Image processing methods, apparatus, devices, storage media, and computer program products

Abstract

This application discloses an image processing method, apparatus, device, storage medium, and computer program product, relating to the field of machine learning. The method includes: acquiring a first registered image and a second registered image; performing a first registration on a second registered image of the second registered image using a first thumbnail image corresponding to the first registered image, thereby determining first registration parameters; determining registration coordinates in the second thumbnail image corresponding to the image coordinates based on the first registration parameters and the image coordinates corresponding to a first local region in the first thumbnail image; determining second registration parameters based on the image coordinates and the registration coordinates; and performing a second registration on the second registered image using the second registration parameters to obtain a registration result. Through this method, a global registration result for a large-size image is obtained based on the second registration parameters, improving the accuracy of image information analysis when performing image analysis on large-size images. This application can be applied to various scenarios such as cloud technology, artificial intelligence, healthcare, and intelligent transportation.

Description

Technical Field

[0001] This application relates to the field of machine learning, and in particular to an image processing method, apparatus, device, storage medium, and computer program product. Background Technology

[0002] With the continuous development of image processing technology, various image processing methods have emerged, such as image enhancement, image segmentation, image recognition, and image registration. Image registration is a technical challenge in the field of image processing research. Its purpose is to compare or fuse images of the same object acquired under different conditions, so that different images correspond one-to-one with points at the same location in space, thereby achieving the purpose of image information fusion.

[0003] In related technologies, it is usually not possible to directly register large images. Generally, the large image is cropped to obtain local images, and each local image is registered to obtain the registration result of the large image.

[0004] However, the above method cannot obtain global registration results for large-size images. When analyzing images based on the registration results of large-size images, the accuracy of the obtained image information is poor. Summary of the Invention

[0005] This application provides an image processing method, apparatus, device, storage medium, and computer program product that can obtain global registration results for large-size images, thereby more effectively improving the accuracy of image information analysis when performing image analysis on large-size images. The technical solution is as follows.

[0006] On the one hand, an image processing method is provided, the method comprising:

[0007] Acquire a first registration image and a second registration image, wherein the first registration image and the second registration image are images acquired from different acquisition angles for the target acquisition point;

[0008] Using the first thumbnail image corresponding to the first registered image, perform a first registration on the second thumbnail image of the second registered image to determine the first registration parameters;

[0009] Based on the first registration parameters and the image coordinates corresponding to the first local region in the first thumbnail image, the registration coordinates of the second local region in the second thumbnail image corresponding to the first local region are determined;

[0010] Based on the image coordinates and the registration coordinates, determine the second registration parameters;

[0011] The second registration image is registered using the second registration parameters to obtain the registration result of the second registration image relative to the first registration image.

[0012] On the other hand, an image processing apparatus is provided, the apparatus comprising:

[0013] The image acquisition module is used to acquire a first registration image and a second registration image, wherein the first registration image and the second registration image are images acquired from different acquisition angles for the target acquisition point;

[0014] The first registration module is used to perform a first registration on the second thumbnail of the second registration image using the first thumbnail image corresponding to the first registration image, and to determine the first registration parameters;

[0015] The coordinate determination module is used to determine the registration coordinates of the second local region in the second thumbnail that corresponds to the first local region, based on the image coordinates corresponding to the first local region in the first thumbnail and the first registration parameters.

[0016] The parameter determination module is used to determine the second registration parameter based on the image coordinates and the registration coordinates;

[0017] The second configuration module is used to perform a second registration on the second registration image using the second registration parameters, so as to obtain a registration result of the second registration image relative to the first registration image.

[0018] On the other hand, a computer device is provided, the computer device including a processor and a memory, the memory storing at least one instruction, at least one program, code set or instruction set, the at least one instruction, the at least one program, the code set or instruction set being loaded and executed by the processor to implement any of the image processing methods described in the embodiments of this application above.

[0019] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction, at least one program, code set, or instruction set is stored therein, wherein the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the image processing method as described in any of the embodiments of this application above.

[0020] On the other hand, a computer program product or computer program is provided, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform any of the image processing methods described in the above embodiments.

[0021] The beneficial effects of the technical solutions provided in this application include at least the following:

[0022] After determining the first registration parameters, the registration coordinates in the second thumbnail image and the second registration parameters are determined based on the first registration parameters and the image coordinates corresponding to the first registered image. The second registered image corresponding to the second thumbnail image is then registered using the second registration parameters to obtain the registration result. This method uses the first thumbnail image corresponding to the first registered image and the second thumbnail image corresponding to the second registered image to obtain the first registration parameters. When processing large images, the influence of size on the registration result is considered. Furthermore, the image coordinates of multiple first local images corresponding to the first thumbnail image are processed separately using the first registration parameters to determine the registration coordinates corresponding to each of the multiple first local images. The second registration parameters, determined using the registration coordinates and image coordinates, are used to register each second local region corresponding to the second registered image, and finally, the registration result of registering the second registered image onto the first registered image is determined, avoiding the situation where only local regions in large images are analyzed. By using thumbnail transformation and the coordinate relationships of different image regions, a more applicable second registration parameter is obtained, and the relationship between the second local regions is determined, thereby obtaining the global registration result of the large-size image, improving the applicability of the registration result, and also more effectively improving the accuracy of image information analysis when performing image analysis on large-size images. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of an implementation environment provided by an exemplary embodiment of this application;

[0025] Figure 2 This is a flowchart of an image processing method provided in an exemplary embodiment of this application;

[0026] Figure 3 This is a schematic diagram illustrating the use of a window sliding method to locally segment an image, provided in an exemplary embodiment of this application.

[0027] Figure 4 This is a flowchart of an image processing method provided in another exemplary embodiment of this application;

[0028] Figure 5 This is a schematic diagram illustrating the application of an image processing method to process an image, provided in an exemplary embodiment of this application.

[0029] Figure 6 This is an overall flowchart of an image processing method provided in an exemplary embodiment of this application;

[0030] Figure 7 This is a schematic diagram of a full-view data slice provided in an exemplary embodiment of this application;

[0031] Figure 8 This is a schematic diagram of a full-view data slice provided in another exemplary embodiment of this application;

[0032] Figure 9 This is a schematic diagram illustrating image registration using a first registration parameter, provided in an exemplary embodiment of this application.

[0033] Figure 10 This is a schematic diagram of the coordinate mapping process provided in an exemplary embodiment of this application;

[0034] Figure 11 This is a schematic diagram illustrating the application of an image processing method for image registration, provided in an exemplary embodiment of this application.

[0035] Figure 12 This is a structural block diagram of an image processing apparatus provided in an exemplary embodiment of this application;

[0036] Figure 13 This is a structural block diagram of an image processing apparatus provided in another exemplary embodiment of this application;

[0037] Figure 14 This is a structural block diagram of a server provided in an exemplary embodiment of this application. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.

[0039] First, a brief introduction to the terms used in the embodiments of this application will be given.

[0040] Artificial Intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0041] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0042] Machine Learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instruction-based learning.

[0043] In related technologies, it is generally not possible to directly register large images. Typically, the large image is cropped to obtain local images, and each local image is then registered to obtain the registration result of the large image. However, the above method cannot obtain the global registration result of a large image, and when analyzing the image based on the registration result of the large image, the accuracy of the obtained image information is poor.

[0044] This application provides an image processing method that can obtain global registration results for large-size images, thereby more effectively improving the accuracy of image information analysis when performing image analysis on large-size images. The image processing method trained according to this application can be applied in at least one of the following scenarios.

[0045] I. Application in the field of pathological image processing

[0046] In the medical field, pathological slides stained with different reagents are typically scanned to create a set of digital slide images. Pathological algorithms combine information from multiple staining agents and perform local matching and comparison of digital slide images with different staining agents to obtain more accurate judgment results. However, during the creation and scanning of digital slides, the images are relatively large, and analysis of these images usually only allows for the analysis of local regions. Illustratively, using the image processing method provided in this application, at least two digital slide images are acquired. After determining the first registration parameter using the corresponding thumbnail images of the at least two digital slide images, the registration coordinates of the digital slide image to be registered are determined based on the coordinate relationship of local regions in the digital slide images and the first registration parameter, thereby determining the second registration parameter. The digital slide image to be registered is then registered onto the target digital slide image to obtain the registration result. Here, the digital slide image to be registered and the target digital slide image are images from at least two digital slide images. The second registration parameter obtained by the above method has stronger applicability. When registering large images, it can obtain the global registration result of large images more quickly, and assist doctors in making a more comprehensive and accurate analysis and judgment of the patient's condition based on the pathological images.

[0047] II. Applications in the field of traffic image processing

[0048] In the transportation field, it is sometimes necessary to analyze road traffic conditions or vehicle conditions using road traffic videos to obtain the required image information. However, when analyzing real-time road conditions or vehicle conditions using road traffic videos, the corresponding images are usually large in size to include more information, so only a portion of the image can typically be analyzed. Illustratively, using the image processing method provided in this application, at least two road images including the target vehicle are acquired. After determining the first registration parameter using the thumbnail images corresponding to the at least two road images, the registration coordinates of the road image to be registered are determined based on the coordinate relationship of local regions in the road images and the first registration parameter, thereby determining the second registration parameter. The road image to be registered is then registered onto the target road image to obtain the registration result. The second registration parameter obtained by the above method has stronger applicability. When registering large-size road images, it can obtain the global registration result of the large-size image. It can not only obtain the information corresponding to the target vehicle (such as the license plate information of the target vehicle, the brand information of the target vehicle, etc.) more accurately, but also obtain the global information of the target vehicle in the road, improve the accuracy of the judgment of the target vehicle in the overall environment, and thus obtain more comprehensive information.

[0049] It is worth noting that the above application scenarios are merely illustrative examples, and the image processing method provided in this embodiment can also be applied to other scenarios, which are not limited in this application.

[0050] It is understood that in the specific implementation of this application, data such as user information, pathological image information, and real-time traffic information are involved. When the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0051] Secondly, the implementation environment involved in the embodiments of this application will be described, for illustrative purposes only. Please refer to [the relevant documentation]. Figure 1 The implementation environment involves a terminal 110 and a server 120, which are connected via a communication network 130.

[0052] In some embodiments, terminal 110 is used to send registration images (including a first registration image and a second registration image) to server 120. In some embodiments, terminal 110 is equipped with an application with image acquisition function.

[0053] The image processing method provided in this application embodiment can be implemented by the terminal 110 alone, by the server 120, or by the terminal 110 and the server 120 through data interaction. This application embodiment does not limit this. In this embodiment, the example is taken where the terminal 110 acquires an image through an application with image acquisition function, sends the acquired registration image to the server 120, and the server 120 registers the second registration image onto the first registration image to obtain the registration result.

[0054] Optionally, after receiving the registration image sent by the terminal 110, the server 120 performs thumbnail processing on the registration image to obtain a first thumbnail image corresponding to the first registration image and a second thumbnail image corresponding to the second registration image. The server then determines the first registration parameter using the first and second thumbnail images. Next, the server determines the registration coordinates corresponding to the image coordinates in the second thumbnail image by using the first registration parameter and the image coordinates corresponding to the first local region in the first thumbnail image. Then, the server determines the second registration parameter based on the image coordinates and the registration coordinates. Finally, the server registers the second registration image using the second registration parameter to obtain the registration result between the first and second registration images, thus realizing the registration process of registering the second registration image onto the first registration image.

[0055] It is worth noting that the aforementioned terminals include, but are not limited to, mobile terminals such as mobile phones, tablets, portable laptops, smart voice interaction devices, smart home appliances, and in-vehicle terminals, as well as desktop computers; the aforementioned servers can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers that provide basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0056] Cloud technology refers to a hosting technology that unifies hardware, applications, and network resources within a wide area network (WAN) or local area network (LAN) to achieve data computation, storage, processing, and sharing. Based on the cloud computing business model, cloud technology encompasses network technology, information technology, integration technology, management platform technology, and application technology. It can form resource pools, providing flexible and convenient on-demand access. Cloud computing technology will become a crucial support. Backend services of technical network systems require substantial computing and storage resources, such as video websites, image websites, and many portal websites. With the rapid development and application of the internet industry, every item may have its own identification mark in the future, requiring transmission to backend systems for logical processing. Data at different levels will be processed separately, and various industry data will require robust system support, which can only be achieved through cloud computing.

[0057] In some embodiments, the server described above can also be implemented as a node in a blockchain system. Blockchain is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and cryptographic algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include a blockchain underlying platform, a platform product service layer, and an application service layer.

[0058] Based on the above introduction to terminology and application scenarios, the image processing method provided in this application will be described, taking the application of this method to a server as an example. Figure 2 As shown, the method includes the following steps 210 to 250.

[0059] Step 210: Obtain the first registration image and the second registration image.

[0060] In illustrative terms, the first and second registered images can be pathological images in the medical field, traffic images in the transportation field, or architectural images, landscape images, etc., in daily life.

[0061] Optionally, depending on the differences in the types of the first and second registration images, the acquisition methods of the first and second registration images can include various methods. For example, when the first and second registration images are pathological images, they include both pathological images obtained from a pathological image database and images obtained by photographing cells, bones, etc.; or, when the first and second registration images are traffic images, they include both traffic images obtained from a traffic image database and images obtained by cropping scenes from road traffic videos; or, when the first and second registration images are architectural images, they include both architectural images obtained from an architectural image database and images obtained by photographing buildings, etc. The above are merely illustrative examples, and the embodiments of this application do not limit the scope of the invention.

[0062] Optionally, the first registration image and the second registration image are images acquired from different acquisition angles for the target acquisition point.

[0063] In illustrative terms, the target acquisition point serves as the reference for acquiring the first and second registered images. Depending on the type of target acquisition point, the first and second registered images can be categorized. For example, when the target acquisition point is a cell, the first and second registered images acquired using this target acquisition point as the reference can be medical images; or, when the target acquisition point is the environment, the first and second registered images acquired using this target acquisition point as the reference can be weather images, etc.

[0064] Schematic illustration: The first and second registration images acquired based on the target acquisition point include the target acquisition point. For example, if the target acquisition point is cell A, then the first and second registration images acquired using this target acquisition point as a reference will include cell A. Optionally, the first and second registration images are images acquired from different angles relative to the target acquisition point, and there is self-similarity between the first and second registration images. In some embodiments, the acquisition methods for the first and second registration images based on the target acquisition point include at least one of the following methods.

[0065] First, obtain the first and second registration images from the image library.

[0066] To illustrate, image set B stores multiple images that are related to each other. For example, the multiple images may be images taken from different angles of the same target acquisition point; or, image set B may be a medical image set, and the multiple images may contain images of the same or similar types of lesions, etc.

[0067] Second, perform fixed-point acquisition around the target acquisition point to obtain the first registration image and the second registration image.

[0068] Indicatively, when the image acquisition device is located at a designated point corresponding to the target acquisition point, it receives an image acquisition command and performs image acquisition according to the command. After acquiring images at multiple designated points, a first registered image and a second registered image are obtained.

[0069] Optionally, when the target acquisition point moves and is within the acquisition range of the image acquisition device, the image acquisition device continuously captures images during the movement of the target acquisition point to obtain a first registration image and a second registration image; or, the image acquisition device acquires a video stream during the movement of the target acquisition point to obtain a first registration image and a second registration image.

[0070] Third, video streams are acquired around the target acquisition point, and the first and second registration images are obtained from the acquired video streams.

[0071] Wherein, the first registration image and the second registration image are images randomly sampled from the video stream; or, the first registration image and the second registration image are images acquired at preset frame intervals in the video stream; or, the first registration image and the second registration image are specified frame images in the video stream, such as keyframe images in the video stream.

[0072] Optionally, the target acquisition point can be a predetermined acquisition point; or, the target acquisition point can be an object randomly determined from the reference image; or, the target acquisition point can be an object determined based on the area of ​​an object in the image. Illustratively, when the image acquisition device performs image acquisition around the target acquisition point, that is, when the target acquisition point is used as the center of the line of sight for image acquisition, the target acquisition point is located at the image center position of the first registered image and the second registered image.

[0073] It is worth noting that the methods for obtaining the first and second registration images described above are merely illustrative examples, and the embodiments of this application do not limit them.

[0074] In an optional embodiment, the first registration image and the second registration image are used for the image registration process. Optionally, a target registration image is determined from the first registration image or the second registration image, and another registration image is used as the image to be registered. For example, the second registration image is registered onto the first registration image, with the first registration image as the target registration image; or, the first registration image is registered onto the second registration image, with the second registration image as the target registration image.

[0075] Optionally, there can be multiple first and second registration images. Illustratively, the first registration image is used as the target registration image, and the second registration image is used as the image to be registered. After acquiring multiple images from the image set, the first registration image is determined from these images. Other images besides the first registration image are used as the second registration images, and the second registration images are registered onto the first registration image. For example: Images C1, C2, and C3 are acquired from image set C. C1 is used as the first registration image, and images C2 and C3 are used as the second registration images. With image C1 as the target registration image, images C2 and C3 are registered with image C1 respectively.

[0076] In an optional embodiment, the first and second registered images are the same size, for example, both the first and second registered images are images with an original size of 200,000 × 100,000; or, the first and second registered images are different sizes, for example, the original size of the first registered image is 200,000 × 100,000; and the original size of the second registered image is 200,000 × 500,000, etc. The above are merely illustrative examples, and the embodiments of this application do not limit the scope of the invention.

[0077] Step 220: Using the first thumbnail image corresponding to the first registered image, perform a first registration operation on the second thumbnail image of the second registered image to determine the first registration parameters.

[0078] Optionally, the first thumbnail image is an image obtained by scaling the first registered image at a first scaling ratio, and the second thumbnail image is an image obtained by scaling the second registered image at a second scaling ratio.

[0079] Schematic, the first thumbnail image is the image obtained by scaling down the first registered image at a first scaling ratio. Optionally, the first scaling ratio is a proportional scaling ratio. For example, if the first registered image is a two-dimensional image, when scaling down the first registered image at the first scaling ratio, the aspect ratio of the length and width of the first registered image remains unchanged. For example, if the original size of the first registered image is 200,000 × 100,000, the size of the first thumbnail image obtained after scaling down the first registered image is 2,000 × 1,000 (keeping the aspect ratio of the length and width of the first registered image at 2:1), wherein the first scaling ratio corresponding to the first registered image is 100.

[0080] Indicatively, the second thumbnail image is the image obtained by further shrinking the second registered image at a second scaling ratio.

[0081] Optionally, the relationship between the first scaling factor and the second scaling factor includes at least one of the following forms.

[0082] (1) The values ​​of the first abbreviated ratio and the second abbreviated ratio are the same.

[0083] To illustrate, the first and second registered images are scaled using the same scaling factor, meaning the first and second scaling factors are the same. For example, if the original size of the first and second registered images is 200,000 × 100,000, scaling them by a scaling factor of 100 (scaling factor of 10,000) results in a first and second thumbnail image with a scaled size of 2,000 × 1,000.

[0084] Alternatively, the original size of the first registered image is 200000×100000, and the original size of the second registered image is 200000×500000. The first and second registered images are scaled with a scaling factor of 100 (scaling ratio of 10000) to obtain a first thumbnail image with a scaling factor of 2000×1000 and a second thumbnail image with a scaling factor of 2000×5000.

[0085] (2) The values ​​of the first abbreviated ratio and the second abbreviated ratio are different.

[0086] To illustrate, the first and second registered images are scaled using different scaling ratios; that is, the first and second scaling ratios are different. For example, if the original size of the first and second registered images is 200,000 × 100,000, scaling the first registered image by a scaling factor of 100 (scaling ratio of 10,000) results in a first thumbnail image with a scaled size of 2,000 × 1,000. Scaling the second registered image by a scaling factor of 50 (scaling ratio of 2,500) results in a second thumbnail image with a scaled size of 4,000 × 2,000.

[0087] Alternatively, the original size of the first registered image is 200000×100000, and the original size of the second registered image is 200000×500000. The first registered image is scaled with a scaling factor of 100 (scaling ratio of 10000) to obtain a first thumbnail image with a scaling factor of 2000×1000. The second registered image is scaled with a scaling factor of 50 (scaling ratio of 2500) to obtain a second thumbnail image with a scaling factor of 4000×1000.

[0088] It is worth noting that the above are merely illustrative examples, and the embodiments of this application are not limited thereto.

[0089] In illustrative terms, after obtaining the first scaled image and the second scaled image, the first scaled image is used as the target image, and the second scaled image after being scaled by the second scaling ratio is registered, that is, the second scaled image is registered onto the first scaled image.

[0090] In an optional embodiment, a first feature point of the first thumbnail image and a second feature point of the second thumbnail image are matched to obtain an intermediate registration image.

[0091] Schematic illustration shows multiple image feature points on the first and second thumbnails, which are used to indicate information in the images. Optionally, various algorithms can be used to extract image feature points, such as the Speeded Up Robust Features (SURF) algorithm, the Scale-invariant feature transform (SIFT) algorithm, and the Oriented Fast and Rotated Brief (ORB) algorithm.

[0092] Optionally, based on the above algorithm, multiple first feature points of the first thumbnail image are determined from the first thumbnail image, and multiple second feature points of the second thumbnail image are determined from the second thumbnail image. Based on the correspondence between the first feature points of the first thumbnail image and the second feature points of the second thumbnail image, the first feature points and the second feature points are matched to obtain an intermediate registered image. The intermediate registered image is the image obtained by registering the second thumbnail image with the first thumbnail image. That is: taking the first thumbnail image as the target image, the second thumbnail image is registered on the first thumbnail image, and the second thumbnail image is adjusted according to the matching relationship between the second feature points and the first feature points to obtain an intermediate registered image.

[0093] Optionally, when matching the first feature point of the first thumbnail image and the second feature point of the second thumbnail image, the correspondence between the first feature point and the second feature point is determined according to the Random Sample Consensus (RANSAC) algorithm. For example, the first thumbnail image and the second thumbnail image are matched to find stable feature point pairs that are similar between the first feature point and the second feature point, and the homography matrix of the corresponding feature point pairs is calculated to realize the process of registering the second thumbnail image to the first thumbnail image, thereby obtaining an intermediate registered image.

[0094] In an optional embodiment, the first registration parameters are determined based on the intermediate registered image and the second thumbnail image.

[0095] The intermediate registration image is the result of registering the second registration image. It is illustrative. After obtaining the intermediate registration image, the first registration parameters are determined based on the relationship between the feature points of the intermediate registration image and the second feature points of the second thumbnail image. For example, the first registration parameters are determined based on the positional changes of the feature points in the intermediate registration image and the second feature points.

[0096] Optionally, intermediate coordinates corresponding to the intermediate registered image and thumbnail coordinates corresponding to the second thumbnail image are determined, and the image distance between the intermediate registered image and the second thumbnail image is determined using the intermediate coordinates and thumbnail coordinates. The intermediate coordinates are used to indicate the positional information of the intermediate registered image, and the thumbnail coordinates are used to indicate the positional information of the second thumbnail image. Illustratively, the first registration parameters are determined with the goal of reducing the image distance.

[0097] It is worth noting that the above are merely illustrative examples, and the embodiments of this application are not limited thereto.

[0098] Step 230: Based on the first registration parameters and the image coordinates corresponding to the first local region in the first thumbnail image, determine the registration coordinates of the second local region in the second thumbnail image that corresponds to the first local region.

[0099] In an optional embodiment, the first registered image is subjected to local segmentation processing to obtain at least one local region corresponding to the first registered image.

[0100] Optionally, the first registered image is a large-size image. When analyzing the first registered image, it is first divided into local segments. For example, a grid-based method is used to divide the first registered image into multiple small units, each unit corresponding to a local region. The sizes of the different small units can be the same or different.

[0101] Schematic illustration: When the different small units are of the same size, i.e., the first registered image is divided at equal intervals in the horizontal direction and at equal intervals in the vertical direction, the shape of the small unit grid is square when the horizontal and vertical division intervals are the same; the shape of the small unit grid is rectangular when the horizontal and vertical division intervals are different. Schematic illustration: When the different small units are of different sizes, i.e., the first registered image is divided at unequal intervals in the horizontal direction or at unequal intervals in the vertical direction.

[0102] In an optional embodiment, such as Figure 3 As shown, the first registered image is a whole-slide image (WSI) 310, which is a pathological image in the field of pathology. When analyzing this whole-slide image 310, a sliding window method is used to locally segment the pathological image. The sliding window is used to indicate the movement of a window 320, and the area corresponding to window 320 is the local region obtained after segmenting the first registered image.

[0103] Optionally, when performing local segmentation processing on the first registered image using the sliding window method, the first local region obtained according to the differences in the sliding window includes at least one of the following cases.

[0104] (1) The local area obtained varies depending on the window acquisition standard.

[0105] Optionally, when the window is formed, the window size is preset. During the window's sliding process, when the right edge of the window slides a certain distance to the right, the left edge of the window also slides a certain distance to the right. (Illustrative example, such as...) Figure 3As shown, window 320 is a pre-defined rectangle. When performing local segmentation processing on the first registered image using the sliding window method, the rectangle corresponding to window 320 starts from the upper left corner of the first registered image and traverses different regions of the first registered image. Each region traversed by the window is taken as a first local region, resulting in multiple first local regions corresponding to the first registered image.

[0106] Optionally, when traversing the first registered image through a window, it is determined whether the region corresponding to the window meets the region acquisition condition. If the region corresponding to the window meets the region acquisition condition, the region corresponding to the window is acquired to obtain the first local region. If the region corresponding to the window does not meet the region acquisition condition, the other regions of the first registered image are further judged, and the region that meets the region acquisition condition is determined as the first local region. For example, when traversing the first registered image through a window, the region acquisition condition is to acquire the D region part of the first registered image (e.g., the upper half of the first registered image). When the region corresponding to the window is located within the D region part, the region corresponding to the window is acquired to obtain the first local region. When the region corresponding to the window is not located within the D region part, the other regions of the first registered image are further judged, and the region that meets the region acquisition condition is determined as the first local region.

[0107] (2) Different local areas can be obtained depending on the window sliding method.

[0108] As an illustration, when traversing the first registered image through a window, the first local regions corresponding to the windows can be either adjacent to each other or partially overlap. (Illustrative example, such as...) Figure 3 As shown, the first local regions corresponding to the windows are adjacent to each other. When performing local segmentation processing on the first registered image through the windows, the obtained first local regions are adjacent to each other. For example, the left boundary of the current window in the first local region is the same as the right boundary of the first local region above the window. Alternatively, the first local regions corresponding to the windows partially overlap. For example, the left boundary of the current window in the first local region is within the first local region above the window.

[0109] (3) The order of the local areas obtained varies depending on the direction of window sliding.

[0110] Optionally, when traversing different regions of the first registered image through a window, the initial position of the window can include any position within the first registered image. Illustratively, when the initial position of the window is located at the upper left corner of the first registered image, traversing the first registered image through the window starts from the upper left corner and proceeds in a certain manner, such as from left to right or from top to bottom; or, when the initial position of the window is located at the center of the first registered image, traversing the first registered image through the window starts from the center and proceeds in a certain manner, such as traversing the first registered image outwards from the center point. Based on the position information of the window on the first registered image, multiple first local regions are obtained.

[0111] (4) The size of the local area obtained varies depending on the size of the window.

[0112] To illustrate, when determining the window size, the corresponding first local region obtained will also differ depending on the window size. For example, if the size of the first matching image is 100×100 and the window size is 5×5, when traversing the first registered image according to this window, 400 first local regions are obtained. Optionally, when the window size is the same as the size of the first registered image, when traversing the first registered image through this window, the size of the first local image obtained is the same as the size of the first registered image.

[0113] The above are merely illustrative examples, and the embodiments of this application are not intended to limit the scope of the application.

[0114] In illustrative terms, when performing local segmentation processing on the first registered image, threshold-based segmentation methods (such as fixed threshold segmentation method, iterative threshold image segmentation method), region-based segmentation methods (such as seed region growth method, region splitting and merging method), edge-based segmentation methods (such as edge detection method), etc. can also be used. The embodiments of this application do not limit this.

[0115] In an optional embodiment, the image coordinates corresponding to the first local region in the first thumbnail image are determined based on the first thumbnail scale and at least one local region.

[0116] Schematic illustration: After performing local segmentation processing on the first registered image, multiple local regions corresponding to the first registered image are obtained, where each local region is a region within the first registered image. Optionally, the coordinate values ​​corresponding to at least two coordinate points in each local region are determined; the coordinate values ​​corresponding to the at least two coordinate points are transformed using a first scaling factor to obtain transformation results corresponding to the at least two coordinate points; and based on the transformation results corresponding to the at least two coordinate points, the image coordinates corresponding to the first local region in the first thumbnail image are determined.

[0117] like Figure 3 As shown, taking any local region 330 as an example, after determining the local region 330, its coordinate information can be determined based on its position information in the first registered image. Optionally, the coordinate information of the local region 330 includes four coordinate points: the upper left coordinate point, the upper right coordinate point, the lower right coordinate point, and the lower left coordinate point. The position information of the local region 330 can be determined using these four coordinate points.

[0118] Optionally, based on the coordinate information of the local region 330 and the first scaling factor, the image coordinates corresponding to the first local region of the first thumbnail image are determined, wherein the first local region in the first thumbnail image corresponds to the local region 330 in the first registered image. For example, the local region 330 in the first registered image is determined by sliding a window, and the coordinate information of the local region 330 is C1, with corresponding coordinate points Pa1, Pb1, Pc1, and Pd1. Optionally, the coordinate information of the local region 330 is represented in the form "C1(Pa1, Pb1, Pc1, Pd1)". In the first thumbnail image, the first local region is the region in the first thumbnail image after scaling down the first registered image by the first scaling factor. Illustratively, the position information of the first local region is represented by image coordinates, that is, image coordinates are used to indicate the position information of the first local region in the first thumbnail image.

[0119] Based on the first scaling factor and the coordinate information of the local region 330—C1(Pa1, Pb1, Pc1, Pd1)—the image coordinates of the first local region in the first thumbnail image can be determined. The relationship between the image coordinates and the coordinate information of the local region is determined by the first scaling factor. Illustratively, the image coordinates of the first local region in the first thumbnail image are C2, and the corresponding coordinate points are Pa2, Pb2, Pc2, and Pd2. Optionally, the image coordinates of the first local region are represented in the form "C2(Pa2, Pb2, Pc2, Pd2)". For example, if the first scaling factor is 0.5, meaning the first registered image is reduced by half to become the first thumbnail image, and the first local region is the region corresponding to the first thumbnail image, then the relationship between the image coordinates and the coordinate information is Pa2, Pb2, Pc2, Pd2 = 1 / 2(Pa1, Pb1, Pc1, Pd1). The above is merely an illustrative example, and the embodiments of this application do not limit this.

[0120] In an optional embodiment, after determining the image coordinates of the first local image, the registration coordinates corresponding to the first local image in the second registered image are determined according to the first registration parameters. The registration coordinates are used to indicate the position information of the second local region in the second thumbnail image.

[0121] Optionally, a correspondence exists between the second local image and the first local image. Illustratively, the first registration parameter is used to indicate the registration relationship between the first thumbnail image and the second thumbnail image. As the first local image in the first thumbnail image, a correspondence can be established between it and the second local image in the second thumbnail image through the first registration parameter. After obtaining the image coordinates corresponding to the first local image, registration coordinates that correspond to the image coordinates in image position are determined based on the image coordinates and the first registration parameter.

[0122] Optionally, the image coordinates are C2(Pa2, Pb2, Pc2, Pd2). When determining the registration coordinates based on the first registration parameter and the image coordinates C2, the coordinate points corresponding to the image coordinates Pa2, Pb2, Pc2, and Pd2, as well as the first registration parameter, are determined sequentially. Schematic, the coordinate points of the registration coordinates are represented by Pa2_reg, Pb2_reg, Pc2_reg, and Pd2_reg, where reg indicates the result after registration using the first registration parameter.

[0123] Optionally, based on the determination of four coordinate points (Pa2_reg, Pb2_reg, Pc2_reg, Pd2_reg) in the registration coordinates, the second local region corresponding to the registration coordinates is determined. That is, the relationship between the second local region and the first local region is determined based on the relationship between the registration coordinates and the image coordinates. Optionally, due to registration reasons, the second local region and the first local region are likely to have a partial overlap but not a complete coincidence. Optionally, since the coordinate points of the registration coordinates of the second local region are the coordinate points obtained after registration, the second local region determined by the coordinate points of the registration coordinates is likely to be an irregular quadrilateral.

[0124] It is worth noting that the above are merely illustrative examples, and the embodiments of this application are not limited thereto.

[0125] Step 240: Determine the second registration parameters based on the image coordinates and registration coordinates.

[0126] Indicatively, the image coordinates are the coordinate information of the first local region in the first thumbnail image, and the registration coordinates are the coordinate information after the image coordinates are processed by the first registration parameter.

[0127] In an optional embodiment, the registration coordinates are standardized to help transform irregular shapes into standard shapes, thereby obtaining a standard region corresponding to the standard shape. Based on the relationship between the coordinate information of the standard region and the image coordinates, a second registration parameter is determined.

[0128] Step 250: Perform a second registration operation on the second registration image using the second registration parameters to obtain the registration result of the second registration image relative to the first registration image.

[0129] Optionally, the first registered image is used as the target image, and the second registered image is registered onto the first registered image. The second registration parameters are used to register the second registered image, enabling the display of key information from different images on the same image.

[0130] In an optional embodiment, a second local region in the second registered image is registered using a second registration parameter to obtain a local registration result corresponding to the second local region.

[0131] Schematic, the second registration parameter is a parameter obtained based on image coordinates and registration coordinates, where the image coordinates and registration coordinates are the positional information corresponding to the first thumbnail image and the second thumbnail image, respectively. Optionally, the second registration parameter is used to register the second local region to enhance the correlation between the second local region and the first local region, thereby obtaining the local registration result corresponding to the second local region.

[0132] Optionally, the second thumbnail image includes multiple second local regions. After performing second registration on each of the multiple second local regions, a local registration result corresponding to each second local region is obtained.

[0133] Optionally, the coordinate information of the second local region in the second registered image is determined. Illustratively, the positional information of the second local image is determined based on the position of the second local region in the second thumbnail image; that is, the coordinate information of the second local region in the second registered image is determined.

[0134] In an optional embodiment, based on coordinate information, the local registration results corresponding to at least two second local regions in the second registration image are stitched together to obtain the registration result between the first registration image and the second registration image.

[0135] Schematic illustration: After determining the coordinate information of the second local region in the second registered image, the local registration results corresponding to different second local regions are stitched together according to the coordinate information. For example, the coordinate information of the second local region M in the second registered image is m1, and the coordinate information of the second local region N in the second registered image is n1. Based on the coordinate information m1 corresponding to the second local region M, the coordinate information of the local registration result m corresponding to the second local region M in the second registered image is determined; based on the coordinate information corresponding to the second local region N, the coordinate information of the local registration result n corresponding to the second local region N in the second registered image is determined. Based on the position (coordinate information) of the local registration results corresponding to different second local regions in the second registered image, the different local registration results are stitched together to obtain the registration result between the first registered image and the second registered image. Optionally, if the first registered image is the target image, then the registration result between the first registered image and the second registered image is used to indicate the registration result of the second registered image being registered onto the first registered image.

[0136] The above are merely illustrative examples, and the embodiments of this application are not intended to limit the scope of the application.

[0137] In summary, after determining the first registration parameters, the registration coordinates in the second thumbnail image and the second registration parameters are determined based on the first registration parameters and the image coordinates corresponding to the first registered image. The second registered image corresponding to the second thumbnail image is then registered using the second registration parameters to obtain the registration result. This method considers the impact of size on the registration result when processing large images. Furthermore, the second registration parameters, determined using the registration coordinates and image coordinates, are used to register each second local region corresponding to the second registered image, ultimately determining the registration result of the second registered image onto the first registered image. This avoids analyzing only local regions within a large image. By using thumbnail transformation and the coordinate relationships of different image regions, more applicable second registration parameters are obtained, determining the relationships between the second local regions, thereby obtaining a global registration result for large images. This improves the applicability of the registration result and can more effectively enhance the accuracy of image information analysis when performing image analysis on large images.

[0138] In an optional embodiment, the second registration parameter is obtained based on image coordinates and registration coordinates. (Illustrative example, such as...) Figure 4 As shown above, Figure 2 Step 240 in the illustrated embodiment can also be implemented as steps 410 to 430.

[0139] Step 410: Based on the second scaling ratio and registration coordinates, determine the candidate coordinates corresponding to the second local region in the second thumbnail image.

[0140] Wherein, the second scaling ratio is the ratio by which the second registered image is converted into the second thumbnail image, and the registration coordinates are used to indicate the positional information of the region corresponding to the first local region. Optionally, based on the four coordinate points corresponding to the coordinates of the first local region image, the registration result of each coordinate point is determined sequentially, then the second thumbnail image has registration coordinates composed of the four registration coordinate points corresponding to the first local image, and the region composed of the four registration coordinate points corresponds to the first local region.

[0141] Schematic, the registration coordinates are determined based on image coordinates and a first registration parameter. The first registration parameter is determined based on a first thumbnail image and a second thumbnail image. The first and second thumbnail images are obtained by scaling down the first and second registered images. Optionally, when determining candidate coordinates in the second local region corresponding to the registration coordinates, the influence of the second scaling ratio on the registration coordinates is considered, thereby determining the position information of the registration coordinates in the second thumbnail image. That is, candidate coordinates of the second local region corresponding to the first local region in the first thumbnail image are determined in the second thumbnail image.

[0142] Optionally, when the second scaling factor is the same as the first scaling factor, the second scaling factor has a smaller impact on the registration coordinates when determining the candidate coordinates of the second local region; when the second scaling factor is different from the first scaling factor, the second scaling factor has a larger impact on the registration coordinates when determining the candidate coordinates of the second local region. Illustratively, the registration coordinates of the region corresponding to the first local region are C2(Pa2, Pb2, Pc2, Pd2), and the second scaling factor is denoted by Z2. Each registration coordinate point Pa2, Pb2, Pc2, and Pd2 in registration coordinates C2 is adjusted using the second scaling factor Z2 to obtain the candidate coordinate points corresponding to each registration coordinate point, i.e., four candidate coordinate points are obtained. These four candidate coordinate points determine the candidate coordinates, which are used to indicate the candidate position information of the second local region in the second thumbnail image corresponding to the first local region.

[0143] Step 420: Standardize the candidate coordinates to obtain the standardized coordinates corresponding to the second local region in the second thumbnail image.

[0144] Indicatively, after adjusting the registration coordinates to the second scaling ratio, candidate coordinates are obtained. The positional relationship of the four candidate coordinate points in the second thumbnail image is irregular. For example, the area formed by the four candidate coordinate points may present a tilted rectangular shape, or an irregular quadrilateral shape.

[0145] In an optional embodiment, the center coordinates of the region corresponding to the second local region in the second thumbnail image are determined based on the candidate coordinates.

[0146] In a schematic way, candidate coordinates are coordinates composed of four coordinate points. When determining the center coordinates of a region based on candidate coordinates, a candidate region is formed by the four coordinate points. The candidate region is a quadrilateral region, and the intersection of the diagonals in the candidate region is used as the center coordinates of the region; or, coordinate operations are performed on the four coordinate points, and the result of the coordinate operations is used as the center coordinates of the region.

[0147] Optionally, the candidate coordinates are coordinates composed of three coordinate points. When determining the center coordinates of the region based on the candidate coordinates, the candidate region is composed of three coordinate points and is a triangular region. The centroid of the candidate region is used as the center coordinates of the region. Alternatively, coordinate operations are performed on the three coordinate points, and the result of the coordinate operations is used as the center coordinates of the region.

[0148] The above are merely illustrative examples, and the embodiments of this application are not intended to limit the scope of the application.

[0149] In an optional embodiment, the circular region range corresponding to the second local region is determined by taking the center coordinates of the region as the point and the specified diagonal of the second local region as the diameter.

[0150] Optionally, the second local region is a quadrilateral region determined based on four coordinate points. Schematic, when determining the second local region based on candidate coordinates, the center coordinates of the region determined based on the candidate coordinates are used as the center point, and the diagonals of the quadrilateral region are connected to obtain the two diagonals corresponding to the quadrilateral region.

[0151] Indicatively, the longest diagonal of the two diagonals is used as the diameter; that is, the longest diagonal is used as the specified diagonal. Using the circle corresponding to the center coordinates of the region as the center and the specified diagonal as the radius, a circular region containing four coordinate points is obtained. This circular region includes two coordinate points, and the other two coordinate points are located within the circular region.

[0152] Optionally, using the circle corresponding to the center coordinates of the region as the center and the specified diagonal as the radius, after obtaining the circular region, the circular region is appropriately expanded to obtain the range of the circular region, so that all four coordinate points are located within the range of the circular region.

[0153] In an optional embodiment, the smallest rectangular region range containing the circular region range is determined as the standardized region range.

[0154] Optionally, after determining the extent of the circular region, the smallest rectangular region encompassing the circular region is determined, using the diameter of the circular region as its side. (Illustrative example follows.) Figure 5The image shown is a local area of ​​a pathological image. Candidate region 510 is used to indicate the region corresponding to candidate coordinates. Candidate region 510 is presented as a tilted quadrilateral image in the pathological image. The candidate coordinates include four candidate coordinate points, namely the first candidate coordinate point 511, the second candidate coordinate point 512, the third candidate coordinate point 513, and the fourth candidate coordinate point 514. Based on the four candidate coordinate points, the center coordinates 520 of the candidate region 510 are determined. With the center coordinates 520 as the center and the longest diagonal 530 of the candidate region 510 as the diameter, a circular region range 540 is obtained. With the diameter (diagonal 530) of the circular region range 540 as the side, the smallest rectangular region range 550 containing the circular region range is determined. This rectangular region range 550 is used as the standardized region range.

[0155] In an optional embodiment, the standardized coordinates corresponding to the second local region are determined based on the regional coordinates corresponding to the standardized region range.

[0156] For illustration, the standardized region is a quadrilateral, with the four vertices of the standardized region serving as the region coordinates; or, the standardized region is a triangle, with the three vertices of the standardized region serving as the region coordinates, etc.

[0157] Optionally, after determining the region coordinates, the coordinate values ​​corresponding to the candidate coordinates are mapped to the region coordinates to obtain the standardized coordinates corresponding to the second local region in the second thumbnail image.

[0158] Coordinate mapping establishes a mapping relationship between two images. For example, it maps any pixel in one image to another. (Illustrative example follows.) Figure 5 As shown, the candidate coordinates are the coordinates corresponding to the candidate region 510, including the first candidate coordinate point 511, the second candidate coordinate point 512, the third candidate coordinate point 513, and the fourth candidate coordinate point 514; the region coordinates are the coordinates corresponding to the standardized region range, that is, the coordinates corresponding to the rectangular region range 550.

[0159] Map the coordinate values ​​corresponding to the candidate coordinates to the region coordinates, that is, map the first candidate coordinate point 511, the second candidate coordinate point 512, the third candidate coordinate point 513, and the fourth candidate coordinate point 514 to the coordinate values ​​corresponding to the rectangular region. For example, the coordinates corresponding to the rectangular region range 550 include: the first region coordinate point 551, the second region coordinate point 552, the third region coordinate point 553, and the fourth region coordinate point 554.

[0160] Optionally, when mapping the coordinate values ​​corresponding to the candidate coordinates to the region coordinates, the coordinate point of the candidate coordinates is used as the minuend, and the first region coordinate point is used as the subtrahend to realize the process of mapping the candidate coordinates to the region coordinates. Based on the mapping result, the standardized coordinates corresponding to the second local region in the second thumbnail image are obtained. The standardized coordinates include four coordinate points, namely the first standard coordinate point, the second standard coordinate point, the third standard coordinate point, and the fourth standard coordinate point. That is: the difference between the first candidate coordinate point 511 and the first region coordinate point 551 is used as the first standard coordinate value (the coordinate value corresponding to the first coordinate point) in the standardized coordinates; the difference between the second candidate coordinate point 512 and the first region coordinate point 551 is used as the second standard coordinate value in the standardized coordinates; the difference between the third candidate coordinate point 513 and the first region coordinate point 551 is used as the third standard coordinate value in the standardized coordinates; and the difference between the fourth candidate coordinate point 514 and the first region coordinate point 551 is used as the fourth standard coordinate value in the standardized coordinates.

[0161] It is worth noting that the process of mapping the candidate coordinates to the region coordinates by subtracting the coordinates is only an illustrative example, and the embodiments of this application do not limit it.

[0162] Step 430: Determine the second registration parameters based on the standardized coordinates and the image coordinates corresponding to the first local region in the first thumbnail image.

[0163] Schematic, the standardized coordinates are the coordinate information of the second local region in the second thumbnail image that corresponds to the first local region. The second registration parameters are determined based on the correspondence between the second local region and the first local region.

[0164] In an optional embodiment, a matrix transformation is performed on the image coordinates corresponding to the first local image and the standardized coordinates corresponding to the second local image to obtain the second registration parameters.

[0165] Schematic illustration: multiple first local regions exist in the first registered image, and multiple second local regions exist in the second registered image. After determining the image coordinates corresponding to the first local regions and the standardized coordinates corresponding to the second local regions, a matrix transformation is performed on the image coordinates and the standardized coordinates.

[0166] Optionally, a similarity matrix transformation method can be used to perform matrix transformation on the image coordinates and the standardized coordinates.

[0167] The similarity matrix transformation method is a calculation method based on the similarity between the first local region and the second local region. Optionally, based on the above transformation, there is a shape similarity between the first local region and the second local region, for example: the first local region and the second local region are rectangular regions; or, the first local region and the second local region are regions obtained based on target feature points, that is: the first local region and the second local region have many similar feature points, etc.

[0168] Indicatively, based on the above transformation method, a similarity matrix between the first local region and the second local region is established using the coordinate information corresponding to the first local region and the second local region. The similarity relationship between the first local region and the second local region is solved using the similarity matrix method to obtain the second registration parameter.

[0169] It is worth noting that when performing matrix transformations on image coordinates and standardized coordinates, the Euler matrix transformation method can also be used. The above is only an illustrative example, and the embodiments of this application limit this.

[0170] In summary, after determining the first registration parameters, the registration coordinates in the second thumbnail image and the second registration parameters are determined based on the first registration parameters and the image coordinates corresponding to the first registered image. The second registered image corresponding to the second thumbnail image is then registered using the second registration parameters to obtain the registration result. This method considers the influence of size on the registration result when processing large images. Furthermore, by using the second registration parameters determined by the registration coordinates and image coordinates, each second local region corresponding to the second registered image is registered, and the final registration result of the second registered image onto the first registered image is determined, avoiding the situation where only local regions in a large image are analyzed. Through thumbnail transformation and the coordinate relationships of different image regions, the relationships between the second local regions are determined, thereby obtaining the global registration result of the large image.

[0171] In this embodiment, the process of determining the second registration parameter is described. Based on the second scaling ratio and the registration coordinates corresponding to the image coordinates in the first local region, candidate coordinates corresponding to the second local region in the second thumbnail image are determined. These candidate coordinates are then standardized to obtain standardized coordinates corresponding to the second local region in the second thumbnail image. Finally, using the standardized coordinates of the second local region and the image coordinates corresponding to the first local region, a matrix transformation method is employed to determine the second registration parameter. This second registration parameter is more applicable; when performing image analysis on large images, it can obtain global registration results for large images, improving the accuracy of image information analysis when analyzing large images.

[0172] In an optional embodiment, the above image processing method is applied to the medical field to register at least two full-view digital slice images. (Illustrative example, such as...) Figure 6 As shown above, Figure 2 The illustrated embodiment can also be implemented as follows: steps 610 to 630.

[0173] Step 610: Register the thumbnails.

[0174] Schematic, at least two thumbnails are thumbnail images corresponding to at least two full-view digital slice images respectively. Optionally, the process of registering at least two thumbnails can be implemented as follows: steps 611 to 612.

[0175] Step 611: Extract thumbnails.

[0176] To illustrate, first acquire at least two full-field digital slide images. For example, acquire at least two full-field digital slide images from a pathological image library; or, acquire at least two full-field digital slide images from different angles for the same pathological cell.

[0177] To illustrate, two full-field digital slide images are obtained from a pathological image database, such as... Figure 7 As shown, this is the first full-view digital slice image, WSI1; Figure 8 As shown, this is the second full-view digital slice image, WSI2. After thumbnailing WSI1 and WSI2, the thumbnail corresponding to WSI1 is obtained. Figure 1 and the abbreviation corresponding to WSI2 Figure 2 .

[0178] Optionally, WSI1 and WSI2 can be images of the same size or images of different sizes; or, when scaling down WSI1 and WSI2, they can be scaled down using the same scaling ratio or using different scaling ratios. The above are merely illustrative examples, and the embodiments of this application do not limit the scope of the application.

[0179] Step 612: Perform the first registration on the thumbnail.

[0180] Indicative, in obtaining the abbreviation corresponding to WSI1 Figure 1 and the abbreviation corresponding to WSI2 Figure 2 After that, the abbreviation Figure 1 and abbreviation Figure 2 Perform the first registration. For example: using the abbreviation... Figure 1 For the target image, the thumbnail will be... Figure 1 Feature points and abbreviations Figure 2 The feature points are matched to achieve the abbreviated version. Figure 2 Register to abbreviated Figure 1 The first registration process. In the abbreviated... Figure 2 After registration, the first registration parameter T is determined based on the registration result when performing the first registration on the thumbnail.

[0181] Step 620: Divide the full-view digital slice image.

[0182] In illustrative terms, full-field digital slides (FDS) are a digital slide technology in the medical field. They utilize a fully automated microscope scanning device to scan and stitch together traditional glass slides, resulting in a digital image that is fade-resistant and can be opened at any time. FDS images are often very large. In this embodiment, the FDS image is divided into multiple regions. In illustrative terms, the process of dividing the FDS image can be implemented as follows: steps 621 to 623.

[0183] Step 621: Divide the first local region.

[0184] Optionally, a sliding window method is used to divide WSI1 into local patches, resulting in multiple first local regions (patch1) corresponding to WSI1. Based on the position information of each first local region, the original coordinates C1 of each first local region are determined. Schematic, C1 consists of four coordinate points Pa1, Pb1, Pc1, and Pd1, representing the four boundary points: upper left, upper right, lower right, and lower left, respectively.

[0185] For example, C1 can be represented in coordinate form as C1(Pa1, Pb1, Pc1, Pd1).

[0186] Step 622: Calculate the thumbnail coordinates.

[0187] Optionally, when scaling down WSI1 and WSI2, WSI1 is scaled down using a scaling factor of Z1, and WSI2 is scaled down using a scaling factor of Z2. Illustratively, based on the original coordinates C1 of patch1 and the scaling factor Z1, the scaling factor of patch1 is calculated. Figure 1 The coordinates in the matrix are C2(Pa2, Pb2, Pc2, Pd2).

[0188] For example, C2 can be represented in coordinate form as C2(Pa2, Pb2, Pc2, Pd2).

[0189] Step 623: Calculate the thumbnail registration coordinates.

[0190] Optionally, based on patch 1 in the abbreviation Figure 1The coordinates C2 in the model and the first registration parameter T are used to calculate the abbreviated form corresponding to C2. Figure 2 The registration coordinates are C2_reg. Schematic, C2_reg consists of four boundary points: Pa2_reg, Pb2_reg, Pc2_reg, and Pd2_reg.

[0191] For example, C2_reg can be represented in coordinate form as C2_reg(Pa2_reg, Pb2_reg, Pc2_reg, Pd2_reg), where reg indicates the result after registration.

[0192] Indicative, such as Figure 9 As shown, this is a method for registering patch 1 using the first registration parameter T in a scaled-down format. Figure 1 The process of registering coordinates C2910 in the image yields a scaled-down image corresponding to C2910. Figure 2 The registration coordinates are C2_reg920.

[0193] Furthermore, due to registration mapping, the calculated C2_reg is mostly not a standard rectangular boundary, but rather... Figure 5 The irregular quadrilateral shown.

[0194] Step 630: Register the full-view digital slice image.

[0195] Indicative, in the case of abbreviation Figure 2 After obtaining the corresponding registration coordinates C2_reg, the registration process for the full-view digital slice image can be implemented as follows: steps 631 to 634.

[0196] Step 631: Calculate the original coordinates.

[0197] In an optional embodiment, the input WSI2 is subjected to a thumbnail extraction process at a scaling factor Z2 to obtain a thumbnail. Figure 2 For illustrative purposes, the abbreviated scale Z2 and the abbreviated scale Z1 can be the same.

[0198] Optionally, for abbreviation Figure 2 The corresponding registration coordinates C2_reg, and the abbreviation Figure 2 The corresponding scaling factor Z2 is used to calculate the coordinates C1_reg of C2_reg in the original WSI2. Illustratively, C1_reg consists of four boundary points: Pa1_reg, Pb1_reg, Pc1_reg, and Pd1_reg.

[0199] For example, C1_reg can be represented in coordinate form as C1_reg(Pa1_reg, Pb1_reg, Pc1_reg, Pd1_reg).

[0200] Step 632: Extract the second local region.

[0201] Optionally, C1_reg is not a standard rectangular boundary (e.g., Figure 5 As shown, to extract a standard rectangular image, a circular region is defined with the center of the area enclosed by C1_reg as the center point and the longest diagonal of that region as the diameter. A rectangular region is then defined with the diameter of this circular region (the longest diagonal of the area enclosed by C1_reg) as its side, thus expanding C1_reg outwards. This is illustrated in the diagram. Figure 5 As shown, the second local region (patch2) is obtained, which is a standard rectangular image. Optionally, patch2 consists of four boundary points, namely Pa1_rect, Pb1_rect, Pc1_rect, and Pd1_rect, where rect is used to indicate the result of expanding to obtain a rectangular image.

[0202] For example, patch2 can be represented in coordinate form as patch2(Pa1_rect, Pb1_rect, Pc1_rect, Pd1_rect).

[0203] Step 633: Calculate the second registration parameters.

[0204] In an optional embodiment, the coordinates are processed using a coordinate mapping method to determine the second registration parameter (local registration parameter).

[0205] Optionally, the coordinate values ​​of C1_reg(Pa1_reg, Pb1_reg, Pc1_reg, Pd1_reg) can be mapped to the local coordinates of patch2(Pa1_rect, Pb1_rect, Pc1_rect, Pd1_rect). For example, subtracting the coordinate values ​​of Pa1_rect from all the coordinate values ​​of C1_reg (the four coordinate points) yields C1_reg_local(Pa1_reg_local, Pb1_reg_local, Pc1_reg_local, Pd1_reg_local).

[0206] That is: Pa1_reg_local=Pa1_reg-Pa1_rect; Pb1_reg_local=Pa2_reg-Pa1_rect; Pc1_reg_local=Pa3_reg-Pa1_rect, Pd1_reg_local=Pa4_reg-Pa1_rect.

[0207] Indicative, such as Figure 10 As shown, Figure 1010 is used to indicate the process of mapping C1_reg to patch2, and Figure 1020 is used to indicate the mapping result obtained after mapping.

[0208] This illustrates how the coordinates of patch1 C1(Pa1, Pb1, Pc1, Pd1) are mapped to the local coordinates of patch1. For example, subtracting the coordinates of Pa1 from all the coordinates of C1 (the four coordinate points) yields C1_local(Pa1_local, Pb1_local, Pc1_local, Pd1_local).

[0209] That is: Pa1_local=Pa1-Pa1; Pb1_local=Pa2-Pa1; Pc1_local=Pa3-Pa1; Pd1_local==Pa4-Pa1.

[0210] Optionally, based on the coordinate mapping, the coordinates corresponding to patch2 and patch1 after mapping are obtained, namely C1_reg_local and C1_local, respectively. Using the coordinates C1_reg_local corresponding to patch2 and C1_local corresponding to patch1, the second registration parameter (local registration parameter) T_local is calculated by using the similarity matrix transformation or the Euler transformation matrix.

[0211] Step 634: Register the second local region.

[0212] Optionally, patch2 can be registered using the second registration parameter T_local to obtain a patch2 registration map corresponding to patch1.

[0213] For each partitioned patch1, after performing the corresponding transformation described above, the WSI2 registration result WSI2_T corresponding to the original WSI1 is obtained, as illustrated below. Figure 11 As shown, Figure 1110 is used to indicate WSI1, and Figure 1120 is used to indicate the registration result WSI2_T of WSI2.

[0214] The above are merely illustrative examples, and the embodiments of this application are not intended to limit the scope of the application.

[0215] In summary, after determining the first registration parameters, the registration coordinates in the second thumbnail image and the second registration parameters are determined based on the first registration parameters and the image coordinates corresponding to the first registered image. The second registered image corresponding to the second thumbnail image is then registered using the second registration parameters to obtain the registration result. This method considers the impact of size on the registration result when processing large images. Furthermore, the second registration parameters, determined using the registration coordinates and image coordinates, are used to register each second local region corresponding to the second registered image, ultimately determining the registration result of the second registered image onto the first registered image. This avoids analyzing only local regions within a large image. By using thumbnail transformation and the coordinate relationships of different image regions, more applicable second registration parameters are obtained, determining the relationships between the second local regions, thereby obtaining a global registration result for large images. This improves the applicability of the registration result and can more effectively enhance the accuracy of image information analysis when performing image analysis on large images.

[0216] Figure 12 This is a structural block diagram of an image processing apparatus provided in an exemplary embodiment of this application, such as... Figure 12 As shown, the device includes the following parts:

[0217] Image acquisition module 1210 is used to acquire a first registration image and a second registration image, wherein the first registration image and the second registration image are images acquired from different acquisition angles for the target acquisition point;

[0218] The first registration module 1220 is used to perform a first registration on a second thumbnail of the second registration image using a first thumbnail image corresponding to the first registration image, and to determine the first registration parameters;

[0219] The coordinate determination module 1230 is used to determine the registration coordinates of the second local region in the second thumbnail that corresponds to the first local region based on the first registration parameters and the image coordinates corresponding to the first local region in the first thumbnail;

[0220] The parameter determination module 1240 is used to determine the second registration parameter based on the image coordinates and the registration coordinates;

[0221] The second registration module 1250 is used to perform a second registration on the second registration image using the second registration parameters, so as to obtain a registration result of the second registration image relative to the first registration image.

[0222] In an optional embodiment, the first thumbnail image is an image obtained by scaling the first registered image at a first scaling ratio, and the second thumbnail image is an image obtained by scaling the second registered image at a second scaling ratio.

[0223] The parameter determination module 1240 is further configured to determine candidate coordinates corresponding to the second local region in the second thumbnail image based on the second thumbnail ratio and the registration coordinates; and to determine the second registration parameters based on the candidate coordinates and the image coordinates.

[0224] In an optional embodiment, the parameter determination module 1240 is further configured to perform normalization processing on the candidate coordinates to obtain normalized coordinates corresponding to the second local region in the second thumbnail image; and determine the second registration parameters based on the normalized coordinates and the image coordinates corresponding to the first local region in the first thumbnail image.

[0225] In an optional embodiment, the parameter determination module 1240 is further configured to: determine the region center coordinates corresponding to the second local region in the second thumbnail image based on the candidate coordinates; determine the circular region range corresponding to the second local region with the region center coordinates as the center point and the specified diagonal of the second local region as the diameter; determine the smallest rectangular region range containing the circular region range as the standardized region range; and determine the standardized coordinates corresponding to the second local region based on the region coordinates corresponding to the standardized region range.

[0226] In an optional embodiment, the parameter determination module 1240 is further configured to map the coordinate values ​​corresponding to the candidate coordinates to the region coordinates to obtain the standardized coordinates corresponding to the second local region in the second thumbnail image.

[0227] In an optional embodiment, the first thumbnail image is an image obtained by scaling the first registered image at a first scaling ratio, and the second thumbnail image is an image obtained by scaling the second registered image at a second scaling ratio.

[0228] like Figure 13 As shown, in an optional embodiment, the device further includes:

[0229] The local segmentation module 1260 is used to perform local segmentation processing on the first registered image to obtain at least one local region corresponding to the first registered image.

[0230] The image determination module 1270 is used to determine the image coordinates corresponding to the first local region in the first thumbnail image based on the first thumbnail ratio and the at least one local region.

[0231] In an optional embodiment, the location information of the local region is represented by at least two coordinate points;

[0232] The image determination module 1270 is further configured to determine the coordinate values ​​corresponding to the at least two coordinate points in the local region; adjust the coordinate values ​​corresponding to the at least two coordinate points according to the first scaling ratio to obtain the adjustment result corresponding to the at least two coordinate points; and determine the image coordinates corresponding to the first local region in the first thumbnail image based on the adjustment result corresponding to the at least two coordinate points.

[0233] In an optional embodiment, the second registration module 1250 is further configured to perform a second registration on the second local region in the second registration image using the second registration parameters to obtain a local registration result corresponding to the second local region; determine the coordinate information of the second local region in the second registration image; and based on the coordinate information, stitch together the local registration results corresponding to at least two second local regions in the second registration image to obtain a registration result of the second registration image relative to the first registration image.

[0234] In an optional embodiment, the parameter determination module 1240 is further configured to perform a matrix transformation on the image coordinates corresponding to the first local region and the standardized coordinates corresponding to the second local image to obtain the second registration parameters.

[0235] In an optional embodiment, the first registration module 1220 is further configured to match a first feature point of the first thumbnail image and a second feature point of the second thumbnail image to obtain an intermediate registration image, wherein the intermediate registration image is an image obtained by registering the second thumbnail image based on the first thumbnail image; and to determine the first registration parameters based on the intermediate registration image and the second thumbnail image.

[0236] In an optional embodiment, the first registration module 1220 is further configured to determine the intermediate coordinates corresponding to the intermediate registered image and the thumbnail coordinates corresponding to the second thumbnail image, wherein the intermediate coordinates are used to indicate the position information of the intermediate registered image and the thumbnail coordinates are used to indicate the position information of the second thumbnail image; determine the image distance between the intermediate registered image and the second thumbnail image using the intermediate coordinates and the thumbnail coordinates; and determine the first registration parameter with the goal of reducing the image distance.

[0237] In summary, after determining the first registration parameters, the registration coordinates in the second thumbnail image and the second registration parameters are determined based on the first registration parameters and the image coordinates corresponding to the first registered image. The second registration image corresponding to the second thumbnail image is then registered using the second registration parameters to obtain the registration result. Through this device, when processing large images, the influence of size on the registration result is considered. The second registration parameters, determined by the registration coordinates and image coordinates, are used to register each second local region corresponding to the second registered image, ultimately determining the registration result. This avoids the situation where only local regions in a large image are analyzed. By using thumbnail transformation and the coordinate relationships of different image regions, more applicable second registration parameters are obtained, determining the relationships between the second local regions, thereby obtaining a global registration result for large images. This improves the applicability of the registration result and can more effectively improve the accuracy of image information analysis when performing image analysis on large images.

[0238] It should be noted that the image processing apparatus provided in the above embodiments is only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the image processing apparatus and the image processing method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0239] Figure 14 This illustration shows a schematic diagram of a server provided in an exemplary embodiment of this application. The server 1400 includes a Central Processing Unit (CPU) 1401, a system memory 1404 including Random Access Memory (RAM) 1402 and Read Only Memory (ROM) 1403, and a system bus 1405 connecting the system memory 1404 and the CPU 1401. The server 1400 also includes a mass storage device 1406 for storing an operating system 1413, application programs 1414, and other program modules 1415.

[0240] Mass storage device 1406 is connected to central processing unit 1401 via a mass storage controller (not shown) connected to system bus 1405. Mass storage device 1406 and its associated computer-readable media provide non-volatile storage for server 1400. That is, mass storage device 1406 may include computer-readable media (not shown) such as hard disk or compact disc read-only memory (CD-ROM) drives.

[0241] Without loss of generality, computer-readable media can include computer storage media and communication media. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented using any method or technology for storing information such as computer-readable instructions, data structures, program modules, or other data. Computer storage media include RAM, ROM, erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other solid-state storage technologies, CD-ROM, digital versatile disc (DVD) or other optical storage, magnetic tape cassettes, magnetic tape, disk storage, or other magnetic storage devices. Of course, those skilled in the art will recognize that computer storage media are not limited to the above-mentioned types. The system memory 1404 and mass storage device 1406 described above can be collectively referred to as memory.

[0242] According to various embodiments of this application, server 1400 can also be connected to a remote computer on a network, such as the Internet. That is, server 1400 can be connected to network 1412 via network interface unit 1411 connected to system bus 1405, or it can also use network interface unit 1411 to connect to other types of networks or remote computer systems (not shown).

[0243] The aforementioned memory also includes one or more programs, which are stored in the memory and configured to be executed by the CPU.

[0244] Embodiments of this application also provide a computer device, which includes a processor and a memory. The memory stores at least one instruction, at least one program, code set, or instruction set. The at least one instruction, at least one program, code set, or instruction set is loaded and executed by the processor to implement the image processing method provided in the above-described method embodiments.

[0245] Embodiments of this application also provide a computer-readable storage medium storing at least one instruction, at least one program, code set, or instruction set, wherein the at least one instruction, at least one program, code set, or instruction set is loaded and executed by a processor to implement the image processing method provided in the above-described method embodiments.

[0246] Embodiments of this application also provide a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform any of the image processing methods described in the above embodiments.

[0247] Optionally, the computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), solid-state drives (SSDs), or optical discs, etc. The random access memory may include resistive random access memory (ReRAM) and dynamic random access memory (DRAM). The sequence numbers of the embodiments in this application are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0248] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0249] The above description is merely an optional embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. An image processing method, characterized in that, The method includes: Acquire a first registration image and a second registration image, wherein the first registration image and the second registration image are images acquired from different acquisition angles for the target acquisition point; Using the first thumbnail image corresponding to the first registered image, perform a first registration on the second thumbnail image of the second registered image to determine the first registration parameters. The second thumbnail image is an image obtained by scaling the second registered image by a second scaling ratio. Based on the first registration parameters and the image coordinates corresponding to the first local region in the first thumbnail image, the registration coordinates of the second local region in the second thumbnail image corresponding to the first local region are determined. Each registration coordinate point in the registration coordinates is adjusted according to the second scaling ratio to determine the candidate coordinates corresponding to the second local region in the second thumbnail image. The candidate coordinate points in the candidate coordinates correspond one-to-one with the registration coordinate points. The candidate coordinates are standardized to obtain the standardized coordinates corresponding to the second local region in the second thumbnail image. The standardization process is used to help the registration coordinates change from an irregular shape to a standard shape so as to obtain the second local region with a standard shape. Based on the standardized coordinates and the image coordinates corresponding to the first local region in the first thumbnail image, the second registration parameters are determined. The second registration image is registered using the second registration parameters to obtain the registration result of the second registration image relative to the first registration image.

2. The method according to claim 1, characterized in that, The first thumbnail image is an image obtained by scaling the first registered image at a first scaling ratio.

3. The method according to claim 1, characterized in that, The step of standardizing the candidate coordinates to obtain the standardized coordinates corresponding to the second local region in the second thumbnail image includes: Based on the candidate coordinates, determine the center coordinates of the region corresponding to the second local region in the second thumbnail image; Using the center coordinates of the region as the point and the designated diagonal of the second local region as the diameter, the range of the circular region corresponding to the second local region is determined; The smallest rectangular region encompassing the circular region is determined as the standardized region range; Based on the regional coordinates corresponding to the standardized region range, the standardized coordinates corresponding to the second local region are determined.

4. The method according to claim 3, characterized in that, Determining the standardized coordinates corresponding to the second local region based on the regional coordinates corresponding to the standardized region range includes: The coordinate values ​​corresponding to the candidate coordinates are mapped to the region coordinates to obtain the standardized coordinates corresponding to the second local region in the second thumbnail image.

5. The method according to any one of claims 1 to 4, characterized in that, The first thumbnail image is an image obtained by scaling the first registered image by a first scaling ratio; Before determining the registration coordinates of the second local region in the second thumbnail corresponding to the first local region based on the first registration parameters and the image coordinates corresponding to the first local region in the first thumbnail image, the method further includes: The first registered image is subjected to local segmentation processing to obtain at least one local region corresponding to the first registered image; Based on the first thumbnail ratio and the at least one local region, determine the image coordinates corresponding to the first local region in the first thumbnail image.

6. The method according to claim 5, characterized in that, The location information of the local area is represented by at least two coordinate points; Determining the image coordinates corresponding to the first local region in the first thumbnail image based on the first thumbnail ratio and the at least one local region includes: Determine the coordinate values ​​corresponding to at least two coordinate points in the local region; The coordinate values ​​corresponding to the at least two coordinate points are transformed according to the first scaling ratio to obtain the transformation results corresponding to the at least two coordinate points; Based on the transformation results corresponding to the at least two coordinate points, the image coordinates corresponding to the first local region in the first thumbnail image are determined.

7. The method according to any one of claims 1 to 4, characterized in that, The step of performing a second registration on the second registered image using the second registration parameters to obtain a registration result of the second registered image relative to the first registered image includes: The second local region in the second registered image is registered using the second registration parameters to obtain the local registration result corresponding to the second local region. Determine the coordinate information of the second local region in the second registered image; Based on the coordinate information, the local registration results corresponding to at least two second local regions in the second registration image are stitched together to obtain the registration result of the second registration image relative to the first registration image.

8. The method according to any one of claims 1 to 4, characterized in that, The step of determining the second registration parameters based on the standardized coordinates and the image coordinates corresponding to the first local region in the first thumbnail image includes: The image coordinates corresponding to the first local region and the standardized coordinates corresponding to the second local image are subjected to matrix transformation to obtain the second registration parameters.

9. The method according to any one of claims 1 to 4, characterized in that, The step of performing a first registration on a second thumbnail of the second registered image using a first thumbnail image corresponding to the first registered image, and determining the first registration parameters, includes: The first feature point of the first thumbnail image and the second feature point of the second thumbnail image are matched to obtain an intermediate registration image, which is the image obtained by registering the second thumbnail image based on the first thumbnail image; The first registration parameters are determined based on the intermediate registered image and the second thumbnail image.

10. The method according to claim 9, characterized in that, Determining the first registration parameters based on the intermediate registered image and the second thumbnail image includes: Determine the intermediate coordinates corresponding to the intermediate registered image and the thumbnail coordinates corresponding to the second thumbnail image. The intermediate coordinates are used to indicate the position information of the intermediate registered image, and the thumbnail coordinates are used to indicate the position information of the second thumbnail image. The image distance between the intermediate registered image and the second thumbnail image is determined using the intermediate coordinates and the thumbnail coordinates; The first registration parameters are determined with the goal of reducing the distance between the images.

11. An image processing apparatus, characterized in that, The device includes: The image acquisition module is used to acquire a first registration image and a second registration image, wherein the first registration image and the second registration image are images acquired from different acquisition angles for the target acquisition point; The first registration module is used to perform a first registration on a second thumbnail of the second registration image using a first thumbnail image corresponding to the first registration image, and to determine a first registration parameter. The second thumbnail image is an image obtained by scaling the second registration image by a second scaling ratio. The coordinate determination module is used to determine the registration coordinates of the second local region in the second thumbnail that corresponds to the first local region, based on the image coordinates corresponding to the first local region in the first thumbnail and the first registration parameters. The parameter determination module is used to adjust each registration coordinate point in the registration coordinates according to the second scaling ratio, determine the candidate coordinates corresponding to the second local region in the second thumbnail image, wherein the candidate coordinate points in the candidate coordinates correspond one-to-one with the registration coordinate points; perform standardization processing on the candidate coordinates to obtain standardized coordinates corresponding to the second local region in the second thumbnail image, wherein the standardization processing is used to assist the registration coordinates in changing from an irregular shape to a standard shape to obtain the second local region of a standard shape; and determine the second registration parameters based on the standardized coordinates and the image coordinates corresponding to the first local region in the first thumbnail image. The second registration module is used to perform a second registration on the second registration image using the second registration parameters, so as to obtain the registration result of the second registration image relative to the first registration image.

12. A computer device, characterized in that, The computer device includes a processor and a memory, the memory storing at least one program, which is loaded and executed by the processor to implement the image processing method as described in any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that, The storage medium stores at least one program segment, which is loaded and executed by a processor to implement the image processing method as described in any one of claims 1 to 10.

14. A computer program product, characterized in that, It includes computer instructions that, when executed by a processor, implement the image processing method as described in any one of claims 1 to 10.

Images