Image matching method based on image feature point cutting area rearrangement

Abstract

The invention discloses an image matching method based on image feature point cutting area rearrangement. For image rotation and blurring, a local image and rearrangement method is used, and the matching stability is enhanced. The system mainly comprises an image cutting part, a data enhancement part, a segmented image rearrangement module and a feature description module. In the image cutting module, a local area with a fixed size around a feature point is obtained according to the position of the feature point, and irrelevant information of the whole image is prevented from interfering feature extraction. In the data enhancement part, an original image is subjected to Gaussian blur and immediate angle rotation to obtain enhanced data. In the segmented image rearrangement module, the small blocks are sequenced in a clockwise direction according to the distance of each segmented region relative to the position of the central point of the image. In this way, errors caused by image rotation can be effectively weakened.

Description

Technical Field

[0001] The invention relates to image analysis, and in particular to an image matching method based on rearrangement of cropped regions of image feature points. Background Art

[0002] The search and description of image feature points are of great significance in the field of computer vision. By identifying key feature points in an image, we can achieve many applications, such as target recognition, image registration, and 3D reconstruction. Common manual extraction methods include Harris corner detection, SIFT (Scale Invariant Feature Transform), and ORB (Oriented FAST and Rotated BRIEF). These methods detect corners, edges, or spots in an image and extract their local feature descriptors, so that the image can maintain stability under different scales, rotations, and lighting conditions. The search and description of feature points are widely used in computer vision. For example, in object recognition, objects are matched by comparing feature points in different images to achieve target detection and recognition; in image stitching, seamless image stitching is achieved by matching feature points in images; in SLAM (Simultaneous Localization and Mapping), robot positioning and map construction are achieved by extracting feature points in images. Therefore, accurate description of image feature points is critical to the realization of various computer vision tasks.

[0003] In image matching, robustness against rotation and blur is crucial. Rotation robustness ensures that the algorithm can recognize and match the same object in the image with different rotation angles, which is particularly critical in practical applications because the shooting angle is often uncontrollable. Blur robustness ensures that the algorithm can still accurately match when faced with blurred or low-definition images, which is particularly important in low-light conditions and motion blur. Image matching algorithms with rotation and blur robustness can handle a variety of scenes more stably, improve matching accuracy and reliability, and enable them to be more widely used in computer vision tasks such as target detection, image stitching, and scene reconstruction.

[0004] However, traditional manual feature point extraction has certain disadvantages. First, the feature extractor needs to be manually designed, which is difficult to adapt to different scenarios and changes. Second, for complex images, the feature points may be unstable or unable to extract effective information. In contrast, the advantage of neural networks is that they can automatically learn abstract features in images without manually designing feature extractors, thus avoiding human bias. In addition, deep learning models can be trained end-to-end through large-scale data, adapt to various complex scenarios and changes, and perform better. Therefore, neural network feature point acquisition has a wider application prospect in image processing and computer vision. Summary of the invention

[0005] The present invention solves the problems existing in the prior art and proposes an image matching method based on rearrangement of cropped regions of image feature points, which is a feature point description method based on an improved Vision Transformer network model.

[0006] The present invention is a feature point description network, so SIFT (Scale Invariant Feature Transform) is used as a feature point search method and SIFT is used as a feature point position acquisition module.

[0007] The present invention uses the feature point as the center and obtains a local area of ​​24*24 pixels around the feature point by cropping, which serves as the image basis for feature description. Because there are multiple advantages to using the local area around the feature point instead of the full image area in image description. First, the local area description can more accurately capture the detailed information around the feature point, while the full image description may contain a lot of redundant information, increasing the computational complexity. Second, the local area description is more robust to partial occlusion and background changes of the image, because it only focuses on specific key areas and is not easily disturbed by other parts of the full image.

[0008] In order to enhance the stability of image feature points, the present invention adopts a series of image processing steps. First, the present invention performs a rotation transformation on the input local image to obtain a rotated cropped image. This step helps to make the performance of feature points more stable in different directions. Secondly, the original image is Gaussian blurred to generate a blurred image. The influence of noise on feature point detection is reduced, thereby improving the stability of feature points. Finally, the original local image, the rotated local image, and the blurred local image are input as homologous images to ensure that the algorithm can identify stable feature points under different conditions. Through these processing steps, we can more reliably obtain the feature points of the image and achieve better results in subsequent image processing or computing tasks.

[0009] The Vision Transformer model modifies the Transformer model and applies it to image classification because the Transformer has significant advantages in image description. Through the self-attention mechanism, the Transformer can capture long-range dependencies and complex contextual information in the image, thereby more comprehensively understanding the global features and local details of the image. Its parallel processing capability improves computational efficiency and adapts to large-scale data. In addition, the Transformer does not need to manually design feature extractors, and automatically learns features, making it more flexible and accurate.

[0010] In the feature point description module of the present invention, the Vision Transformer model is improved in a targeted manner. Specifically, we removed the original sequence part used for classification tasks in the design of the model. In the traditional Vision Transformer model, the classification task involves label prediction for the image, which is not required in our application scenario. Therefore, we removed this part of the structure and redesigned the decoder part of the model. After being processed by the decoder, we splice the results of the decoder output, and then process them through a fully connected dimensionality reduction module to finally generate a feature descriptor for the input local image. This process effectively converts the feature information in the image into a compact descriptor form, which is convenient for subsequent feature matching and recognition tasks.

[0011] A method for image matching based on rearrangement of cropped regions of image feature points comprises the following steps:

[0012] 1) Obtain the original image, randomly rotate the original image to obtain a rotated image, randomly blur the original image to obtain a blurred image, and then splice the original image, the rotated image and the blurred image to obtain an enhanced data set;

[0013] 2) Using the scale-invariant feature transformation method to obtain the feature point positions from the original image in the enhanced data set, and obtaining the local area around the feature point positions from the original image, the rotated image, and the blurred image in the enhanced data set according to the feature point positions as the feature description image;

[0014] 3) cropping and rearranging the feature description image to obtain a rearranged cropping region;

[0015] 4) Input the rearranged cropped areas into the Vision Transformer model to obtain the descriptors corresponding to each cropped area, concatenate and descend the descriptors corresponding to each cropped area, obtain the final description vector of the local image of the feature point, and obtain the feature description network;

[0016] 5) Optimizing the feature description network through the loss function to obtain an optimized feature description model;

[0017] 6) Match the two images to be matched through the optimized feature description model.

[0018] In step 1), the original image is randomly rotated to obtain a rotated image, which specifically includes:

[0019] The original image is randomly rotated within the range of 50 degrees clockwise to 50 degrees counterclockwise to obtain a rotated image.

[0020] In step 1), the original image is randomly blurred to obtain a blurred image, which specifically includes:

[0021] The original image is subjected to a random convolution with a Gaussian blur of a kernel size of 3*3 or 5*5 to obtain a blurred image.

[0022] In step 2), the local area around the feature point position refers to a (20-28)*(20-28) pixel area within the feature point position range.

[0023] In step 3), the feature description image is cropped and rearranged, specifically including:

[0024] 3.1) First, the obtained feature description image is cropped to obtain multiple cropped areas, and they are numbered;

[0025] 3.2) According to the coordinates of the center point of the cropped area, find the cropped area closest to the center point of the image and mark it as 1. Then mark the cropped areas in a circle outside the cropped image clockwise as 2 to 4, and mark the outer circle as 6-9. Rearrange the cropped areas from small to large according to the numbers.

[0026] In step 3.2), the coordinates of the center point of the cropping area are determined by the formula:

[0027]

[0028] Among them, (x i ,y i ) is the coordinate of the center point of the cropping area, i is the cropping area numbered from 1 to 9, (x_left, y_left) is the coordinate of the upper left corner of the cropping area, w, h are the two side lengths of the cropping area, w, h are both 8 pixels, and the size of the feature description image is 24*24 pixels.

[0029] In step 5), the final description vectors of the local image of the feature points are divided into groups of three and calculated separately. The loss function is as follows:

[0030] Loss = Loss1 + Loss2; (1)

[0031]

[0032] In formula (1), Loss is the final loss function value used for back propagation;

[0033] In formula (2), n is the number of feature points of the original image, that is, the final description vector of the local image of n feature points; j is the image type, where 0 represents the original image, and when j is 1-2, it represents a rotated image and a blurred image respectively; is the final description vector of the local image of the feature point; mean() means finding the mean.

[0034] In formula (3), n is the number of feature points of the original image, that is, the final description vector of the local image of n feature points; j is the image type, when j is 1-3, it represents the original image, the rotated image, and the blurred image respectively; is the final description vector of the local image corresponding to the feature point; Max is the corrected loss value, ensuring that the entire loss value Loss2 is a non-negative value; mean() means finding the mean.

[0035] In the encoding module, we use the standard transformer module. The specific architecture of this part is as follows Figure 1 As shown in Figure 2, the transformer encoding module extracts high-level features from the image through the self-attention mechanism and feedforward network, ensuring the accuracy and robustness of the feature description.

[0036] The complete process of the network is as follows Figure 2 As shown, it includes various processing steps from input image to final feature descriptor. Figure 3 The original structure of the Vision Transformer model is shown to facilitate comparison and understanding of the improvements we made. In the model of the present invention, we focus on optimizing the accuracy and efficiency of feature point description so that it can better serve feature matching and recognition tasks in practical applications.

[0037] In order to enhance the rotation stability of the image, the present invention proposes a method based on segmentation and sorting. First, the input local image is segmented into several small blocks. Then, according to the distance of each segmented area relative to the center point of the image, these small blocks are sorted in a clockwise direction. In this way, the error caused by image rotation can be effectively reduced, thereby improving the robustness of the model to the rotated image. The specific process is as follows: Figure 4 shown.

[0038] In the loss function design, the present invention adopts the sum of the similarity between homologous local images and the difference between the same images as the loss function of back propagation to enhance the ability of the network.

[0039] For local images of the same source, the descriptors obtained by the network for the original image, the rotated image, and the blurred image are required to be as close as possible, and the network matching accuracy is positively correlated with the similarity between the two. Therefore, the calculation method of the loss function (Loss1) is as follows: n is the number of feature points of the original image, that is, the final description vector of the local image with n feature points; j is the image type, where 0 represents the original image, and when j is 1-2, it represents the rotated image and the blurred image respectively; is the final description vector of the local image of the feature point; mean() means to find the mean:

[0040]

[0041] For the descriptors obtained from the local images in the same image, we need the descriptor gap to be as large as possible to better distinguish different feature points. The network matching accuracy is negatively correlated with the similarity between the two, so we set a larger value (Max). The specific loss function Loss2 expression is as follows, where n is the number of feature points in the original image, that is, the final description vector of the local image with n feature points; j is the image type, and when j is 1-3, it represents the original image, the rotated image, and the blurred image respectively; is the final description vector of the local image corresponding to the feature point; Max is the corrected loss value, ensuring that the entire loss value Loss2 is a non-negative value; mean() means finding the mean:

[0042]

[0043] Finally, the final loss function Loss used for back propagation, where n1 is the calculation formula as follows:

[0044] Loss = Loss1 + Loss2

[0045] The present invention relates to a feature point cropping region description method based on a Vision Transformer neural network for rotation and blur, which is divided into image enhancement, local image segmentation shift and feature point description network. In the data enhancement part, the original image is rotated and Gaussian blurred to become three homologous images. In the local image segmentation shift module, the segmented images are rearranged according to the position of the local image in the complete image to enhance the robustness of the feature description network for rotation. In the feature description module, the present invention modifies the VIT model, removes its branch module for classification, and focuses on image description, connects the descriptors of the segmented part, and obtains a complete description of the local image. Then it is reduced in dimension to obtain the final local area descriptor. Its loss function consists of two parts, including loss1, a descriptor similarity of local images in the same image, and loss2, a similarity between homologous local images. Among them, for loss2, we need it to be as large as possible, so that the difference between feature point descriptors is greater, and the matching accuracy is improved. Then the sum of loss1 and loss2 is taken as the final loss value.

[0046] Compared with the prior art, the present invention has the following advantages:

[0047] (1) A local image description method based on the Vision Transformer neural network is proposed. Compared with other neural network methods, it greatly enhances the stability and accuracy of image rotation and blur.

[0048] (2) Local images are used to obtain descriptors of feature points, so that the network can focus on the features of the area around the feature points, reduce the details of irrelevant information in other areas of the entire image, and enhance the accuracy of the description.

[0049] (3) The use of data enhancement technology to generate homologous images can improve the neural network's ability to recognize feature points at different angles and blur levels, enhance the network's generalization ability, and make it more robust in actual scenarios.

[0050] (4) After the input image is segmented, the segmentation map is rearranged according to the relative positions of its feature points, which enhances the network's robustness to rotation.

[0051] (5) In the loss function, the sum of the similarities between corresponding feature points and the differences between non-corresponding feature points is used as the loss value, and no standard labels are required. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is the transformer encoding module;

[0053] Figure 2 It is the overall network flow diagram of the present invention;

[0054] Figure 3 Original Vision Transformer neural network flow chart;

[0055] Figure 4 Crop image reordering instructions;

[0056] Figure 5 is the matching result between the original image and the Gaussian blurred image;

[0057] Figure 6 is the matching result between the original image and the rotated image; DETAILED DESCRIPTION

[0058] The present invention is further described in detail below in conjunction with embodiments, but the protection scope of the present invention is not limited thereto.

[0059] The invention relates to a feature point cropping region description method based on an improved Vision Transformer neural network model, which mainly comprises image cropping, a data enhancement part, a segmented image rearrangement module and a feature description module.

[0060] The present invention is a feature point description network, so SIFT (Scale Invariant Feature Transform) is used as a feature point search method, and the SIFT method is adopted to obtain the position information of the feature points.

[0061] The present invention takes the feature point as the center and obtains a local area of ​​24*24 pixels around the feature point by cropping, which serves as the image basis for feature description. This is because the local area description can more accurately capture the detailed information around the feature point, while the full image description may contain a large amount of redundant information, increasing the computational complexity. Secondly, the local area description is more robust to partial occlusion and background changes of the image, because it only focuses on specific key areas and is not easily interfered by other parts of the full image. The reason for choosing an area of ​​24*24 pixels is that an area of ​​this size can focus on the detailed information around the feature point, avoid including too much irrelevant background or other features, and at the same time process less data, thereby improving computational efficiency.

[0062] In the data enhancement part, the present invention performs a rotation transformation on the input local image to obtain a rotated image. Then the original image is subjected to Gaussian blur processing to generate a blurred image. Finally, the original image, the rotated image and the blurred image are input as homologous images. The original image list with a length of n is enhanced to an enhanced image with a length of 3n.

[0063] In the segmented image rearrangement module, the input full image size is 3*224*224 pixels, the local area size is 3*24*24 pixels, and the local area is divided into 9 3×3 cropping areas starting from the upper left corner. The channel, width, and height of each cropping area are 3*8*8 pixels. Then we get the coordinates of the center point of each cropping area. The specific formula is as follows, where i is the cropping area number from 1 to 9, (x left ,y left ) is the coordinate of the upper left corner of the cropping area, w,h is the width of the local area, here is 24 pixels:

[0064]

[0065] According to the obtained coordinates of the cropped area, the distances from each to the center point (112, 112) of the image are obtained, and the corner area with the closest coordinates is obtained and marked as 1. Then, the cropped area in the outer circle of the cropped image is marked as 2 to 4 clockwise, and the outer circle is marked as 6-9. The cropped areas are rearranged from small to large according to the numbers. In this way, when the image is rotated, the relative invariance between the areas can be guaranteed. The error caused by image rotation can be effectively reduced.

[0066] The obtained cropped small blocks are flattened into a sequence and input into the feature description module one by one. The feature description module is composed of three stacked Transformer encoders, and each encoder layer consists of a multi-head self-attention mechanism (Multi-Head Self-Attention) and a feed-forward neural network (Feed-Forward Neural Network).

[0067] The multi-head attention has h heads. For each head i, calculate the attention weight matrix A i The expression is as follows, where Q i ,k i , and V i are query, key, and value matrices, d i is the dimension of each head. :

[0068]

[0069] The output of the self-attention mechanism is passed to the feedforward neural network, which consists of two linear layers and a nonlinear activation function. The mathematical expression is as follows:

[0070] F x =Relu(w1*x+b1)*w2+b2

[0071] The network obtains the description vectors corresponding to each segmented small block. Because the complete local image is composed of segmented small blocks, we concatenate the description vectors of the segmented small blocks to obtain [x1,x2…x8,x9], where x1 to x9 are the feature vectors of the cropped area. Then, the fully connected module is used for dimensionality reduction to obtain the final 16-dimensional descriptor.

[0072] In order to enhance the matching stability of the network, in the loss function design, the present invention adopts the sum of the similarity between homologous local images and the difference between the same images as the back-propagation loss function to enhance the network's capabilities.

[0073] For local images of the same source, the descriptors obtained by the network for the original image, the rotated image, and the blurred image are required to be as close as possible, and the network matching accuracy is positively correlated with the similarity between the two. Therefore, the calculation method of this part of the loss function is as follows, where n is the number of feature points in the original image, that is, the final description vector of the local image with n feature points; j is the image type, where 0 represents the original image, and when j is 1-2, it represents the rotated image and the blurred image respectively; is the final description vector of the local image of the feature point; mean() means to find the mean:

[0074]

[0075] For the descriptors obtained from the local images in the same image, we need the descriptor gap to be as large as possible to better distinguish different feature points. The network matching accuracy is negatively correlated with the similarity between the two, so we set a larger value (Max). The specific loss function Loss2 expression is as follows, where n is the number of feature points in the original image, that is, the final description vector of the local image with n feature points; j is the image type, and when j is 1-3, it represents the original image, the rotated image, and the blurred image respectively; is the final description vector of the local image corresponding to the feature point; Max is the corrected loss value, ensuring that the entire loss value Loss2 is a non-negative value; mean() means finding the mean:

[0076]

[0077] The final loss function Loss calculation formula used for back propagation is as follows:

[0078] Loss = Loss1 + Loss2

[0079] 1) Experimental conditions

[0080] The model is implemented using the PyTorch framework. All experiments are conducted on the Microsoft Windows Server 2019 Datacenter 10.0.17763 operating system. The CPU is Intel (R) Xeon (R) Gold 6326 CPU @ 2.90 GHz, the memory is 256 GB, and the RTX A6000 is used for acceleration during training. The training images come from the caltech101 dataset, which consists of 101 categories of object images. Each category contains about 40 to 800 images, totaling about 9k images. When the training images are input into the model, they are uniformly adjusted to 224 × 224 pixels. The model is trained in an end-to-end manner using the Adam optimization algorithm, with the learning rate set to 1e-4 and the batch size set to 15. Finally, it reaches convergence after 200 epochs.

[0081] 2) Experimental results

[0082] Figure 5 This is the result of feature point matching of blurred image. It can be seen that the network has a high accuracy in matching feature points of blurred image.

[0083] Figure 6 This is the feature point matching result of the image rotated at a large angle. It can be seen that the network has a high accuracy rate in matching feature points of the image rotated at a small angle.

[0084] The present invention adopts matching accuracy as an evaluation index.

[0085] Table 1 shows the matching accuracy of the model proposed in the present invention under 3*3 convolution kernel Gaussian blur, 5*5 convolution kernel Gaussian blur, range random angle rotation and range random angle rotation plus 3*3 convolution kernel Gaussian blur.

[0086] Table 1

[0087]

[0088] Comprehensive experimental results show that the matching model proposed in this invention has good learning and generalization capabilities. Compared with classic and popular matching methods in the field, it has higher precision, more accurate matching effects and faster reasoning speed in both quantitative indicators and qualitative segmentation results, and can effectively and efficiently cope with various complex matching scenarios.

[0089] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0090] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0091] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0092] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0093] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0094] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for image matching based on rearrangement of cropped regions of image feature points, characterized in that: The following steps are involved: 1) Obtain the original image, randomly rotate the original image to obtain a rotated image, randomly blur the original image to obtain a blurred image, and then splice the original image, the rotated image and the blurred image to obtain an enhanced data set; 2) Use the scale-invariant feature transformation method to obtain the feature point positions from the original image in the enhanced dataset, and obtain the local area around the feature point positions from the original image, rotated image and blurred image of the enhanced dataset according to the feature point positions as the feature description image; 3) cropping and rearranging the feature description image to obtain a rearranged cropping region; 4) Input the rearranged cropped areas into the Vision Transformer model to obtain the descriptors corresponding to each cropped area, concatenate and descend the descriptors corresponding to each cropped area, obtain the final description vector of the local image of the feature point, and obtain the feature description network; 5) Optimize the feature description network through the loss function to obtain the optimized feature description model; 6) Match the two images to be matched through the optimized feature description model.

2. The image matching method based on rearrangement of image feature point cropping regions according to claim 1, characterized in that: In step 1), the original image is randomly rotated to obtain a rotated image, which specifically includes: The original image is randomly rotated within the range of 50 degrees clockwise to 50 degrees counterclockwise to obtain a rotated image.

3. The image matching method based on rearrangement of image feature point cropping regions according to claim 1, characterized in that: In step 1), the original image is randomly blurred to obtain a blurred image, which specifically includes: The original image is subjected to a random convolution with a Gaussian blur of a kernel size of 3*3 or 5*5 to obtain a blurred image.

4. The image matching method based on rearrangement of image feature point cropping regions according to claim 1, characterized in that: In step 2), the local area around the feature point position refers to a (20-28)*(20-28) pixel area within the feature point position range.

5. The image matching method based on rearrangement of image feature point cropping regions according to claim 1, characterized in that: In step 3), the feature description image is cropped and rearranged, specifically including: 3.1) First, the obtained feature description image is cropped to obtain multiple cropped areas, and they are numbered; 3.2) According to the coordinates of the center point of the cropped area, find the cropped area closest to the center point of the image and mark it as 1. Then mark the cropped areas in a circle outside the cropped image clockwise as 2 to 4, and mark the outer circle as 6-9. Rearrange the cropped areas from small to large according to the numbers.

6. The image matching method based on rearrangement of image feature point cropping regions according to claim 5, characterized in that: In step 3.2), the coordinates of the center point of the cropping area are determined by the formula: Among them, (x i ,y i ) is the coordinate of the center point of the cropping area, i is the cropping area number from 1 to 9, (x left ,y left ) is the coordinate of the upper left corner of the cropped area, w and h are the two side lengths of the cropped area, w and h are both 8 pixels, and the size of the feature description image is 24*24 pixels.

7. The image matching method based on rearrangement of image feature point cropping regions according to claim 1, characterized in that: In step 5), the loss function is as follows: Loss = Loss1 + Loss2; (1) In formula (1), Loss is the final loss function value used for back propagation; In formula (2), n is the number of feature points of the original image; j is the image type; is the final description vector of the local image of the feature point; mean() means to calculate the mean; In formula (3), n is the number of feature points of the original image; j is the image type; is the final description vector of the local image corresponding to the feature point; Max is the corrected loss value; mean() means finding the mean.

Images