Eye fundus image feature point matching method based on improved feature descriptor and KNN search

Abstract

The present invention relates to computer vision. Disclosed is an eye fundus image feature point matching method based on an improved feature descriptor and KNN search. The method comprises: inputting an eye fundus image set requiring stitching, and performing image preprocessing on the image set to obtain an eye fundus representation image; constructing an improved image scale space pyramid, comparing pixel values of the eye fundus representation image by using a difference of Gaussian pyramid technique to preliminarily determine an extreme point, determining the position and direction of a key pixel point, and generating a feature descriptor by using an improved feature vector; and searching for the nearest and second-nearest neighbor matching points of a key point in each group of eye fundus images by means of a KNN algorithm, eliminating erroneous matching, and retaining a high-quality matching pair. The present invention achieves efficient processing and feature extraction of eye fundus images, optimizes the gray scale distribution of images, and improves the contrast and information expression capability of images; and in addition, the matching effect of key points in images is remarkably improved, and mismatching is eliminated by means of a distance ratio determination mechanism, thereby ensuring the reliability of a matching result.

Description

Fundus image feature point matching method based on improved feature descriptor and KNN search Technical Field

[0001] The present invention relates to the field of computer vision, and in particular to a fundus image feature point matching method based on improved feature descriptors and KNN search. Background Art

[0002] Feature recognition and stitching and fusion of fundus images are crucial components of many medical research fields and advanced artificial intelligence (AI) technology development. In the medical industry, fundus health screening is a key indicator of human health screening. Because most medical imaging cameras have limited shooting angles, multiple fundus images taken by a single instrument must be stitched and fused to form a panoramic fundus image before medical screening. This panoramic image is then used for case screening and health testing. In some advanced AI research and development fields, the accuracy of human organ recognition and detection, such as eye and lens recognition, is extremely high. This requires that recognition instruments be able to perform detailed and accurate characterization of the eye before recognition. Therefore, accurate feature extraction and image stitching are also required within the device. Feature extraction and matching of fundus images play a crucial role in these applications.

[0003] Due to the great differences and small size of the human eyeball, as well as the accuracy of the shooting instrument, the light during shooting, and other issues, the fundus images captured often have problems such as many noise points, uneven color tones, and low clarity, resulting in poor overall quality. Therefore, feature extraction and matching of fundus images are difficult, and the research on corresponding fast and high-precision response methods is of great value.

[0004] The feature extraction and matching of fundus images includes three steps: image preprocessing, image feature extraction, and image feature matching. Preprocessing often uses corresponding image enhancement operations to highlight image features based on different image characteristics. Feature extraction mainly includes classic local feature extraction methods such as SIFT, SURF, ORB, and KAZE. In the process, the FAST detector is added to detect feature points and the MLD feature description method is added to improve detection accuracy and facilitate feature matching. With the development of artificial intelligence, deep learning methods such as neural networks and genetic algorithms are also added. Although there are many mature cases of image feature extraction algorithms, the conventional image feature extraction and matching algorithms used in fundus images are not very effective in medical applications due to the problems of many noise points, large image shadow areas, and low clarity. At present, research in this area is not sufficient, and there are no improved feature extraction and matching methods based on the unique feature distribution and feature information of fundus images. Summary of the Invention

[0005] The present invention is proposed in view of the problems existing in the existing fundus image feature point matching method based on improved feature descriptors and KNN search. Therefore, the problem to be solved by the present invention is how to provide a fundus image feature point matching method based on improved feature descriptors and KNN search.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for matching feature points of fundus images based on an improved feature descriptor and KNN search, which includes inputting a set of fundus images to be spliced ​​and performing image preprocessing to obtain a fundus representation image;

[0008] An improved image scale space pyramid is constructed, and the Gaussian difference pyramid technique is used to compare the pixel values ​​of the fundus representation image to preliminarily determine the extreme points, locate the position and direction of the key pixels, and use the improved feature vector to generate feature descriptors;

[0009] The KNN algorithm is used to search for the nearest neighbor and next nearest neighbor matching points of the key points in each group of fundus images, eliminate erroneous matching, and retain excellent matching pairs.

[0010] As a preferred solution of the fundus image feature point matching method based on improved feature descriptor and KNN search of the present invention, the image preprocessing includes the following steps:

[0011] Acquire a fundus image dataset;

[0012] Input the fundus image and convert it into a grayscale image, calculate the grayscale histogram of the fundus image, calculate the grayscale value transformation table, and generate a uniformly distributed fundus image histogram based on the given fundus image grayscale histogram;

[0013] The grayscale values ​​and spatial domain values ​​of the pixels around the center of each pixel in the evenly distributed fundus image histogram are calculated according to the grayscale value transformation table. The pixel grayscale values ​​and spatial domain values ​​are weighted averaged and bilateral filtering is performed to enhance the image and obtain the fundus representation image.

[0014] As a preferred solution of the fundus image feature point matching method based on the improved feature descriptor and KNN search of the present invention, the generating of the uniformly distributed fundus image histogram includes the following steps:

[0015] Convert the fundus image into a grayscale image, and scan each pixel of the original grayscale image to obtain a grayscale histogram of the fundus image;

[0016] Calculate the cumulative distribution function of the grayscale histogram of the fundus image. The calculation formula is as follows:

[0017] Among them, n represents the number of pixels in an image, l represents the total number of gray levels, and n k Indicates gray level r k The number of pixels;

[0018] The mapping function is calculated based on the histogram gray value uniformity principle and the cumulative distribution function. The calculation formula is as follows:

[0019] The mapped value of each pixel is calculated according to the mapping function to generate a uniformly distributed fundus image histogram.

[0020] As a preferred solution of the fundus image feature point matching method based on improved feature descriptor and KNN search of the present invention, the specific steps of the bilateral filtering enhancement include:

[0021] Input the fundus image after histogram uniformization, and calculate the point distance weight and point value weight in the bilateral filter for each image pixel. The calculation formula is as follows:

[0022] Among them, W d Represents the point distance weight, W i represents the point value weight, ||pq|| is the Euclidean distance between point p and point q, |I p -I q | is the pixel difference between two points, represents the pixel variance, represents the distance variance;

[0023] The fusion weight of each point on the image is calculated based on the point distance weight and the point value weight. The calculation formula is as follows:

[0024] Among them, W p Represents the weight after fusion, and A represents the calculated image pixel set;

[0025] The filtered pixel intensity is calculated by taking the weighted average of the pixels surrounding each pixel. The calculation formula is as follows:

[0026] Among them, I' p Represents the pixel intensity after filtering.

[0027] As a preferred solution of the fundus image feature point matching method based on improved feature descriptor and KNN search of the present invention, the specific steps of generating feature descriptors include:

[0028] A Gaussian pyramid is constructed based on fundus representation images. The first three layers of each image pyramid are taken to calculate the difference image, generating a Gaussian difference pyramid. The improved scale space image is obtained by convolving the Gaussian difference pyramid with the original image. The pixel values ​​of the pixels in the area of ​​the pyramid are compared to determine the extreme points.

[0029] According to the scale invariance of extreme points, a directional reference is assigned to each local feature of the image. The value and direction of the key pixel are calculated by combining the scale-direction distribution characteristics of adjacent pixels. The calculation formula of the value and direction of the key pixel is as follows:

[0030] Among them, I (x,y) Indicates the value of the extreme point, Indicates the direction of the extreme point, (x+1, y+1) and (x-1, y-1) represent the information representation of the adjacent pixels of the extreme point in the scale space;

[0031] Select the pixel area around the key point and divide it into sub-areas. Make the largest inscribed hexagon in the sub-square with the central pixel of each sub-area as the center. Calculate the feature vectors from different directions as the feature vectors of the sub-area center point. Superimpose the feature vectors of the sub-area on the key feature point with distance as the weight to generate the key point descriptor.

[0032] As a preferred solution of the fundus image feature point matching method based on the improved feature descriptor and KNN search of the present invention, the specific steps of determining the extreme point include:

[0033] An image scale space displaying comprehensive information under multi-scale parameters is constructed by convolving a two-dimensional Gaussian function with the original fundus image. The image is downsampled using the Gaussian function to generate an image pyramid model.

[0034] Cut the first three layers of the Gaussian pyramid of each image, and subtract the three layers of the Gaussian pyramid in sequence to obtain a two-layer Gaussian difference pyramid;

[0035] The pixel values ​​of the difference pyramid image are compared with the pixel values ​​in the surrounding adjacent scale range to determine the extreme points.

[0036] As a preferred solution of the fundus image feature point matching method based on improved feature descriptor and KNN search of the present invention, the error elimination matching includes the following steps:

[0037] Search the nearest neighbor feature points and the next nearest neighbor feature points of the input key point descriptor according to the KNN algorithm;

[0038] Calculate the ratio of the nearest neighbor distance to the next nearest neighbor distance of the feature point, and calculate the proportional threshold based on different images. The relevant calculation formula is as follows:

[0039] Among them, R represents the proportional threshold, N represents the total number of feature points, and w k represents the weight of the kth feature point, d(x k ,y k ) represents the Euclidean distance between the kth feature point and its nearest neighbor, α represents the normalization coefficient, and w j Represents the weight coefficient for normalizing the weights of all feature points;

[0040] When the ratio of the nearest neighbor distance to the next nearest neighbor distance is less than the proportional threshold, it means that the matching point pair meets the feature point requirements and the matching point pair is accepted;

[0041] When the ratio of the nearest neighbor distance to the next nearest neighbor distance is greater than the proportional threshold, it means that the matching point pair does not meet the feature point requirements and the matching point pair is eliminated.

[0042] In a second aspect, the present invention provides a fundus image feature point matching system based on improved feature descriptors and KNN search, comprising: the system includes an image preprocessing module, a feature point detection module, a feature descriptor generation module, and a feature matching module;

[0043] The image preprocessing module is used to obtain fundus images, convert them into grayscale images, perform histogram equalization and bilateral filtering enhancement, and obtain fundus representation images;

[0044] The feature point detection module is used to obtain improved scale space and Gaussian difference pyramid technology, compare pixel values ​​to determine key pixels, and calculate the position and direction information of the points;

[0045] The feature descriptor generation module is used to divide the sub-region around the key point, calculate the feature vector of each sub-region, and generate the feature descriptor of the key point by combining the distance weight superposition;

[0046] The feature matching module is used to find the nearest neighbor matching point and the next nearest neighbor matching point of each key point using the KNN nearest neighbor search algorithm, and calculate the distance ratio to eliminate false matches.

[0047] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the fundus image feature point matching method based on improved feature descriptors and KNN search as described in the first aspect of the present invention is implemented.

[0048] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the fundus image feature point matching method based on improved feature descriptors and KNN search as described in the first aspect of the present invention.

[0049] The beneficial effects of the present invention are that it realizes efficient processing and feature extraction of fundus images, provides a reliable basis for subsequent matching, and helps to improve the accuracy and robustness of matching. By optimizing image preprocessing, the quality and characterization ability of the image are improved, providing better input for subsequent steps, helping to enhance the characteristics of the fundus characterization image, and improving the reliability of matching. By generating a uniformly distributed fundus image histogram, the grayscale distribution of the image is optimized, which helps to improve the contrast and information expression ability of the image, and provides a more discriminative image representation for feature extraction. Through the bilateral filtering enhancement step, the calculation of the point distance weight and the point value weight achieves the bilateral filtering effect of smoothing the image while maintaining the image edge details, further improving the image quality. Through the descriptor generation step, the distance weight superposition method of the sub-region feature vector combines local and global information, making the descriptor more robust to scene changes. Overall, the multiple steps in the present invention cooperate with each other to significantly improve the image key point matching effect, eliminate false matching through the distance ratio judgment mechanism, and ensure the reliability of the matching result. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0051] Figure 1 is a scene diagram of the fundus image feature point matching method based on improved feature descriptors and KNN search.

[0052] FIG2 is a structural diagram of the fundus image feature point matching method based on the improved feature descriptor and KNN search. DETAILED DESCRIPTION

[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0054] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0055] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0056] Example 1

[0057] 1 and 2 , which are the first embodiment of the present invention, provide a method for matching feature points of fundus images based on an improved feature descriptor and KNN search.

[0058] Input the fundus image set to be stitched, perform image preprocessing on the image set, and obtain the fundus representation image.

[0059] Furthermore, image preprocessing includes the following steps:

[0060] Acquire a fundus image dataset;

[0061] Input the fundus image and convert it into a grayscale image, calculate the grayscale histogram of the fundus image, calculate the grayscale value transformation table, and generate a uniformly distributed fundus image histogram based on the given fundus image grayscale histogram;

[0062] The grayscale values ​​and spatial domain values ​​of the pixels around the center of each pixel in the evenly distributed fundus image histogram are calculated according to the grayscale value transformation table. The pixel grayscale values ​​and spatial domain values ​​are weighted averaged and bilateral filtering is performed to enhance the image and obtain the fundus representation image.

[0063] Furthermore, generating a uniformly distributed fundus image histogram includes the following steps:

[0064] Convert the fundus image into a grayscale image, and scan each pixel of the original grayscale image to obtain a grayscale histogram of the fundus image;

[0065] Calculate the cumulative distribution function of the grayscale histogram of the fundus image. The calculation formula is as follows:

[0066] Among them, n represents the number of pixels in an image, l represents the total number of gray levels, and n k Indicates gray level r k The number of pixels;

[0067] The mapping function is calculated based on the histogram gray value uniformity principle and the cumulative distribution function. The calculation formula is as follows:

[0068] The mapped value of each pixel is calculated according to the mapping function to generate a uniformly distributed fundus image histogram.

[0069] Furthermore, the specific steps of bilateral filtering enhancement include:

[0070] Input the fundus image after histogram uniformization, and calculate the point distance weight and point value weight in the bilateral filter for each image pixel. The calculation formula is as follows:

[0071] Among them, W d Represents the point distance weight, W i represents the point value weight, ||pq|| is the Euclidean distance between point p and point q, |I p -I q | is the pixel difference between two points, represents the pixel variance, represents the distance variance;

[0072] The fusion weight of each point on the image is calculated based on the point distance weight and the point value weight. The calculation formula is as follows:

[0073] Among them, W p Represents the weight after fusion, and A represents the calculated image pixel set;

[0074] The filtered pixel intensity is calculated by taking the weighted average of the pixels surrounding each pixel. The calculation formula is as follows:

[0075] Among them, I' p Represents the pixel intensity after filtering.

[0076] An improved image scale space pyramid is constructed, and the Gaussian difference pyramid technology is used to compare the pixel values ​​of the fundus representation image to preliminarily determine the extreme points, locate the position and direction of the key pixels, and use the improved feature vector to generate feature descriptors.

[0077] Furthermore, the specific steps of generating feature descriptors include:

[0078] A Gaussian pyramid is constructed based on fundus representation images. The first three layers of each image pyramid are taken to calculate the difference image, generating a Gaussian difference pyramid. The improved scale space image is obtained by convolving the Gaussian difference pyramid with the original image. The pixel values ​​of the pixels in the area of ​​the pyramid are compared to determine the extreme points.

[0079] According to the scale invariance of extreme points, a directional reference is assigned to each local feature of the image. The value and direction of the key pixel are calculated by combining the scale-direction distribution characteristics of adjacent pixels. The calculation formula of the value and direction of the key pixel is as follows:

[0080] Among them, I(x,y) Indicates the value of the extreme point, Indicates the direction of the extreme point, (x+1, y+1) and (x-1, y-1) represent the information representation of the adjacent pixels of the extreme point in the scale space;

[0081] Select the pixel area around the key point and divide it into sub-areas. Make the largest inscribed hexagon in the sub-square with the central pixel of each sub-area as the center. Calculate the feature vectors from different directions as the feature vectors of the sub-area center point. Superimpose the feature vectors of the sub-area on the key feature point with distance as the weight to generate the key point descriptor.

[0082] Specifically, the sub-region around the key pixel point is divided into 4 sub-regions, and each sub-region is defined to include 16 directional features. Combined with the key point descriptor generated by 64-dimensional vector, the feature information of the key pixel points of the fundus image is represented for feature matching.

[0083] Furthermore, the specific steps for determining extreme points include:

[0084] An image scale space displaying comprehensive information under multi-scale parameters is constructed by convolving a two-dimensional Gaussian function with the original fundus image. The image is downsampled using the Gaussian function to generate an image pyramid model.

[0085] Cut the first three layers of the Gaussian pyramid of each image, and subtract the three layers of the Gaussian pyramid in sequence to obtain a two-layer Gaussian difference pyramid;

[0086] The pixel values ​​of the difference pyramid image are compared with the pixel values ​​in the surrounding adjacent scale range to determine the extreme points.

[0087] The KNN algorithm is used to search for the nearest neighbor and next nearest neighbor matching points of the key points in each group of fundus images, eliminate erroneous matching, and retain excellent matching pairs.

[0088] Furthermore, eliminating error matching includes the following steps:

[0089] Search the nearest neighbor feature points and the next nearest neighbor feature points of the input key point descriptor according to the KNN algorithm;

[0090] Calculate the ratio of the nearest neighbor distance to the next nearest neighbor distance of the feature point, and calculate the proportional threshold based on different images. The relevant calculation formula is as follows:

[0091] Among them, R represents the proportional threshold, N represents the total number of feature points, and w k represents the weight of the kth feature point, d(x k ,y k ) represents the Euclidean distance between the kth feature point and its nearest neighbor, α represents the normalization coefficient, and wj Represents the weight coefficient for normalizing the weights of all feature points;

[0092] When the ratio of the nearest neighbor distance to the next nearest neighbor distance is less than the proportional threshold, it means that the matching point pair meets the feature point requirements and the matching point pair is accepted;

[0093] When the ratio of the nearest neighbor distance to the next nearest neighbor distance is greater than the proportional threshold, it means that the matching point pair does not meet the feature point requirements and the matching point pair is eliminated.

[0094] Specifically, the k value in the KNN algorithm is 2.

[0095] This embodiment also provides a computer device, which is applicable to the fundus image feature point matching method based on improved feature descriptors and KNN search, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the fundus image feature point matching method based on improved feature descriptors and KNN search proposed in the above embodiment.

[0096] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0097] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the fundus image feature point matching method based on improved feature descriptors and KNN search proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk.

[0098] In summary, the present invention realizes efficient processing and feature extraction of fundus images, provides a reliable basis for subsequent matching, and helps to improve the accuracy and robustness of matching. By optimizing image preprocessing, the quality and characterization ability of the image are improved, providing better input for subsequent steps, helping to enhance the characteristics of the fundus characterization image, and improving the reliability of matching. By generating a uniformly distributed fundus image histogram, the grayscale distribution of the image is optimized, which helps to improve the contrast and information expression ability of the image, and provides a more discriminative image representation for feature extraction. Through the bilateral filtering enhancement step, the calculation of the point distance weight and the point value weight achieves a bilateral filtering effect that smoothes the image while maintaining the edge details of the image, further improving the quality of the image. Through the descriptor generation step, the distance weight superposition method of the sub-region feature vector combines local and global information, making the descriptor more robust to scene changes. Overall, the multiple steps in the present invention cooperate with each other to significantly improve the image key point matching effect, eliminate false matching through the distance ratio judgment mechanism, and ensure the reliability of the matching result.

[0099] Example 2

[0100] 1 and 2 , which illustrate a second embodiment of the present invention, provide a fundus image feature point matching method based on an improved feature descriptor and KNN search. To verify the beneficial effects of the present invention, scientific demonstration is performed through economic benefit calculations and simulation experiments.

[0101] Experimental preparation

[0102] Dataset preparation: Two sets of fundus image datasets are selected, one for applying the method of the present invention and the other for applying the traditional SIFT algorithm.

[0103] Software and hardware environment: Prepare a computer with high-performance computing capabilities and install appropriate image processing software and algorithm libraries.

[0104] Implementation details

[0105] Image preprocessing: Grayscale conversion, histogram equalization and bilateral filtering are performed on the two sets of fundus image data sets.

[0106] Feature point detection and description: In the method of the present invention, an improved scale space pyramid is constructed, and Gaussian difference pyramid technology is used to determine the extreme points, and then an improved feature descriptor is generated. In the traditional SIFT method, a standard feature detection and description algorithm is used.

[0107] Feature point matching: The KNN algorithm is used for feature point matching. The method of the present invention adds a distance ratio judgment mechanism to reduce mismatching, while the traditional SIFT algorithm performs matching directly.

[0108] Make data judgment based on the comparison data table in Table 1.

[0109] Table 1 Comparison data table

[0110] Comparing the proposed method with the traditional SIFT algorithm, the table data shows that the proposed method improves the accuracy of feature point matching by an average of 8.7% and the processing speed by an average of 0.7 seconds. This significant improvement is mainly attributed to the following innovations:

[0111] Improved scale-space pyramid: more effectively captures image features and enhances the detection accuracy of feature points.

[0112] Optimized feature descriptors: More accurate descriptions of feature points, enhancing accuracy during the matching process.

[0113] The KNN algorithm combined with the distance ratio judgment mechanism effectively reduces mismatches and improves the overall matching accuracy.

[0114] The embodiments of the present invention demonstrate its advantages in matching feature points in fundus images, particularly in terms of matching accuracy and processing speed. These advantages are of great significance for fundus image analysis and improve the reliability and efficiency of the algorithm in practical applications.

[0115] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A fundus image feature point matching method based on improved feature descriptors and KNN search, characterized by: include, Input the fundus image set to be stitched and perform image preprocessing to obtain the fundus representation image; An improved image scale space pyramid is constructed, and the Gaussian difference pyramid technique is used to compare the pixel values ​​of the fundus representation image to preliminarily determine the extreme points, locate the position and direction of the key pixels, and use the improved feature vector to generate feature descriptors; The specific steps of generating the feature descriptor include: A Gaussian pyramid is constructed based on fundus representation images. The first three layers of each image pyramid are taken to calculate the difference image, generating a Gaussian difference pyramid. The improved scale space image is obtained by convolving the Gaussian difference pyramid with the original image. The pixel values ​​of the pixels in the area of ​​the pyramid are compared to determine the extreme points. According to the scale invariance of extreme points, a directional reference is assigned to each local feature of the image. The value and direction of the key pixel are calculated by combining the scale-direction distribution characteristics of adjacent pixels. The calculation formula of the value and direction of the key pixel is as follows: Among them, I (x,y) Indicates the value of the extreme point, Indicates the direction of the extreme point, (x+1, y+1) and (x-1, y-1) represent the information representation of the adjacent pixels of the extreme point in the scale space; Select the pixel area around the key point and divide it into sub-areas. Make the largest inscribed hexagon in the sub-square with the central pixel of each sub-area as the center. Calculate the feature vectors from different directions as the feature vector of the sub-area center point. Superimpose the feature vectors of the sub-area on the key feature point with distance as the weight to generate the key point descriptor. The KNN algorithm is used to search for the nearest neighbor and next nearest neighbor matching points of key points in each set of fundus images, eliminating erroneous matching and retaining excellent matching pairs; The error elimination matching comprises the following steps: Search the nearest neighbor feature points and the next nearest neighbor feature points of the input key point descriptor according to the KNN algorithm; Calculate the ratio of the nearest neighbor distance to the next nearest neighbor distance of the feature point, and calculate the proportional threshold based on different images. The relevant calculation formula is as follows: Among them, R represents the proportional threshold, N represents the total number of feature points, and w k represents the weight of the kth feature point, d(x k ,y k ) represents the Euclidean distance between the kth feature point and its nearest neighbor, α represents the normalization coefficient, and w j Represents the weight coefficient for normalizing the weights of all feature points; When the ratio of the nearest neighbor distance to the next nearest neighbor distance is less than the proportional threshold, it means that the matching point pair meets the feature point requirements and the matching point pair is accepted; When the ratio of the nearest neighbor distance to the next nearest neighbor distance is greater than the proportional threshold, it means that the matching point pair does not meet the feature point requirements and the matching point pair is eliminated.

2. The method for matching fundus image feature points based on improved feature descriptors and KNN search according to claim 1, wherein: The image preprocessing includes the following steps: Acquire a fundus image dataset; Input the fundus image and convert it into a grayscale image, calculate the grayscale histogram of the fundus image, calculate the grayscale value transformation table, and generate a uniformly distributed fundus image histogram based on the given fundus image grayscale histogram; The grayscale values ​​and spatial domain values ​​of the pixels around the center of each pixel in the evenly distributed fundus image histogram are calculated according to the grayscale value transformation table. The pixel grayscale values ​​and spatial domain values ​​are weighted averaged and bilateral filtering is performed to enhance the image and obtain the fundus representation image.

3. The method for matching fundus image feature points based on improved feature descriptors and KNN search according to claim 2, wherein: The method of generating a uniformly distributed fundus image histogram comprises the following steps: Convert the fundus image into a grayscale image, and scan each pixel of the original grayscale image to obtain a grayscale histogram of the fundus image; Calculate the cumulative distribution function of the grayscale histogram of the fundus image. The calculation formula is as follows: Among them, n represents the number of pixels in an image, l represents the total number of gray levels, and n k Indicates gray level r k The number of pixels; The mapping function is calculated based on the histogram gray value uniformity principle and the cumulative distribution function. The calculation formula is as follows: The mapped value of each pixel is calculated according to the mapping function to generate a uniformly distributed fundus image histogram.

4. The method for matching feature points of fundus images based on improved feature descriptors and KNN search according to claim 2, wherein: The specific steps of the bilateral filtering enhancement include: Input the fundus image after histogram uniformization, and calculate the point distance weight and point value weight in the bilateral filter for each image pixel. The calculation formula is as follows: Among them, W d Represents the point distance weight, W i represents the point value weight, ||pq|| is the Euclidean distance between point p and point q, |I p -I q | is the pixel difference between two points, represents the pixel variance, represents the distance variance; The fusion weight of each point on the image is calculated based on the point distance weight and the point value weight. The calculation formula is as follows: Among them, W p Represents the weight after fusion, and A represents the calculated image pixel set; The filtered pixel intensity is calculated by taking the weighted average of the pixels surrounding each pixel. The calculation formula is as follows: Among them, I' p Represents the pixel intensity after filtering.

5. The method for matching fundus image feature points based on improved feature descriptors and KNN search according to claim 1, wherein: The specific steps of determining the extreme value point include: An image scale space displaying comprehensive information under multi-scale parameters is constructed by convolving a two-dimensional Gaussian function with the original fundus image. The image is downsampled using the Gaussian function to generate an image pyramid model. Cut the first three layers of the Gaussian pyramid of each image, and subtract the three layers of the Gaussian pyramid in sequence to obtain a two-layer Gaussian difference pyramid; The pixel values ​​of the difference pyramid image are compared with the pixel values ​​in the surrounding adjacent scale range to determine the extreme points.

6. A fundus image feature point matching system based on improved feature descriptors and KNN search, based on the fundus image feature point matching method based on improved feature descriptors and KNN search according to any one of claims 1 to 5, characterized in that: The system includes an image preprocessing module, a feature point detection module, a feature descriptor generation module and a feature matching module; The image preprocessing module is used to obtain fundus images, convert them into grayscale images, perform histogram equalization and bilateral filtering enhancement, and obtain fundus representation images; The feature point detection module is used to obtain improved scale space and Gaussian difference pyramid technology, compare pixel values ​​to determine key pixels, and calculate the position and direction information of the points; The feature descriptor generation module is used to divide the sub-region around the key point, calculate the feature vector of each sub-region, and generate the feature descriptor of the key point by combining the distance weight superposition; The feature matching module is used to find the nearest neighbor matching point and the next nearest neighbor matching point of each key point using the KNN nearest neighbor search algorithm, and calculate the distance ratio to eliminate false matches.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the fundus image feature point matching method based on improved feature descriptors and KNN search are implemented as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the fundus image feature point matching method based on improved feature descriptors and KNN search are implemented as described in any one of claims 1 to 5.

Images