A metrology method and system for photoreceptor density

Abstract

The application discloses a kind of metrology method and system for visual cell density, it is related to visual cell metrology technical field, the method comprises: obtaining target area video based on first preset angle threshold;Frame by frame extraction obtains target area image set;Registration processing obtains target area registration image;Intelligent processing model is constructed and pretreated, and pretreated target area registration image is obtained;Generate pixel abstract graph;The segmentation processing result of the pixel abstract graph is obtained by using superpixel segmentation technology processing, which includes a plurality of segmentation regions;Extract segmentation region and collect to obtain segmentation region data set;The visual cell density of the target retina macular region is calculated.The technical problem that visual cell diameter and density cannot be accurately measured in real time when observing living non-invasive visual cells in the prior art is solved.The technical effect of improving the real-time and accuracy of visual cell diameter and density measurement is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of optic cell measurement, and particularly relates to a measurement method and system for optic cell density. BACKGROUND

[0002] With the rapid development of computer information technology, observation and analysis of human retinas by means of computer technology has become an important approach for research on fundus and retinas. In the prior art, the dissected retinal tissues are often observed and researched by means of microscopic technology. When the living body is non-invasively observed, the clear imaging of optic cells is often realized by combining adaptive optical technology with existing fundus imaging equipment. Exemplary devices include adaptive optical laser confocal scanning ophthalmoscopes, optical coherence tomography scanners, adaptive optical fundus cameras and the like. However, due to small imaging angle and great difficulty in shooting, the diameter and density of optic cells cannot be measured in real time and accurately, which leads to the difficulty in clinical practice. Therefore, it is of great significance to research non-invasive observation of optic cells in living bodies and obtain quantitative data such as the diameter and density of optic cells.

[0003] However, in the prior art, when the living body is non-invasively observed, there is the technical problem that the shooting of optic cells is difficult, the imaging angle is small, and thus the diameter and density of optic cells cannot be measured in real time and accurately. SUMMARY

[0004] The present application relates to the technical field of optic cell measurement, and particularly relates to a measurement method and system for optic cell density.

[0005] In view of the above problems, the present application provides a measurement method and system for optic cell density.

[0006] In a first aspect, the present application provides a method for measuring the density of photoreceptor cells, which is realized by a system for measuring the density of photoreceptor cells, wherein the method comprises: collecting a video of a target macular region of a retina based on a first preset angle threshold to obtain a target region video; extracting images from the target region video frame by frame to obtain a target region image set, wherein the target region image set comprises a plurality of target region images; performing registration processing on the plurality of target region images to obtain a target region registration image; constructing an intelligent processing model and using the intelligent processing model to pre-process the target region registration image to obtain a pre-processed target region registration image; processing the pre-processed target region registration image based on a simple linear iterative clustering principle to generate a pixel abstraction graph of the pre-processed target region registration image; processing the pixel abstraction graph using a superpixel segmentation technique to obtain a segmentation processing result of the pixel abstraction graph, wherein the segmentation processing result comprises a plurality of segmentation regions; extracting any one of the plurality of segmentation regions and performing data collection on the segmentation region to obtain a segmentation region data set; and calculating the density of photoreceptor cells of the target macular region of the retina based on the segmentation region data set.

[0007] In a second aspect, the present application further provides a system for measuring the density of retinal photoreceptor cells, which is used to perform the method for measuring the density of retinal photoreceptor cells according to the first aspect, wherein the system comprises: a video acquisition module, which is used to acquire a video of a target macular region of the retina based on a first preset angle threshold, to obtain a target region video; a video processing module, which comprises: an image extraction module, which is used to extract images from the target region video frame by frame, to obtain a target region image set, wherein the target region image set comprises a plurality of target region images; an image registration module, which is used to perform registration processing on the plurality of target region images, to obtain a target region registration image; an image enhancement module, which is used to construct an intelligent processing model, and use the intelligent processing model to pre-process the target region registration image, to obtain a pre-processed target region registration image; an image abstraction module, which is used to process the pre-processed target region registration image based on a simple linear iterative clustering principle, to generate a pixel abstraction image of the pre-processed target region registration image; an image segmentation module, which is used to use a superpixel segmentation technique to process the pixel abstraction image, to obtain a segmentation result of the pixel abstraction image, wherein the segmentation result comprises a plurality of segmentation regions; and an analysis and calculation module, which comprises: a data acquisition module, which is used to extract any one of the plurality of segmentation regions, and acquire data of the segmentation region, to obtain a segmentation region data set; and a data calculation module, which is used to calculate the density of retinal photoreceptor cells of the target macular region of the retina based on the segmentation region data set.

[0008] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0009] The video information of the target macular region of the retina is collected at the first preset angle threshold, and the collected video is extracted frame by frame and registered, so as to obtain a target region registration image, that is, an image of the retinal region to be observed and measured, thereby achieving the technical effect of providing an image basis for subsequent processing and analysis of the retina. Then, the target region registration image obtained by shooting and processing is preprocessed, such as restoration enhancement, noise reduction and contrast enhancement, to obtain a preprocessed target region registration image, thereby achieving the technical target of improving the clarity of the retinal image. Further, the preprocessed target region registration image is subjected to pixel drawing and superpixel segmentation processing, and the segmented regions are calculated and analyzed to obtain a segmented region data set, thereby achieving the target of obtaining a segmented image of the rod cell and providing a data basis for subsequent accurate measurement of the rod cell. Finally, based on the related data of the segmented region, the rod cell diameter and density of the target macular region of the retina are calculated, thereby achieving the technical effect of improving the real-time and accuracy of the measurement of the rod cell diameter and density.

[0010] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, the specific embodiments of the present application can be implemented according to the content of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only exemplary, and those skilled in the art can obtain other drawings according to the provided drawings without creating any creative labor.

[0012] Figure 1 A flowchart of a method for measuring the density of rod cells according to the present application;

[0013] Figure 2 A flowchart of obtaining a target region registration image in a method for measuring the density of rod cells according to the present application;

[0014] Figure 3 A flowchart of constructing an intelligent processing model in a method for measuring the density of rod cells according to the present application;

[0015] Figure 4 A flowchart of calculating the density of rod cells in a target macular region of the retina in a method for measuring the density of rod cells according to the present application;

[0016] Figure 5 A structural diagram of a system for measuring the density of rod cells according to the present application.

[0017] Reference signs:

[0018] Video acquisition module M100, video processing module M200, image extraction module M210, image registration module M220, image enhancement module M230, image abstraction module M240, image segmentation module M250, analysis calculation module M300, data acquisition module M310, data calculation module M320. DETAILED DESCRIPTION

[0019] The present application provides a kind of for the metrology method and system of visual cell density, the technical problem that the existing technology is solved when carrying out living noninvasive visual cell observation, there is big, imaging view angle small, and then cannot real-time accurately to the diameter and density of visual cell is metered.The technical effect of improving the real-time, accuracy of visual cell diameter and density measurement is achieved.

[0020] In the technical scheme of the present application, the acquisition, storage, use and processing of data comply with the relevant provisions of national laws and regulations.

[0021] The technical solutions in the present application will be described clearly and completely below with reference to the drawings.Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application. In addition, it should be noted that, for convenience of description, only the parts related to the present application are shown in the drawings, not all.

[0022] Embodiment one

[0023] Please refer to the drawings Figure 1 The present application provides a kind of for the metrology method of visual cell density, wherein the method is applied to a kind of for the metrology system of visual cell density, the method specifically includes the following steps:

[0024] Step S100: based on the first preset angle threshold, target retinal macular region is carried out video acquisition, and target region video is obtained;

[0025] Step S200: the target region video is carried out frame-by-frame image extraction, and target region image set is obtained, wherein the target region image set includes multiple target region images;

[0026] Specifically, the one kind for the counting method of visual cell density is applied to the one kind for the counting system of visual cell density, can be through clinical confocal laser scanning ophthalmoscope SLO noninvasive retinal visual cell shooting in vivo, then the visual cell image is restored, the noise reduction and the enhancement processing, and then the diameter and the density of the visual cell are calculated by using the super pixel technology processing, and the effect that the quality of clinical noninvasive visual cell observation is improved, the real-time nature, the accuracy of visual cell diameter and density measurement are improved.

[0027] Firstly, based on the first preset angle threshold, the video information of the retinal region to be measured for the diameter and the density of the visual cell is collected by using the confocal laser scanning ophthalmoscope SLO, so that the target region video is obtained. Wherein, the retinal region to be measured for the diameter and the density of the visual cell is the target retinal macular region. Then, the target region video collected is extracted frame by frame to obtain each frame image in the target region video, and the target region image set is formed, that is, the target region image set includes each frame image of the target region video, that is, the plurality of target region images.

[0028] Through the frame-by-frame extraction of the video information, the plurality of target retinal macular region images are obtained, the noninvasive shooting of the visual cell is realized, and the technical effect of providing an image basis for the subsequent processing and calculation of the diameter and the density of the visual cell is achieved.

[0029] Step S300: performing registration processing on the plurality of target region images to obtain a target region registration image;

[0030] Further, as shown in the accompanying Figure 2 The step S300 of the present application further comprises:

[0031] Step S310: extracting a first frame image in the plurality of target region images, and taking the first frame image as a registration reference image;

[0032] Step S320: obtaining a non-first frame image set after eliminating the first frame image in the plurality of target region images, wherein the non-first frame image set includes a plurality of non-first frame images;

[0033] Step S330: taking the plurality of non-first frame images as to-be-registered images;

[0034] Step S340: performing offset registration on the to-be-registered images based on the registration reference image to obtain the target region registration image.

[0035] Specifically, before obtaining the target region registration image based on the plurality of target region images extracted from the target region video frame by frame, the plurality of target region images are sequentially subjected to registration processing.

[0036] First, the first frame image is extracted from the multiple target region images, and this first frame image, i.e., the first frame image of the target region video, is used as the registration reference image to provide a benchmark for the registration of the second frame and subsequent frames. Then, based on the multiple target region images, a set of non-first frame images is obtained. This set of non-first frame images refers to the collection of other floating images after removing the first frame image from the multiple target region images, and includes multiple non-first frame images, i.e., multiple floating images. In other words, the set of non-first frame images includes the images of the second frame and all other frames in the target region video. These multiple non-first frame images are then used as images to be registered, and image offset registration is performed based on the registration reference image. The offset registration of each image in the non-first frame image set is divided into two steps: First, the deviation between each floating image (i.e., each non-first frame image) and the registration reference image is calculated based on phase information, and coarse localization is performed on each floating image (i.e., each non-first frame image). Second, based on the coarse localization registration, registration is performed using feature point matching principles and affine transformation. Finally, the registered image of the target region is obtained after registration.

[0037] By registering multiple frames of images to obtain a registered image of the target region, the corresponding points of each non-first frame image in the non-first frame image set are matched with the corresponding points of the registration reference image. This achieves the technical goal of ensuring consistency in the spatial and anatomical positions of the target retinal macular region, thereby improving the image clarity of the target retinal macular region and providing a reliable and realistic image basis for subsequent processing and analysis.

[0038] Step S400: Construct an intelligent processing model and use the intelligent processing model to preprocess the target region registration image to obtain a preprocessed target region registration image;

[0039] Further details are attached. Figure 3 As shown, step S400 of the present invention further includes:

[0040] Step S410: Construct a restoration processing layer based on the principle of blind deconvolution algorithm, wherein the restoration processing layer is used to perform a first preprocessing on the registered image of the target region;

[0041] Step S420: Construct a noise reduction processing layer based on the median filter principle, wherein the noise reduction processing layer is used to perform a second preprocessing on the registered image of the target region;

[0042] Step S430: Construct an enhancement processing layer based on the principle of adaptive histogram enhancement algorithm, wherein the enhancement processing layer is used to perform a third preprocessing on the registered image of the target region;

[0043] Step S440: constructing the intelligent processing model according to the restoration processing layer, the noise reduction processing layer and the enhancement processing layer.

[0044] Specifically, the intelligent processing model is constructed before the target region registration image obtained by registration is preprocessed by using the intelligent processing model.

[0045] Firstly, the restoration processing layer is constructed based on the principle of blind deconvolution algorithm, wherein the restoration processing layer is used for first preprocessing the target region registration image. Then, the noise reduction processing layer is constructed based on the principle of median filter, wherein the noise reduction processing layer is used for second preprocessing the target region registration image. Since the blind deconvolution algorithm based on PSF estimation is used in the restoration processing layer to restore and enhance the target region registration image, the real PSF and the original image are unknown, so that the noise information is also amplified synchronously when the image is restored. Therefore, the second preprocessing is performed on the target region registration image after the first preprocessing by using the noise reduction processing layer. For example, the median filter with a filter window of 5*5 is selected to filter the mis-restored noise, so that the rod cells in the image are clearer and the blood vessel edges are more prominent. Further, the enhancement processing layer is constructed based on the principle of adaptive histogram enhancement algorithm, wherein the enhancement processing layer is used for third preprocessing the target region registration image. The adaptive histogram enhancement algorithm can effectively improve the contrast of the image. Finally, the intelligent processing model is constructed according to the restoration processing layer, the noise reduction processing layer and the enhancement processing layer.

[0046] By constructing the intelligent processing model, the target region registration image obtained by shooting and processing is preprocessed by restoration and enhancement, noise reduction and contrast improvement, so as to obtain the preprocessed target region registration image, and the technical goal of improving the clarity of the retinal image is achieved.

[0047] Step S500: processing the preprocessed target region registration image based on the principle of simple linear iterative clustering to generate a pixel abstraction map of the preprocessed target region registration image.

[0048] Step S600: processing the pixel abstraction map by using a superpixel segmentation technology to obtain a segmentation processing result of the pixel abstraction map, wherein the segmentation processing result includes a plurality of segmentation regions.

[0049] Specifically, before performing superpixel segmentation processing on the pre-processed target region registration image, pixel abstraction processing is performed on the pre-processed target region registration image, so as to realize image dimension reduction of the pre-processed target region registration image. When performing pixel abstraction processing on the pre-processed target region registration image, processing is performed based on the simple linear iterative clustering principle. The simple linear iterative clustering is a simple and efficient method for constructing superpixels extended on the basis of the K-Means clustering algorithm. Further, the pixel abstraction image is segmented by using a superpixel segmentation technology to obtain a segmentation result of the pixel abstraction image. The segmentation result includes a plurality of segmentation regions. Through superpixel segmentation, a plurality of segmentation regions are obtained, which provides a basis for measuring the diameter of the optic cell.

[0050] Step S700: Extract any one of the plurality of segmentation regions, and perform data acquisition on the segmentation region to obtain a segmentation region data set;

[0051] Further, the step S700 of the present application further comprises:

[0052] Step S710: Obtain the total number of pixel points of the pixel abstraction image;

[0053] Step S720: Based on the segmentation result, obtain the number of segmentation regions;

[0054] Step S730: Based on the total number of pixel points and the number of segmentation regions, calculate the number of pixel points of the segmentation region, wherein the calculation formula of the number of pixel points of the segmentation region is as follows:

[0055] n=N / k;

[0056] Step S740: Wherein n refers to the number of pixel points of the segmentation region, N refers to the total number of pixel points, and k refers to the number of segmentation regions;

[0057] Step S750: According to the number of pixel points of the segmentation region, calculate the inner diameter of the segmentation region, wherein the calculation formula of the inner diameter of the segmentation region is as follows:

[0058] s=sqrt(n)=sqrt(N / k);

[0059] Step S760: Wherein s refers to the inner diameter of the segmentation region;

[0060] Step S770: According to the number of pixel points of the segmentation region and the inner diameter of the segmentation region, the segmentation region data set is composed.

[0061] Further, the present application further comprises the following steps:

[0062] Step S781: image collection is performed on the target retina based on a second preset angle threshold to obtain a target retina image, and the target retina image is taken as a positioning map;

[0063] Step S782: a macular center of the target retina image is obtained, and the macular center is taken as a positioning origin;

[0064] Step S783: a positioning rectangular coordinate system is constructed for the positioning map based on the positioning origin;

[0065] Step S784: a centroid position of the segmentation region is obtained according to the positioning rectangular coordinate system, and the coordinate is denoted as (x, y);

[0066] Step S785: an Euclidean distance between the centroid position and the positioning origin is calculated;

[0067] Step S786: the segmentation region inner diameter is corrected by using the Euclidean distance to obtain a segmentation region corrected inner diameter, and a calculation formula of the segmentation region corrected inner diameter is as follows:

[0068] s′=a*s,

[0069]

[0070] Step S787: wherein s′ denotes the segmentation region corrected inner diameter, a denotes a correction factor, denotes the Euclidean distance between the centroid position and the positioning origin.

[0071] Specifically, any one of the plurality of segmentation regions is processed and analyzed, for example, each of the plurality of segmentation regions is processed in turn, and before the segmentation region data set is obtained, the pixel point number and the inner diameter of the segmentation region are calculated and analyzed.

[0072] First, the total pixel point number of the pixel abstraction map is counted, that is, the total pixel number of the target retina macular region before the superpixel segmentation processing is performed. Then, the number of segmentation regions obtained after the pixel abstraction map of the target retina macular region is subjected to superpixel segmentation is counted, that is, the segmentation region number. Further, based on the total pixel point number and the segmentation region number, the segmentation region pixel point number is calculated. In addition, the pixel point numbers of each segmentation region are counted in turn, and the pixel point numbers of each segmentation region are subjected to mean value calculation, and the mean value calculation result is used to correct the segmentation region pixel point number. The specific calculation formula for calculating the segmentation region pixel point number based on the total pixel point number and the segmentation region number is as follows:

[0073] n=N / k;

[0074] wherein, n refers to the number of pixels in the segmentation region, N refers to the total number of pixels, and k refers to the number of segmentation regions.

[0075] Further, the inner diameter of the segmentation region is calculated according to the number of pixels in the segmentation region, and the calculation formula of the inner diameter of the segmentation region is as follows:

[0076] s = sqrt(n) = sqrt(N / k);

[0077] wherein, s refers to the inner diameter of the segmentation region.

[0078] Finally, the segmentation region dataset is obtained based on the number of pixels in the segmentation region and the inner diameter of the segmentation region.

[0079] Further, after the inner diameter of the segmentation region is calculated, the inner diameter of the segmentation region is processed and corrected again to obtain the corrected inner diameter of the segmentation region with more accurate data.

[0080] First, the target retina is image collected based on a second preset angle threshold to obtain a target retina image, and the target retina image is taken as a positioning map. The second preset angle threshold is much larger than the first preset angle threshold. For example, the first preset angle threshold is 8°, and the second preset angle threshold is 60°. Then, the macular center of the target retina image is found, and the macular center is taken as the positioning origin of the target retina image, i.e., the positioning map. For example, the fovea position of the macular center is taken as the positioning origin. Further, a positioning rectangular coordinate system is constructed for the positioning map based on the positioning origin, for example, the horizontal vector direction of the macula to the optic disc is taken as the positive direction of the X direction, and the positive direction of the X direction is rotated counterclockwise by 90° to be the positive direction of the Y direction. Finally, the accurate coordinates of the centroid position of the segmentation region in the positioning rectangular coordinate system, i.e., (x, y), are obtained, and the Euclidean distance between the centroid position and the positioning origin, i.e., the coordinates (x0, y0) = (0, 0), is calculated. Further, the Euclidean distance is used to correct the inner diameter of the segmentation region to obtain the corrected inner diameter of the segmentation region. The specific calculation formula is as follows:

[0081] s' = a * s,

[0082]

[0083] wherein, s' refers to the corrected inner diameter of the segmentation region, a refers to a correction factor, refers to the Euclidean distance between the centroid position and the positioning origin.

[0084] The pixel point number in the segmentation region and the inner diameter data of the segmentation region are calculated by processing the segmentation region, wherein the inner diameter of the segmentation region is regarded as the cell diameter after the superpixel segmentation result is regarded as a regular shape.

[0085] Step S800: Based on the segmentation region data set, the cell density of the target macular region of the retina is calculated.

[0086] Further, as shown in the accompanying Figure 4 The step S800 of the present application further comprises:

[0087] Step S810: The corrected inner diameter of the segmentation region is taken as the cell diameter of the target macular region of the retina.

[0088] Step S820: The region size of the target macular region of the retina is obtained, and the region area of the target macular region of the retina is calculated according to the region size.

[0089] Step S830: The cell density of the target macular region of the retina is calculated according to the cell diameter and the region area.

[0090] Specifically, the segmentation region after superpixel segmentation is irregularly shaped, and the inner diameter of the segmentation region is the diameter of the hexagon-like cell after it is regarded as a regular square. Therefore, the corrected inner diameter of the segmentation region is first taken as the cell diameter of the target macular region of the retina. Then the region size of the target macular region of the retina is measured, and the region area of the target macular region of the retina is calculated according to the region size. Finally, the cell density of the target macular region of the retina is calculated according to the cell diameter and the region area. By statistically analyzing the length, width and other related data of the target macular region of the retina and calculating the region area, the cell density is obtained by dividing the cell diameter by the region area, that is, the accurate quantitative data of the cell density is obtained, which achieves the technical effect of quickly and accurately measuring the corresponding cell diameter and density based on the real-time video of the target macular region of the retina.

[0091] In summary, the present application provides a kind of for cell density measurement method with following technical effects:

[0092] The video information of the target macular region of the retina is collected at a first preset angle threshold, and the collected video is extracted frame by frame and registered, so as to obtain a target region registration image, that is, an image of the retinal region to be observed and measured, thereby achieving the technical effect of providing an image basis for subsequent processing and analysis of the retina. Then, the target region registration image obtained by shooting and processing is preprocessed, such as restoration enhancement, noise reduction and contrast enhancement, to obtain a preprocessed target region registration image, thereby achieving the technical target of improving the clarity of the retinal image. Further, the preprocessed target region registration image is subjected to pixel drawing and superpixel segmentation processing, and the segmented regions are calculated and analyzed to obtain a segmented region data set, thereby achieving the goal of obtaining a segmented image of the rod cell and providing a data basis for subsequent accurate measurement of the rod cell. Finally, based on the related data of the segmented region, the diameter and density of the rod cell in the target macular region of the retina are calculated, thereby achieving the technical effect of improving the real-time and accuracy of the measurement of the diameter and density of the rod cell.

[0093] Embodiment Two

[0094] Based on the aforementioned embodiment, the present application also provides a system for measuring the density of rod cells, which is based on the same inventive concept. Please refer to the accompanying drawings. Figure 5 , the system comprises:

[0095] A video acquisition module M100 is configured to acquire a video of a target macular region of the retina based on a first preset angle threshold, thereby obtaining a target region video.

[0096] A video processing module M200 comprises:

[0097] An image extraction module M210 is configured to extract images from the target region video frame by frame, thereby obtaining a target region image set, wherein the target region image set comprises a plurality of target region images.

[0098] An image registration module M220 is configured to register the plurality of target region images, thereby obtaining a target region registration image.

[0099] An image enhancement module M230 is configured to construct an intelligent processing model and use the intelligent processing model to preprocess the target region registration image, thereby obtaining a preprocessed target region registration image.

[0100] An image abstraction module M240 is configured to process the preprocessed target region registration image based on the principle of simple linear iterative clustering, thereby generating a pixel abstraction image of the preprocessed target region registration image.

[0101] an image segmentation module M250, configured to process the pixel abstraction map by using a superpixel segmentation technique to obtain a segmentation result of the pixel abstraction map, wherein the segmentation result comprises a plurality of segmentation regions;

[0102] an analysis and calculation module M300, comprising:

[0103] a data acquisition module M310, configured to extract any one of the plurality of segmentation regions and perform data acquisition on the segmentation region to obtain a segmentation region data set;

[0104] a data calculation module M320, configured to calculate the photoreceptor cell density of the target retinal macular region based on the segmentation region data set.

[0105] Further, the image registration module M220 in the system is further configured to:

[0106] extract a first frame image in the plurality of target region images and take the first frame image as a registration reference image;

[0107] obtain a non-first frame image set after removing the first frame image in the plurality of target region images, wherein the non-first frame image set comprises a plurality of non-first frame images;

[0108] take the plurality of non-first frame images as to-be-registered images;

[0109] perform offset registration on the to-be-registered images based on the registration reference image to obtain the target region registration image.

[0110] Further, the image enhancement module M230 in the system is further configured to:

[0111] construct a restoration processing layer based on a blind deconvolution algorithm principle, wherein the restoration processing layer is configured to perform first preprocessing on the target region registration image;

[0112] construct a noise reduction processing layer based on a median filter principle, wherein the noise reduction processing layer is configured to perform second preprocessing on the target region registration image;

[0113] construct an enhancement processing layer based on an adaptive histogram enhancement algorithm principle, wherein the enhancement processing layer is configured to perform third preprocessing on the target region registration image;

[0114] construct the intelligent processing model according to the restoration processing layer, the noise reduction processing layer, and the enhancement processing layer.

[0115] Further, the data acquisition module M310 in the system is further used for:

[0116] obtaining a total pixel number of the pixel abstract graph;

[0117] obtaining a segmentation region number based on the segmentation processing result;

[0118] calculating a segmentation region pixel number based on the total pixel number and the segmentation region number, wherein a calculation formula of the segmentation region pixel number is as follows:

[0119] n = N / k;

[0120] wherein n refers to the segmentation region pixel number, N refers to the total pixel number, and k refers to the segmentation region number;

[0121] calculating a segmentation region inner diameter based on the segmentation region pixel number, wherein a calculation formula of the segmentation region inner diameter is as follows:

[0122] s = sqrt(n) = sqrt(N / k);

[0123] wherein s refers to the segmentation region inner diameter;

[0124] composing the segmentation region data set according to the segmentation region pixel number and the segmentation region inner diameter.

[0125] Further, the data acquisition module M310 in the system is further used for:

[0126] performing image acquisition on a target retina based on a second preset angle threshold to obtain a target retina image, and taking the target retina image as a positioning graph;

[0127] obtaining a macular center of the target retina image, and taking the macular center as a positioning origin;

[0128] constructing a positioning rectangular coordinate system for the positioning graph based on the positioning origin;

[0129] obtaining a centroid position of the segmentation region according to the positioning rectangular coordinate system, with a coordinate being (x, y);

[0130] calculating a Euclidean distance between the centroid position and the positioning origin;

[0131] correcting the segmentation region inner diameter by using the Euclidean distance to obtain a segmentation region corrected inner diameter, wherein a calculation formula of the segmentation region corrected inner diameter is as follows:

[0132] s' = a*s,

[0133]

[0134] wherein s' refers to the segmented area corrected inner diameter, a refers to a correction factor, refers to the Euclidean distance between the centroid position and the positioning origin.

[0135] Further, the data calculation module M320 in the system is further used for:

[0136] taking the segmented area corrected inner diameter as the rod cell diameter of the target retinal macular area;

[0137] obtaining the area size of the target retinal macular area, and calculating the area size of the target retinal macular area according to the area size;

[0138] calculating the rod cell density of the target retinal macular area according to the rod cell diameter and the area size.

[0139] Each embodiment in the specification is described in a progressive manner, and each embodiment focuses on the difference from other embodiments. The foregoing Figure 1 The measurement method for rod cell density and the specific example in embodiment one are also applicable to the measurement system for rod cell density in the present embodiment. Those skilled in the art can clearly understand the measurement system for rod cell density in the present embodiment through the foregoing detailed description of the measurement method for rod cell density. Therefore, in order to make the specification concise, the measurement system for rod cell density in the present embodiment will not be described in detail. As for the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part can be referred to the method part.

[0140] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application also intends to include these modifications and variations.

Claims

1. A metrology method for photoreceptor cell density, characterized by, The method comprises the following steps: Based on the first preset angle threshold, the target retinal macular region is video collected to obtain a target region video; Frame-by-frame image extraction is performed on the target region video to obtain a target region image set, wherein the target region image set comprises a plurality of target region images; Registration processing is performed on the plurality of target region images to obtain a target region registration image; An intelligent processing model is constructed, and the target region registration image is preprocessed using the intelligent processing model to obtain a preprocessed target region registration image; Based on the principle of simple linear iterative clustering, the preprocessed target region registration image is processed to generate a pixel abstraction graph of the preprocessed target region registration image; The pixel abstraction graph is processed using a superpixel segmentation technique to obtain a segmentation processing result of the pixel abstraction graph, wherein the segmentation processing result comprises a plurality of segmentation regions; The total number of pixel points of the pixel abstraction graph is obtained; based on the segmentation processing result, the number of segmentation regions is obtained; based on the total number of pixel points and the number of segmentation regions, the number of pixel points of the segmentation region is calculated, wherein the calculation formula of the number of pixel points of the segmentation region is as follows: n=N / k; wherein n is the number of pixel points of the segmentation region, N is the total number of pixel points, and k is the number of segmentation regions; according to the number of pixel points of the segmentation region, the inner diameter of the segmentation region is calculated, wherein the calculation formula of the inner diameter of the segmentation region is as follows: s=sqrt(n)=sqrt(N / k); wherein s is the inner diameter of the segmentation region; the segmentation region data set is composed according to the number of pixel points of the segmentation region and the inner diameter of the segmentation region; The corrected inner diameter of the segmentation region is taken as the cell body diameter of the target retinal macular region; the region size of the target retinal macular region is obtained, and the region area of the target retinal macular region is calculated according to the region size; according to the cell body diameter and the region area, the cell density of the target retinal macular region is calculated.

2. The method of claim 1, wherein, The registration processing of the plurality of target region images to obtain a target region registration image comprises: Extracting a first frame image in the plurality of target region images, and taking the first frame image as a registration reference image; Obtaining a non-first frame image set after removing the first frame image in the plurality of target region images, wherein the non-first frame image set comprises a plurality of non-first frame images; Taking the plurality of non-first frame images as to-be-registered images; Based on the registration reference image, the to-be-registered images are offset-registered to obtain the target region registration image.

3. The method of claim 1, wherein, The construction of the intelligent processing model comprises: Based on the principle of blind deconvolution algorithm, a restoration processing layer is constructed, wherein the restoration processing layer is used for first preprocessing of the target region registration image; Based on the principle of median filter, a noise reduction processing layer is constructed, wherein the noise reduction processing layer is used for second preprocessing of the target region registration image; Based on the principle of adaptive histogram enhancement algorithm, an enhancement processing layer is constructed, wherein the enhancement processing layer is used for third preprocessing of the target region registration image; According to the recovery processing layer, the noise reduction processing layer, and the enhancement processing layer, the intelligent processing model is constructed.

4. The method of claim 1, wherein, Further comprising: Based on the second preset angle threshold, the target retina is image collected to obtain a target retina image, and the target retina image is taken as a positioning map; Obtain the macular center of the target retina image, and take the macular center as a positioning origin; Based on the positioning origin, a positioning rectangular coordinate system is constructed for the positioning map; According to the positioning rectangular coordinate system, the centroid position of the segmentation region is obtained, and the coordinates are marked as (x, y); The Euclidean distance between the centroid position and the positioning origin is calculated; The inner diameter of the segmentation region is corrected by using the Euclidean distance to obtain a segmentation region corrected inner diameter, wherein the The calculation formula of the segmentation region corrected inner diameter is as follows: ； Wherein, s' refers to the segmentation region corrected inner diameter, a refers to a correction factor, and refers to the Euclidean distance between the centroid position and the positioning origin.

5. A metrology system for photoreceptor cell density, characterized by, The system is applied to the steps of the method of any one of claims 1-4, and the system comprises: A video acquisition module, configured to acquire a target macular region of a target retina based on a first preset angle threshold to obtain a target region video; A video processing module, comprising: An image extraction module, configured to extract images from the target region video frame by frame to obtain a target region image set, wherein the target region image set comprises a plurality of target region images; An image registration module, configured to perform registration processing on the plurality of target region images to obtain a target region registration image; An image enhancement module, configured to construct an intelligent processing model, and perform preprocessing on the target region registration image by using the intelligent processing model to obtain a preprocessed target region registration image; An image abstraction module, configured to process the preprocessed target region registration image based on a simple linear iterative clustering principle to generate a pixel abstraction map of the preprocessed target region registration image; An image segmentation module, configured to process the pixel abstraction map by using a superpixel segmentation technology to obtain a segmentation processing result of the pixel abstraction map, wherein the segmentation processing result comprises a plurality of segmentation regions; An analysis and calculation module, comprising: The data acquisition module is configured to obtain a total number of pixel points of the pixel abstract graph; based on the segmentation processing result, a number of segmentation regions is obtained; based on the total number of pixel points and the number of segmentation regions, a number of segmentation region pixel points is calculated, wherein a calculation formula of the number of segmentation region pixel points is as follows: n=N / k; wherein n represents the number of segmentation region pixel points, N represents the total number of pixel points, and k represents the number of segmentation regions; according to the number of segmentation region pixel points, a segmentation region inner diameter is calculated, wherein a calculation formula of the segmentation region inner diameter is as follows: s=sqrt(n)=sqrt(N / k); wherein s represents the segmentation region inner diameter; and according to the number of segmentation region pixel points and the segmentation region inner diameter, the segmentation region data set is formed. The data calculation module is configured to take the segmentation region corrected inner diameter as a rod cell diameter of the target retinal macular region; obtain a region size of the target retinal macular region, and calculate a region area of the target retinal macular region according to the region size; and according to the rod cell diameter and the region area, the rod cell density of the target retinal macular region is calculated.

Images