross-sectional image is taken every 0.2 mm along the center line to ensure the continuity of the cross section and the reconstruction accuracy. Threshold segmentation (such as Canny) is applied to each cross section to extract the closed contour, calculate the cross-sectional parameters such as contour area, perimeter and major and minor axis size. The cross-sectional contours are connected by spline interpolation in the longitudinal order to ensure smooth transition between adjacent cross sections. For the eardrum end cross section, identify the closed contour with prominent curvature change, and seal the necessary position to construct the complete eardrum surface. The point cloud of all cross-sectional contours is spline interpolated and spatially registered along the ear canal center line to generate a continuous three-dimensional eardrum surface mesh. The smoothness of the closed contour at the eardrum end and the continuity of the surface are verified by surface normal calculation and curvature analysis.

[0065] In another embodiment, assuming that the longitudinal range of the collected external ear canal image is 20 mm, a total of 100 cross sections are taken every 0.2 mm; a smooth curve is obtained by center line fitting, and the three-dimensional coordinate range is , which ensures that the center line is smooth and continuous in the longitudinal direction. Assuming that among the 100 cross-sectional contours, the 95th to 100th frames are located at the eardrum end, the cross-sectional areas are , the perimeters are ; the high change point of the curvature appears as a sharp peak at the 98th contour, and the closed contour is completed by local sealing. After spline interpolation connection, the eardrum surface height is about 5 mm, which ensures the continuous surface reconstruction from the external ear canal to the eardrum end. Assuming that the eardrum surface mesh after spline interpolation connection contains about 1200 vertices, the maximum local normal deviation is 2.1°, and the average normal deviation is 0.8°, the surface continuity and smoothness meet the requirements of eardrum shape reconstruction, and can be used for subsequent three-dimensional ear canal modeling or surgical navigation simulation.

[0066] Preferably, the present specification also provides an acquisition system for a three-dimensional model of an ear mold for performing the acquisition method for a three-dimensional model of an ear mold as described above, the acquisition system for a three-dimensional model of an ear mold comprising: An image acquisition module 101 is configured to control the insertion of the acquisition probe into the entrance of the ear canal if the acquisition parameter of the acquisition probe is lower than the preset acquisition parameter threshold, and to acquire the concha cavity image; A fitting point determination module 102 is configured to identify the concha cavity contour boundary by using the concha cavity image, to generate an entrance reference surface, and to determine the retroconcha fitting point; An eardrum surface construction module 103 is configured to identify the external ear canal region according to the concha cavity contour boundary, to collect the external ear canal image, to extract the ear canal center line and to measure the cross-sectional size, and to construct the eardrum surface; The ear mold file export module 104 is configured to perform hole repairing and smoothing processing on the eardrum surface, verify rationality of the eardrum structure, and export a standard three-dimensional ear mold file.

[0067] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, the scope of the application being defined by the appended claims and not by the above description, and all changes falling within the meaning and range of equivalency of the elements of the patent file are therefore intended to be embraced within the present application.

[0068] The foregoing is considered as illustrative only of the principles of the application. Numerous modifications and adaptations will be apparent to those skilled in the art in view of the above descriptions of the embodiments. This specification and the embodiments are not therefore to be taken in a limiting sense, but are made within the spirit of the scope of the application, and the general inventive concept as defined herein.

Claims

1. A method for obtaining a three-dimensional model of an ear mold, characterized in that, The method includes the following steps: Step S1: If the acquisition parameters of the acquisition probe are lower than the preset acquisition parameter threshold, control the acquisition probe to be inserted into the ear canal entrance to acquire an image of the concha cavity; Step S2: Use concha images to identify the concha contour boundary, generate the entrance reference plane, and determine the resonator fitting point; Step S3: Identify the external auditory canal region based on the concha contour boundary, acquire images of the external auditory canal, extract the center line of the ear canal and measure the cross-sectional dimensions, and construct the tympanic membrane surface; Step S4: Repair the holes in the curved surface of the eardrum and smooth it out. Verify the rationality of the eardrum structure and export the standard three-dimensional ear mold file.

2. The method for obtaining a three-dimensional model of an ear mold according to claim 1, characterized in that, Step S2, which involves identifying the conchae contour boundary using conchae images, includes: The concha cavity image was converted into a grayscale image, and the grayscale gradient magnitude was extracted as edge features. For any pixel in a grayscale image, calculate the grayscale difference between each pixel in the neighborhood and the center pixel; take the line connecting the pixel with the largest grayscale difference to the pixel with the smallest grayscale difference as the texture feature of the center pixel. Edge features and texture features are used to determine the concha cavity contour boundary.

3. The method for obtaining a three-dimensional model of an ear mold according to claim 2, characterized in that, Determining the conchae contour boundary using edge and texture features includes: The weighted results of fusing edge features and texture features are used to generate an edge strength score. Pixels that meet the preset edge strength threshold are selected from the edge strength score as candidate points for the concha contour and connected to form a continuous edge curve. Noise points are removed to serve as the boundary of the concha contour.

4. The method for obtaining a three-dimensional model of an ear mold according to claim 1, characterized in that, Step S2 involves generating the inlet reference plane and determining the earpiece contact point, including: After obtaining the conchae contour boundary, equidistant sampling is performed to obtain discrete boundary coordinates; the initial reference plane is fitted using the discrete boundary coordinates, and the normal vector of the plane is calculated. The normal vector is compared with the axial baseline of the ear canal. If the deviation between the two exceeds the preset angle threshold, the initial reference plane is rotated and corrected to make it orthogonal to the longitudinal direction of the ear canal, so as to form the entrance reference plane. The earpiece contact point is determined using the inlet reference plane.

5. The method for obtaining a three-dimensional model of an ear mold according to claim 4, characterized in that, Determining the earpiece contact point using the inlet reference plane includes: The conchae cavity contour boundary is projected onto the entrance reference plane to form a closed projection curve; Identify the point with the smallest radius of curvature in the closed projection curve that faces the ear canal opening, determine it as the in-ear contact point, and output the position coordinates of this point in the three-dimensional coordinate system.

6. The method for obtaining a three-dimensional model of an ear mold according to claim 1, characterized in that, Step S3, which involves identifying the external auditory canal region based on the conchae contour boundary, includes: Spatial extension analysis is performed on the boundary of the concha cavity to determine the direction of boundary contraction; in this direction, continuous image slices are extracted into the depth region, and the cross-sectional grayscale of the continuous image slices is calculated; if the cross-sectional grayscale continuously decreases with the depth direction, then this direction is determined to be the extension direction of the external auditory canal. Using the conchae contour boundary as the starting section, the section contour is traced frame by frame along the extension direction of the external auditory canal to establish a section sequence. In the section sequence, the centroid position of each section is calculated and connected to form a depth reference line. At the same time, the section closure around the section is monitored, and discontinuous boundary points are eliminated. The section sequence and the depth reference line are defined together as the external auditory canal region.

7. The method for obtaining a three-dimensional model of an ear mold according to claim 6, characterized in that, Spatial extension analysis of the conchae contour boundary was performed to determine the direction of boundary contraction, including: Sampling points are selected on the boundary of the concha cavity contour, and the extension vector of each sampling point in the normal direction is calculated; as the extension vector advances, image slices are extracted layer by layer and the cross-sectional area of ​​the slices is calculated. If the cross-sectional area gradually decreases with the extension depth, then the extension direction is determined to be the initial contraction direction of the boundary; if the initial contraction direction of the boundary is consistent among the extension vector results of multiple sampling points, then the direction is determined to be the boundary contraction direction.

8. The method for obtaining a three-dimensional model of an ear mold according to claim 1, characterized in that, Step S3 involves acquiring images of the external auditory canal, extracting the centerline of the ear canal, measuring the cross-sectional dimensions, and constructing the tympanic membrane surface, including: Acquire images of the external auditory canal, extract the center line of the ear canal, and perform cross-sectional fitting on the external auditory canal images to extract the cross-sectional contour and calculate the cross-sectional dimensions. Connect the cross-sectional contours by spline interpolation in the order of depth, and identify the closed contour with high curvature at the eardrum end. Seal the closed contour to construct the eardrum surface.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for obtaining a three-dimensional model of an ear mold as described in any one of claims 1 to 8.

10. A system for acquiring a three-dimensional model of an ear mold, characterized in that, For performing the method for acquiring a three-dimensional model of an ear mold as described in claim 1, the system for acquiring a three-dimensional model of an ear mold comprises: The image acquisition module is used to control the acquisition probe to be inserted into the ear canal entrance to acquire images of the concha cavity if the acquisition parameters of the acquisition probe are lower than the preset acquisition parameter threshold. The fitting point determination module is used to identify the concha cavity contour boundary using concha cavity images, generate an entrance reference plane, and determine the ear resonator fitting point; The eardrum surface construction module is used to identify the external auditory canal region based on the concha cavity contour boundary, acquire external auditory canal images, extract the ear canal centerline and measure cross-sectional dimensions to construct the eardrum surface. The earmold file export module is used to repair holes in the curved surface of the eardrum, smooth the surface, verify the rationality of the eardrum structure, and export a standard three-dimensional earmold file.

Images