Image processing method, device and electronic equipment

Abstract

The present disclosure relates to an image processing method, device and electronic device, the method comprising: normalizing an image to obtain an image function of the image; calculating mixed continuous and discrete orthogonal moments of the image; wherein the mixed continuous and discrete orthogonal moments are mixed orthogonal moments with multiple normalized polynomials as basis functions, and are used for feature extraction of the image; and using the mixed continuous and discrete orthogonal moments to extract features of the image.

Description

Technical Field

[0001] The present disclosure belongs to the technical field of image processing, and in particular relates to an image processing method, device and electronic device. Background Art

[0002] In the field of computer vision, accurate recognition of target images requires the design of features with significant representation and distinction capabilities. Compared with traditional two-dimensional images, three-dimensional images provide richer and more specific target information, especially the representation of the target's spatial information. With the development of computer and imaging technology, three-dimensional images appear more and more frequently in people's lives. Such as magnetic resonance imaging (MRI), computed tomography (CT) and three-dimensional scanning in medical image processing. Therefore, it is of great practical significance to study the feature design method for three-dimensional images.

[0003] At present, there are mainly the following methods for feature design and extraction of three-dimensional images. The first method is to use the structural features of the three-dimensional image itself, such as points, surfaces, and axes, to establish the features of the target. The establishment of this type of feature is closely related to the storage format of the target image. Common three-dimensional image formats, such as off, obj, etc., contain target point and surface information, which provides convenience for the establishment of this type of feature. The main disadvantages of this type of method are low dimension of features, weak representation ability, and feature design is too dependent on file format. The second type is based on spherical harmonics. This type of method regards the three-dimensional image as a sphere with different radii, and on this basis maps different spheres to spherical harmonics of different orders to establish image features. This type of method has a large amount of calculation, and the calculation of spherical harmonics also has a certain complexity. Summary of the invention

[0004] The purpose of the present disclosure is to provide an image processing method, device, storage medium and electronic device, which are used to solve the technical problems that the features extracted by the existing feature design and extraction method for three-dimensional images have low dimension, weak representation ability, the feature design is too dependent on the file format and has a large amount of calculation and complexity.

[0005] In order to achieve the above objective, the first aspect of the present disclosure provides an image processing method, comprising:

[0006] Performing normalization processing on the image to obtain an image function of the image;

[0007] Calculating mixed continuous and discrete orthogonal moments of the image; wherein the mixed continuous and discrete orthogonal moments are mixed orthogonal moments with multiple normalized polynomials as basis functions, and are used for feature extraction of the image;

[0008] The mixed continuous and discrete orthogonal moments are used to extract features from the image.

[0009] Optionally, the normalizing the original image to obtain the image function of the image includes:

[0010] Get the image coordinates (x, y, z) of the original image;

[0011] According to the transformation relationship, the transformed normalized image coordinates (x', y', z') are calculated;

[0012] determining an image function of the image according to the normalized image coordinates;

[0013] The image is represented as f(x, y, z), and its size is M×N×O, that is, there are M voxels in the horizontal X-axis direction, N voxels in the Y-axis direction, and O voxels in the Z-axis direction; the transformation relationship can be expressed by the following formula:

[0014]

[0015] y'=y,1≤y≤N,

[0016] z'=z,1≤z≤O,

[0017] M, N and O are all positive integers.

[0018] Optionally, the normalizing the original image to obtain the image function of the image further includes:

[0019] Get the image coordinates (x, y, z) of the original image;

[0020] According to the transformation relationship, the transformed normalized image coordinates (x', y', z') are calculated;

[0021] determining an image function of the image according to the normalized image coordinates;

[0022] The image is represented as f(x, y, z), and its size is M×N×O, that is, there are M voxels in the horizontal X-axis direction, N voxels in the Y-axis direction, and O voxels in the Z-axis direction; the transformation relationship can be expressed by the following formula:

[0023]

[0024]

[0025] z'=z,1≤z≤O,

[0026] M, N and O are all positive integers.

[0027] Optionally, the calculating a mixed continuous and discrete orthogonal moment of the image includes:

[0028] Determine a first orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the first orthogonal polynomial is a projection function of the image in the X direction;

[0029] Determine a second orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the second orthogonal polynomial is a projection function of the image in the Y direction;

[0030] Determine a third orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the third orthogonal polynomial is a projection function of the image in the Z direction;

[0031] The first polynomial, the second polynomial and the third polynomial are respectively one of two types: Hermite polynomial and discrete Krawtchouk polynomial.

[0032] Optional,

[0033] The normalized Hermite polynomial is determined by the following formula:

[0034]

[0035] in, is the pth order Hermite polynomial, σ is the scale parameter, and is also the standard deviation of the Gaussian function.

[0036] The regularized discrete Krawtchouk polynomial is determined by the following formula:

[0037]

[0038] Where N is the total number of discrete sampling points; α is the control parameter of the Krawtchouk polynomial, which controls the center position of the polynomial; w(v) is the weight function, which is defined as:

[0039]

[0040] Optionally, the calculating of the mixed continuous and discrete orthogonal moments of the image further includes:

[0041] The first polynomial in the X direction, the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the basis function of the mixed continuous orthogonal moment; wherein the Hermite polynomial is used as the first polynomial in the X direction, the discrete Krawtchouk polynomial is used as the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the mixed continuous and discrete HKKMs orthogonal moment;

[0042] The mixed continuous and discrete HKKMs orthogonal moments are determined by the following formula:

[0043]

[0044] Among them, the basis function is the product of the orthogonal moments in the three coordinate axis directions:

[0045]

[0046] f'(x',y',z') is the new image function corresponding to the original three-dimensional image coordinate transformation, and has:

[0047] f'(x',y',z')=f(x,y,z).

[0048] Among them, the mixed continuous and discrete HKKMs orthogonal matrix A pqr Determined using different orders (p, q, r).

[0049] Optionally, the calculating of the mixed continuous and discrete orthogonal moments of the image further includes:

[0050] The first polynomial in the X direction, the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the basis function of the mixed continuous orthogonal moment; wherein the Hermite polynomial is used as the first polynomial in the X direction and the second polynomial in the Y direction, and the discrete Krawtchouk polynomial is used as the third polynomial in the Z direction for point multiplication to obtain the mixed continuous and discrete HHKMs orthogonal moment;

[0051] The mixed continuous and discrete HHKMs orthogonal moments are determined by the following formula:

[0052]

[0053] Among them, the basis function is the product of the orthogonal moments in the three coordinate axis directions:

[0054]

[0055] Among them, the mixed continuous and discrete HHKMs orthogonal matrix B pqr Determined using different orders (p, q, r).

[0056] In a second aspect, a three-dimensional image feature extraction device based on mixed continuous orthogonal polynomials is provided, the device comprising:

[0057] An acquisition module, used to acquire a transformation relationship of image coordinates, calculate normalized image coordinates transformed according to the transformation relationship based on the image coordinates of the original three-dimensional image, and acquire an image matrix of the image based on the normalized image coordinates;

[0058] A calculation module is used to calculate the mixed continuous and discrete orthogonal moments of the image; wherein the mixed continuous and discrete orthogonal moments are mixed orthogonal moments with multiple normalized polynomials as basis functions, and are used for feature extraction of the image.

[0059] According to a third aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method according to the first aspect are implemented.

[0060] In a fourth aspect, an electronic device is provided, comprising: a memory having a computer program stored thereon;

[0061] A processor is used to execute the computer program in the memory to implement the steps of the method described in the first aspect above.

[0062] In the above technical solution, the traditional statistical moments for two-dimensional images are extended to three-dimensional space to obtain three-dimensional orthogonal moments. Different from the complex methods of extracting three-dimensional image features in the prior art, orthogonal moments have the advantages of convenient calculation and low information redundancy. Each type of orthogonal polynomial, that is, the basis function of the orthogonal moment has different characteristics. In view of the different characteristics of the two different types of orthogonal polynomials of Hermite and Krawtchouk when performing feature mapping, the basis functions and three-dimensional orthogonal moments that can simultaneously reflect the advantages of the two orthogonal polynomials are used to extract the features of the target image. The extracted features not only reflect the characteristics of one type of polynomial, but also reflect the common characteristics and features of the two polynomials. Orthogonal polynomials make the calculation of three-dimensional orthogonal moments very convenient, because each type of orthogonal polynomial supports recursive formula calculation. Secondly, mapping the target image to the orthogonal basis further reduces the information redundancy between orthogonal moments of different orders. Therefore, the present disclosure has the advantages of convenient calculation and low information redundancy, further improving the user experience when extracting three-dimensional image features, and the two different types of orthogonal polynomials Hermite and Krawtchouk used in the present disclosure, while considering that continuous and discrete orthogonal moments have different advantages in analyzing and processing digital images, a new type of three-dimensional orthogonal moment is designed by using mixed continuous and discrete orthogonal polynomials with different properties as basis functions, and the orthogonal moment is used to construct the features of the target image. This feature has the advantages of convenient calculation, excellent scalability, and good numerical stability.

[0063] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 is a flowchart of an image processing method provided by an embodiment of the present disclosure;

[0065] Figure 2 is a flow chart of a method for calculating a mixed continuous and discrete orthogonal moment of the image provided by an embodiment of the present disclosure;

[0066] Figure 3 It is a block diagram of an image processing device provided by an embodiment of the present disclosure.

[0067] Figure 4 are two three-dimensional binary images, Figure 4 (a) Part of the skull structure constructed from slices of MRI brain images. Figure 4 (b) is a cheetah shape displayed in the Princeton Shape Benchmark (PSB) 3D shape library. Both images have a voxel size of 200×200×200.

[0068] Figure 5 It is the skull structure using (0,0,0) to (50,50,50) order orthogonal moments to reconstruct the MRI image and its error.

[0069] Figure 6 It is the cheetah shape using (0,0,0) to (20,20,20) order orthogonal moments to reconstruct the MRI image and its error.

[0070] Figure 7 are three different spiders and owls in the PSB dataset; Figure 7 (a), (b) and (c) are three different spiders. Figure 7 (d) An owl.

[0071] Figure 8 It is displayed in the feature space constructed by HKKMs and HHKMs Figure 7 Classification results of three spiders and an owl: (a) Classification results using HKKMs (σ=0.1895, α1=α2=0.5); (b) Classification results using HHMs (σ1=σ2=0.1895, α=0.5). DETAILED DESCRIPTION

[0072] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.

[0073] First, the application scenario of the present invention is explained. The present invention can be applied to the feature design and extraction of three-dimensional images such as magnetic resonance imaging (MRI), computed tomography (CT) and three-dimensional scanning in medical image processing. For example, in CT scanning, the CT image needs to be reconstructed and the feature extraction of the three-dimensional image needs to be performed. In the prior art, the structural features of the three-dimensional image itself, such as points, surfaces and axes, are often used to establish the features of the target, and the feature extraction is performed based on spherical harmonics. However, the feature dimension established based on the structural features of the three-dimensional image itself is low, the representation ability is weak, and the feature design is too dependent on the file format; the method based on spherical harmonics regards the three-dimensional image as a sphere with different radii, and on this basis, different spheres are mapped to spherical harmonics of different orders to establish image features. This type of method has a large amount of calculation, and the calculation of spherical harmonics also has a certain complexity.

[0074] In order to solve the above problems, the present invention provides a three-dimensional image feature design method based on mixed continuous and discrete orthogonal polynomials, which uses two continuous orthogonal polynomials with different properties as basis functions to design a new three-dimensional orthogonal moment based on the advantages of different orthogonal polynomials, and uses the orthogonal moment to construct the features of the target image. This feature has the advantages of convenient calculation, excellent scalability and good numerical stability.

[0075] The present disclosure is described below in conjunction with specific embodiments. Figure 1 A flowchart of an image processing method provided in an embodiment of the present disclosure is as follows: Figure 1 As shown, the method may include:

[0076] S101, normalize an image to obtain an image function of the image.

[0077] The original images may include magnetic resonance images in medical image processing, computer tomography scans, and three-dimensional images such as three-dimensional scans.

[0078] This step also includes:

[0079] Get the image coordinates (x, y, z) of the original image;

[0080] According to the transformation relationship, the transformed normalized image coordinates (x', y', z') are calculated;

[0081] determining an image function of the image according to the normalized image coordinates;

[0082] The image is represented as f(x, y, z), and its size is M×N×O, that is, there are M voxels in the horizontal x-axis direction, N voxels in the y-axis direction, and O voxels in the z-axis direction; the transformation relationship can be expressed by the following formula:

[0083]

[0084] y'=y,1≤y≤N,

[0085] z'=z, 1≤z≤O, M, N and O are all positive integers.

[0086] It should be noted that, in another method, it also includes:

[0087] The image is represented as f(x, y, z), and its size is M×N×O, that is, there are M voxels in the horizontal X-axis direction, N voxels in the Y-axis direction, and O voxels in the Z-axis direction; the transformation relationship can be expressed by the following formula:

[0088]

[0089]

[0090] z'=z,1≤z≤O,

[0091] M, N and O are all positive integers.

[0092] S102, calculating mixed continuous and discrete orthogonal moments of the image; wherein the mixed continuous and discrete orthogonal moments are mixed orthogonal moments with multiple normalized polynomials as basis functions, and are used for feature extraction of the image.

[0093] In this step, the mixed continuous and discrete orthogonal moments of the image are calculated, including:

[0094] Determine a first orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the first orthogonal polynomial is a projection function of the image in the X direction;

[0095] Determine a second orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the second orthogonal polynomial is a projection function of the image in the Y direction;

[0096] Determine a third orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the third orthogonal polynomial is a projection function of the image in the Z direction;

[0097] The first polynomial, the second polynomial and the third polynomial are respectively one of two types: Hermite polynomial and discrete Krawtchouk polynomial.

[0098] The normalized Hermite polynomial is determined by the following formula:

[0099]

[0100] in, is the pth order Hermite polynomial, σ is the scale parameter, and is also the standard deviation of the Gaussian function.

[0101] The regularized discrete Krawtchouk polynomial is determined by the following formula:

[0102]

[0103] Where N is the total number of discrete sampling points; α is the control parameter of the Krawtchouk polynomial, which controls the center position of the polynomial; w(v) is the weight function, which is defined as:

[0104]

[0105] It should be noted that the calculation of the mixed continuous and discrete orthogonal moments of the image also includes:

[0106] The first polynomial in the X direction, the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the basis function of the mixed continuous orthogonal moment; wherein the Hermite polynomial is used as the first polynomial in the X direction, the discrete Krawtchouk polynomial is used as the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the mixed continuous and discrete HKKMs orthogonal moment;

[0107] The mixed continuous and discrete HKKMs orthogonal moments are determined by the following formula:

[0108]

[0109] Among them, the basis function is the product of the orthogonal moments in the three coordinate axis directions:

[0110]

[0111] f'(x',y',z') is the new image function corresponding to the original three-dimensional image coordinate transformation, and has:

[0112] f'(x',y',z')=f(x,y,z).

[0113] Among them, the mixed continuous and discrete HKKMs orthogonal matrix A pqr Determined using different orders (p, q, r).

[0114] And, the calculating of the mixed continuous and discrete orthogonal moments of the image further includes:

[0115] The first polynomial in the X direction, the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the basis function of the mixed continuous orthogonal moment; wherein the Hermite polynomial is used as the first polynomial in the X direction and the second polynomial in the Y direction, and the discrete Krawtchouk polynomial is used as the third polynomial in the Z direction for point multiplication to obtain the mixed continuous and discrete HHKMs orthogonal moment;

[0116] The mixed continuous and discrete HHKMs orthogonal moments are determined by the following formula:

[0117]

[0118] Among them, the basis function is the product of the orthogonal moments in the three coordinate axis directions:

[0119]

[0120] Among them, the mixed continuous and discrete HHKMs orthogonal matrix B pqr Determined using different orders (p, q, r).

[0121] S103: Use the mixed continuous and discrete orthogonal moments to extract features from the image.

[0122] For example, in the process of extracting features from a three-dimensional image using mixed continuous and discrete orthogonal moments, the effectiveness and advantages of the features of the three-dimensional image proposed by the present invention can be effectively verified.

[0123] Regarding the representation ability of image features,

[0124] When orthogonal moments are used as features, the ability of orthogonal moment features to identify images can be tested and evaluated by image reconstruction. When the same number and order of orthogonal moments are used to reconstruct the original image, the more similar the reconstructed image is to the original image, the better the image representation capability of the corresponding orthogonal moment feature. We test the representation capability of HKKMs and HHKMs by reconstructing a three-dimensional binary image. Figure 4 (a) Part of the skull structure constructed from slices of MRI brain images. Figure 4 (b) is a cheetah shape shown in the Princeton Shape Benchmark (PSB) 3D shape library. We fill these two shapes to create a binary stereo image, with all the background voxels of the object being 0 and all the voxels of the object being 1. The reconstruction error e is defined as the sum of the number of different voxels between the original 3D image and the binary reconstructed image, that is:

[0125]

[0126] T(·) is a binarization operation. In this experiment, the three-dimensional Gaussian-Hermite moment is used as a comparative method for reconstruction.

[0127] The experimental results are as follows Figure 5 When a higher order is used to reconstruct the image, such as the highest order is 50, HKKMs, HHKMs and three-dimensional Gaussian-Hermite moments can all reconstruct the overall shape of the original image, as shown in Figure 5 (a), (b) and (c). However, the reconstruction errors of HKKMs and HHKMs are smaller than those of the three-dimensional Gaussian-Hermite moments. When the order is small, such as when the order is 20, HKKMs is closer to the original image in the corresponding reconstructed image than the three-dimensional Gaussian-Hermite moments, as shown in Figure 6 As shown in (a) and (c), Figure 6 (a) shows the general head, two front limbs, and two hind limbs of the cheetah; Figure 6 (c) It is difficult to distinguish the head and forelimbs.

[0128] Left Figure 5 (a) is the HKKMs moment (σ=0.1177,α1=α2=0.5); middle image Figure 5 (b) is the HHKMs (σ1=σ2=0.1177,α=0.5) moment; (right figure) Figure 5 (c) Gaussian-Hermite moment (σ=0.1177) is used.

[0129] Left Figure 6 (a) is the HKKMs moment (σ=0.1895,α1=α2=0.5); middle image Figure 6 (b) Using HHKMs moment (σ1=σ2=0.1895, α=0.5); (right figure) Figure 6 (c) Gaussian-Hermite moment (σ=0.1895) is used.

[0130] Regarding the ability to distinguish image features. As a feature description method for three-dimensional images, HKKMs and HHMs should have good image distinguishing ability, that is, the features constructed using HKKMs and HHMs can represent three-dimensional images while also being able to distinguish other images. An image classification experiment was used to test the feature distinguishing ability of the two.

[0131] Figure 7 are three different spiders and owls in the PSB dataset. We use three HKKMs, (A 444 ,A 624 ,A777 )σ=0.1895,α1=α2=0.5 and three HHKMs(H 444 ,H 624 ,H 777 )σ1=σ2=0.1895,α=0.5Build a feature space and perform Figure 7 The three spiders shown in (a), (b) and (c) and Figure 7 (d) Classify the owl shown. Figure 7 The three spiders in (a), (b) and (c) have common characteristics of this species, and Figure 7 The owl in (d) has a significantly different appearance from the spider. A good feature should be able to distinguish between insects and birds while also being able to distinguish between three different spiders. Figure 8 Shown in the feature space constructed by HKKMs and HHKMs Figure 7 The classification results of three spiders and owls in the experiment are shown in Figure 2. The results show that no matter the three selected HKKMs or HHKMs, they can clearly distinguish the three types of spiders and owls. The points corresponding to the four shapes do not overlap in the feature space and are difficult to distinguish due to their close distribution. Therefore, this experiment confirms that the invented HKKMs and HHKMs are effective, convenient and potential 3D image feature description methods.

[0132] Figure 3 is a schematic diagram of the structure of an image processing device provided by an embodiment of the present disclosure, such as Figure 3 As shown, including:

[0133] The acquisition module 301 is used to acquire the transformation relationship of the image coordinates, calculate the normalized image coordinates transformed according to the transformation relationship based on the image coordinates of the original three-dimensional image, and acquire the image matrix of the image based on the normalized image coordinates;

[0134] The calculation module 302 is used to calculate the mixed continuous and discrete orthogonal moments of the image; wherein the mixed continuous and discrete orthogonal moments are mixed orthogonal moments with multiple normalized polynomials as basis functions, and are used for feature extraction of the image.

[0135] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0136] The present disclosure also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the method for reporting a fault provided by the present disclosure are implemented.

[0137] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings. However, the present disclosure is not limited to the specific details in the above embodiments. Within the technical concept of the present disclosure, the technical solution of the present disclosure can be subjected to a variety of simple modifications, and these simple modifications all belong to the protection scope of the present disclosure. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.

[0138] In addition, various embodiments of the present disclosure may be arbitrarily combined, and as long as they do not violate the concept of the present disclosure, they should also be regarded as the contents disclosed by the present disclosure.

Claims

1. An image processing method, characterized in that: include: Performing normalization processing on the image to obtain an image function of the image; Calculating mixed continuous and discrete orthogonal moments of the image; wherein the mixed continuous and discrete orthogonal moments are mixed orthogonal moments with multiple normalized polynomials as basis functions, and are used for feature extraction of the image; Extract features from the image using the mixed continuous and discrete orthogonal moments; The step of calculating mixed continuous and discrete orthogonal moments of the image comprises: Determine a first orthogonal polynomial of an orthogonal moment basis function of a three-dimensional image, wherein the first orthogonal polynomial is a projection function of the image in the X direction; Determine a second orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the second orthogonal polynomial is a projection function of the image in the Y direction; Determine a third orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the third orthogonal polynomial is a projection function of the image in the Z direction; The first polynomial, the second polynomial and the third polynomial are respectively one of two types: Hermite polynomial and discrete Krawtchouk polynomial; The normalized Hermite polynomial is determined by the following formula: in, is the pth order Hermite polynomial, σ is the scale parameter, which is also the standard deviation of the Gaussian function; The regularized discrete Krawtchouk polynomial is determined by the following formula: Where N is the total number of discrete sampling points; α is the control parameter of the Krawtchouk polynomial, which controls the center position of the polynomial; w(v) is the weight function, which is defined as: Calculating mixed continuous and discrete orthogonal moments of the image, further comprising: The first polynomial in the X direction, the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the basis function of the mixed continuous orthogonal moment; wherein the Hermite polynomial is used as the first polynomial in the X direction, the discrete Krawtchouk polynomial is used as the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the mixed continuous and discrete HKKMs orthogonal moment; The mixed continuous and discrete HKKMs orthogonal moments are determined by the following formula: Among them, the basis function is the product of the orthogonal moments in the three coordinate axis directions: f'(x',y',z') is the new image function corresponding to the original three-dimensional image coordinate transformation, and has: f'(x',y',z')=f(x,y,z). Among them, the mixed continuous and discrete HKKMs orthogonal matrix A pqr Determined using different orders (p, q, r); or, The first polynomial in the X direction, the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the basis function of the mixed continuous orthogonal moment; wherein the Hermite polynomial is used as the first polynomial in the X direction and the second polynomial in the Y direction, and the discrete Krawtchouk polynomial is used as the third polynomial in the Z direction for point multiplication to obtain the mixed continuous and discrete HHKMs orthogonal moment; The mixed continuous and discrete HHKMs orthogonal moments are determined by the following formula: Among them, the basis function is the product of the orthogonal moments in the three coordinate axis directions: Among them, the mixed continuous and discrete HHKMs orthogonal matrix B pqr Determined using different orders (p, q, r).

2. The method according to claim 1, characterized in that The normalizing the original image to obtain the image function of the image includes: Get the image coordinates (x, y, z) of the original image; According to the transformation relationship, the transformed normalized image coordinates (x', y', z') are calculated; determining an image function of the image according to the normalized image coordinates; The image is represented as f(x, y, z), and its size is M×N×O, that is, there are M voxels in the horizontal X-axis direction, N voxels in the Y-axis direction, and O voxels in the Z-axis direction; the transformation relationship can be expressed by the following formula: y'=y,1≤y≤N, z'=z,1≤z≤O, M, N and O are all positive integers.

3. The method according to claim 1, characterized in that The step of normalizing the original image to obtain the image function of the image also includes: Get the image coordinates (x, y, z) of the original image; According to the transformation relationship, the transformed normalized image coordinates (x', y', z') are calculated; determining an image function of the image according to the normalized image coordinates; The image is represented as f(x, y, z), and its size is M×N×O, that is, there are M voxels in the horizontal X-axis direction, N voxels in the Y-axis direction, and O voxels in the Z-axis direction; the transformation relationship can be expressed by the following formula: z'=z,1≤z≤O, M, N and O are all positive integers.

4. A three-dimensional image feature extraction device, characterized in that: The device comprises: An acquisition module, used to acquire a transformation relationship of image coordinates, calculate normalized image coordinates transformed according to the transformation relationship based on the image coordinates of the original three-dimensional image, and acquire an image matrix of the image based on the normalized image coordinates; A calculation module, used for calculating mixed continuous and discrete orthogonal moments of the image; wherein the mixed continuous and discrete orthogonal moments are mixed orthogonal moments with multiple normalized polynomials as basis functions, and are used for feature extraction of the image; The step of calculating mixed continuous and discrete orthogonal moments of the image comprises: Determine a first orthogonal polynomial of an orthogonal moment basis function of a three-dimensional image, wherein the first orthogonal polynomial is a projection function of the image in the X direction; Determine a second orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the second orthogonal polynomial is a projection function of the image in the Y direction; Determine a third orthogonal polynomial of the orthogonal moment basis function of the three-dimensional image, wherein the third orthogonal polynomial is a projection function of the image in the Z direction; The first polynomial, the second polynomial and the third polynomial are respectively one of two types: Hermite polynomial and discrete Krawtchouk polynomial; The normalized Hermite polynomial is determined by the following formula: in, is the pth order Hermite polynomial, σ is the scale parameter, which is also the standard deviation of the Gaussian function; The regularized discrete Krawtchouk polynomial is determined by the following formula: Where N is the total number of discrete sampling points; α is the control parameter of the Krawtchouk polynomial, which controls the center position of the polynomial; w(v) is the weight function, which is defined as: Calculating mixed continuous and discrete orthogonal moments of the image, further comprising: The first polynomial in the X direction, the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the basis function of the mixed continuous orthogonal moment; wherein the Hermite polynomial is used as the first polynomial in the X direction, the discrete Krawtchouk polynomial is used as the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the mixed continuous and discrete HKKMs orthogonal moment; The mixed continuous and discrete HKKMs orthogonal moments are determined by the following formula: Among them, the basis function is the product of the orthogonal moments in the three coordinate axis directions: f'(x',y',z') is the new image function corresponding to the original three-dimensional image coordinate transformation, and has: f'(x',y',z')=f(x,y,z). Among them, the mixed continuous and discrete HKKMs orthogonal matrix A pqr Determined using different orders (p, q, r); or, The first polynomial in the X direction, the second polynomial in the Y direction and the third polynomial in the Z direction are point-multiplied to obtain the basis function of the mixed continuous orthogonal moment; wherein the Hermite polynomial is used as the first polynomial in the X direction and the second polynomial in the Y direction, and the discrete Krawtchouk polynomial is used as the third polynomial in the Z direction for point multiplication to obtain the mixed continuous and discrete HHKMs orthogonal moment; The mixed continuous and discrete HHKMs orthogonal moments are determined by the following formula: Among them, the basis function is the product of the orthogonal moments in the three coordinate axis directions: Among them, the mixed continuous and discrete HHKMs orthogonal matrix B pqr Determined using different orders (p, q, r).

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is operated to execute the image processing method described in any one of claims 1 to 3.

6. An electronic device, characterized in that: include: The computer-readable storage medium as claimed in claim 5, having a computer program stored thereon; A processor, configured to execute the computer program in the computer-readable storage medium to implement the image processing method according to any one of claims 1 to 3.

Images