Method and system for image registration

Abstract

The invention relates to the technical field of image registration, in particular to an image registration method and system. The method comprises the following steps: acquiring a splicing key point between a replacement sub-image and an adjacent clear sub-image, acquiring a comparison registration rate in combination with a key point database, and determining an image registration state. According to the method, the standard image and the real-time image are subjected to image segmentation processing, subsequent image overall comparison is converted into regional sequential comparison processing, the relevance of image registration is improved, meanwhile, the splicing key points between the adjacent sub-images are obtained to serve as comparison reference of the blurred sub-images, and the image registration accuracy is improved. According to the image registration method, the fuzzy sub-images are obtained, the fuzzy sub-images are replaced with the clear sub-images at the corresponding positions, and the registration state of the real-time image is defined through the splicing key point matching rate between the adjacent clear sub-images, so that the image registration efficiency is improved, and the image registration steps are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of image registration, and in particular to a method and system for image registration. Background Art

[0002] Image registration is the process of matching and superimposing two or more images acquired at different times, with different sensors or under different conditions. It has been widely used in remote sensing data analysis, computer vision, image processing and other fields.

[0003] The traditional image registration process includes the following steps:

[0004] 1. First, extract the features of the two images to obtain feature points;

[0005] 2. Find matching feature point pairs by performing similarity measurement;

[0006] 3. Then, the image space coordinate transformation parameters are obtained through the matched feature point pairs;

[0007] 4. Finally, the coordinate transformation parameters are used for image registration.

[0008] In the process of image monitoring of some specific areas, such as the timed image acquisition of outdoor fixed areas, since most of the acquired images tend to be the same, only some areas are blurred due to changes in the outdoor environment. If the above-mentioned image registration is used, the overall blurred image needs to be clarified and then processed through the same registration process, which reduces the image registration efficiency and wastes system processing resources. It is not suitable for timed image acquisition of outdoor fixed areas.

[0009] In order to address the above problems, a method and system for image registration are urgently needed. Summary of the invention

[0010] The object of the present invention is to provide a method and system for image registration to solve the problems raised in the above background technology.

[0011] To achieve the above object, one of the objects of the present invention is to provide a method for image registration, comprising the following steps:

[0012] S1. Obtain standard image information and perform image segmentation processing on the standard image to generate a standard segmented image Se image ;

[0013] S2, combined with the standard segmentation image Se in S1 image , get the splicing key points Sk between adjacent sub-images points , establish a key point database;

[0014] S3, according to the standard image information in S1, collect the real-time images corresponding to different time points, and synchronously perform image segmentation to generate the real-time segmented image Rt image ;

[0015] S4, real-time segmentation image Rt image Evaluate image clarity and divide fuzzy sub-images Fus image And the clear sub-image Cls image ;

[0016] S5. Obtain each fuzzy sub-image Fus image In real-time segmentation image Rt image Internal position points, replace the standard segmentation image Se image The sub-image corresponding to the position points is marked as the replacement sub-image Res image ;

[0017] S6. Obtain replacement sub-image Res image With the adjacent clear sub-image Cls image The splicing key point Sk points , combined with the key point database in S2, obtain the comparison and registration rate and determine the image registration status.

[0018] As a further improvement of the technical solution, the standard image information in S1 includes the image acquisition time point, the image shooting angle and the continuous image shooting time interval;

[0019] The image acquisition time point is the time when the standard image is captured. For example, the image shooting angle at the same time point on different dates is the shooting angle of the monitoring instrument at the image acquisition time point.

[0020] The continuous image shooting time interval is the time interval between adjacent images when acquiring multiple groups of monitoring images.

[0021] As a further improvement of the technical solution, the method of performing image segmentation processing on the standard image in S1 comprises the following steps:

[0022] S1.1, loading the image to be processed into the memory and preprocessing it;

[0023] S1.2, using edge detection algorithm to detect edge points of the image to be processed, setting edge threshold and gradient direction, and connecting edge points into contours;

[0024] S1.3, refine the edge contour, eliminate redundant pixels, and obtain accurate contour;

[0025] S1.4, segment the image into different regions according to the edge contour, and establish marking coordinate points to mark each sub-region;

[0026] S1.5. Save the segmented image.

[0027] As a further improvement of the technical solution, the step of obtaining the splicing key points Sk between adjacent sub-images in S2 points The method comprises the following steps:

[0028] S2.1, identifying the connected regions between adjacent sub-images;

[0029] S2.2. Mark the landmarks that span adjacent sub-images in the connection area as splicing key points Sk points ;

[0030] S2.3. Distinguishing each splicing key point Sk points The positional relationship of the connected area, and remove the redundant splicing key points Sk points ;

[0031] S2.4, and locate each splicing key point Sk according to the marked coordinate points in S1.4 points Location.

[0032] As a further improvement of the technical solution, the real-time segmentation image Rt in S4 image The method for evaluating image clarity includes the following steps:

[0033] S4.1. Use gradient algorithm to obtain the pixel gray value Pi of each sub-image value ;

[0034] S4.2. Setting pixel grayscale threshold Compare the pixel grayscale values ​​Pi of each sub-image value ;

[0035] The pixel grayscale threshold will be lower than The sub-image is marked as fuzzy sub-image Fus image ;

[0036] Will not be lower than the pixel grayscale threshold The sub-image is marked as the clear sub-image Cls image .

[0037] As a further improvement of the technical solution, the method for obtaining the comparison and registration rate in S6 includes the following steps:

[0038] S6.1. Obtaining the replacement sub-image Res image With the adjacent clear sub-image Cls image The splicing key point Skpoints total;

[0039] S6.2. Combine the key point database to obtain the splicing key point Sk points The whole image structure is compared with the replaced whole image structure to obtain the splicing key point Sk corresponding to the successfully compared whole image structure. points ;

[0040] S6.3, combined with the splicing key point Sk in S6.1 points Total number, calculate the successful splicing key points Sk points The ratio of , marked as the alignment ratio;

[0041] S6.4. Establish a comparison and registration rate threshold, mark the real-time images whose comparison and registration rate is not less than the comparison and registration rate threshold as registered images, and mark the real-time images whose comparison and registration rate is less than the comparison and registration rate threshold as abnormal images.

[0042] As a further improvement of the technical solution, the comparison and registration rate is obtained in S6 using a threshold comparison algorithm, and the algorithm formula is as follows:

[0043]

[0044]

[0045] Among them, Cr rate is the alignment rate, Cot amount is the splicing key point Sk points Total, Cot amount Construct the corresponding stitching key point Sk for the successful overall image comparison points Total, F(Cr rate ) is the comparison function of the registration rate, is the alignment rate threshold.

[0046] The second object of the present invention is to provide a system for realizing a method for image registration, comprising an image segmentation module, a splicing key point acquisition module, a key point database, a sub-image clarity evaluation module, a fuzzy image replacement module and a registration state recognition module;

[0047] The image segmentation module is used to obtain standard image information and perform image segmentation processing on the standard image to generate a standard segmented image;

[0048] The splicing key point acquisition module is used to acquire the splicing key points between adjacent sub-images, and the key point database is used to store the splicing key point information between adjacent sub-images;

[0049] The sub-image clarity evaluation module is used to evaluate the image clarity of the real-time segmented image and divide it into a fuzzy sub-image and a clear sub-image;

[0050] The fuzzy image replacement module is used to obtain the position points of each fuzzy sub-image in the real-time segmented image, replace the sub-image at the corresponding position point of the standard segmented image, and mark it as a replacement sub-image;

[0051] The registration state recognition module is used to obtain the splicing key points between the replacement sub-image and the adjacent clear sub-image, and obtain the comparison registration rate in combination with the key point database to determine the image registration state.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] In the image registration method and system, image segmentation processing is performed on the standard image and the real-time image, and the subsequent overall image comparison is converted into a region-by-region comparison processing, so as to improve the relevance of the image registration, and at the same time, the splicing key points between adjacent sub-images are obtained as the comparison reference of the fuzzy sub-image, and the fuzzy sub-image is replaced with a clear sub-image at the corresponding position. The registration state of the real-time image is defined by the splicing key point matching rate between adjacent clear sub-images, thereby improving the image registration efficiency and reducing the image registration steps. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is a flowchart of the overall method of the present invention;

[0055] Figure 2 A flowchart of a method for performing image segmentation processing on a standard image according to the present invention;

[0056] Figure 3 This is one of the position point relationship diagrams of the present invention;

[0057] Figure 4 This is the second position point relationship diagram of the present invention;

[0058] Figure 5 The third position point relationship diagram of the present invention;

[0059] Figure 6 The invention obtains the splicing key point Sk between adjacent sub-images points Flow chart of the method steps;

[0060] Figure 7 The real-time segmentation image Rt of the present invention image A flowchart of the method for image clarity assessment;

[0061] Figure 8 A flow chart of a method for obtaining a comparison and registration rate according to the present invention;

[0062] Fig. 9 It is a system structure block diagram of the present invention. DETAILED DESCRIPTION

[0063] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] See also Figure 1-Figure 4 As shown, one of the purposes of the present invention is to provide a method for image registration, comprising the following steps:

[0065] S1. Obtain standard image information and perform image segmentation processing on the standard image to generate a standard segmented image Se image ;

[0066] S2, combined with the standard segmentation image Se in S1 image , get the splicing key points Sk between adjacent sub-images points , establish a key point database;

[0067] S3, according to the standard image information in S1, collect the real-time images corresponding to different time points, and perform image segmentation synchronously to generate the real-time segmented image Rt image ;

[0068] S4, real-time segmentation image Rt image Evaluate image clarity and divide fuzzy sub-images Fus image And the clear sub-image Cls image ;

[0069] S5. Obtain each fuzzy sub-image Fus image In real-time segmentation image Rt image Internal position points, replace the standard segmentation image Se image The sub-image corresponding to the position points is marked as the replacement sub-image Res image ;

[0070] S6. Obtaining the replacement sub-image Res image With the adjacent clear sub-image Cls image The splicing key point Sk points , combined with the key point database in S2, obtain the comparison and registration rate and determine the image registration status.

[0071] When used specifically, in the process of acquiring timed images of a fixed outdoor area, first obtain standard image information, that is, perform image acquisition of a specific area when the environment is in a stable state, such as when the light is sufficient to clearly capture the various geographical features of the monitored area, as a reference image for later comparison. Due to different monitoring conditions in the later stage, the acquired real-time image is prone to partial blurring, resulting in the inability to identify the geographical features of the image area. At this time, it is necessary to perform image segmentation processing on the standard image to generate a standard segmented image Se image , and perform refined segmentation processing for later comparison.

[0072] Complete the standard segmentation image Se image After processing, the standard segmentation image Se is obtained image The splicing key points Sk between adjacent sub-images points , establish a key point database. After the image segmentation processing, the connection relationship between the segmented sub-images is the same, that is, no matter at any time point, the same acquisition conditions are used, and the connection relationship between adjacent sub-images in the segmented image is the same. Specifically, in the real object, for example, it is necessary to perform image acquisition on a running machine. In the acquired mechanical image, after segmentation processing, different parts are in different sub-images. The parts in adjacent sub-images are fixed by bolts, welding and clamps, and their connecting parts are in the connecting area between adjacent sub-images. Even if one of the sub-images is in a blurred state, that is, the mechanical part in the sub-image cannot be identified, its adjacent connecting parts can be used as a reference for later comparison.

[0073] In the later specific acquisition process, it is necessary to collect real-time images corresponding to different time points according to the standard image information, and perform image segmentation simultaneously to generate real-time segmented images Rt image , and then the real-time segmentation image Rt image Evaluate image clarity and divide fuzzy sub-images Fus image And the clear sub-image Cls image , that is, the image content of the unrecognizable sub-image is divided into a fuzzy sub-image Fus image , which can identify the image content of the sub-image and divide it into clear sub-images Cls image ,like Figure 4 As shown in the figure, the sub-image marked as "M" is a fuzzy sub-image. When a fuzzy sub-image Fus appears in the real-time image being monitored image , it means that the real-time image cannot be subsequently registered. At this time, it is necessary to obtain various fuzzy sub-images Fus image In real-time segmentation image Rt image Internal position points, replace the standard segmentation image Se imageThe sub-image corresponding to the position points is marked as the replacement sub-image Res image ,like Figure 3-5 As shown in the figure, the position of the sub-image marked as "M" is (4, 5). During the replacement process, the sub-image at the corresponding position in the standard image is replaced to the position of the blurred sub-image, and finally the replacement sub-image Res is obtained. image With the adjacent clear sub-image Cls image The splicing key point Sk points , combined with the key point database, obtain the comparison and registration rate, and determine the image registration status.

[0074] In addition, the standard image information in S1 includes the image acquisition time point, the image shooting angle, and the time interval between consecutive image shootings;

[0075] The image acquisition time point is the time when the standard image is captured, such as the same time point on different dates.

[0076] The image shooting angle is the shooting angle of the monitoring instrument at the time of image acquisition;

[0077] The continuous image shooting time interval is the time interval between adjacent images when acquiring multiple sets of monitoring images.

[0078] Furthermore, the method for performing image segmentation processing on the standard image in S1 includes the following steps:

[0079] S1.1, loading the image to be processed into the memory and preprocessing it;

[0080] S1.2, using edge detection algorithm to detect edge points of the image to be processed, setting edge threshold and gradient direction, and connecting edge points into contours;

[0081] S1.3, refine the edge contour, eliminate redundant pixels, and obtain accurate contour;

[0082] S1.4, segment the image into different regions according to the edge contour, and establish marking coordinate points to mark each sub-region;

[0083] S1.5. Save the segmented image.

[0084] In the process of image segmentation of a standard image, the image to be processed is first loaded into the memory and preprocessed, including image denoising, smoothing and enhancement, to improve the effect of edge detection. Then, an edge detection algorithm is used to detect the edge points of the image to be processed. The Canny edge detection algorithm adopted in the present invention connects the edge points of the image to be processed into a contour by setting the edge threshold and gradient direction of the image. In order to distinguish the boundaries between each sub-region, it is necessary to refine the edge contour, eliminate redundant pixels, obtain an accurate contour, and divide the image into different regions according to the edge contour. The present invention adopts a region growing algorithm to fill the sub-region, and establishes a marking coordinate point to mark each sub-region. Figure 3 As shown in the figure, the entire image is divided into 5×5 image regions, and each sub-region is distinguished by coordinate points. The sub-region here is the sub-image, which is used as the later positioning position point points The reference basis is finally saved.

[0085] Furthermore, the splicing key points Sk between adjacent sub-images are obtained in S2 points The method comprises the following steps:

[0086] S2.1, identifying the connected regions between adjacent sub-images;

[0087] S2.2. Mark the landmarks that span adjacent sub-images in the connection area as splicing key points Sk points ;

[0088] S2.3. Distinguishing each splicing key point Sk points The positional relationship of the connected area, and remove the redundant splicing key points Sk points ;

[0089] S2.4, and locate each splicing key point Sk according to the marked coordinate points in S1.4 points Location.

[0090] In most cases, the shape and structure of the objects being photographed are complex, so the splicing key points Sk between different adjacent sub-images are points Different, in order to facilitate later recognition, it is first necessary to identify the connected areas between adjacent sub-images, such as Figure 3 The circle points shown in , then the landmarks across adjacent sub-images in the connection area are marked as splicing key points Sk points For example, the parts in adjacent sub-images are connected by bolts. In this case, the bolts span across two sub-images. The bolts can be used as reference objects and marked as splicing key points Sk pointsHowever, since there may be multiple identical reference objects in adjacent parts, in order to reduce the later recognition time, it is necessary to distinguish each splicing key point Sk in advance. points The position relationship of the connected area, such as the position height of the layer, and remove the redundant splicing key points Sk points For example, there are multiple connected parts with the same connection area between adjacent sub-images. Finally, according to the marked coordinate points, each splicing key point Sk is located. points Position, stored in the key point database.

[0091] Specifically, in S4, the real-time segmentation image Rt image The method for evaluating image clarity includes the following steps:

[0092] S4.1. Use gradient algorithm to obtain the pixel gray value Pi of each sub-image value ;

[0093] S4.2. Setting pixel grayscale threshold Compare the pixel grayscale values ​​P of each sub-image ivalue ;

[0094] The pixel grayscale threshold will be lower than The sub-image is marked as fuzzy sub-image Fus image ;

[0095] Will not be lower than the pixel grayscale threshold The sub-image is marked as the clear sub-image Cls image .

[0096] When used in real-time segmentation image Rt image In the process of image clarity evaluation, the gradient algorithm, such as Tenengrad gradient function, Laplacian gradient function and Brenner gradient method, is first used to obtain the pixel gray value Pi of each sub-image. value , pixel gray value Pi value Proportional to the image clarity, the pixel gray value Pi value The larger the value, the sharper the image edge and the richer the details, that is, the clearer the image. In order to further distinguish the fuzzy sub-image Fus image With the clear sub-image Cls image , the pixel grayscale threshold needs to be set in advance Compare the pixel grayscale values ​​Pi of each sub-image value , and will be lower than the pixel grayscale threshold The sub-image is marked as fuzzy sub-image Fus image , will not be lower than the pixel grayscale threshold The sub-image is marked as the clear sub-image Cls image .

[0097] In addition, the method for obtaining the comparison and registration rate in S6 includes the following steps:

[0098] S6.1. Obtaining the replacement sub-image Res image With the adjacent clear sub-image Cls image The splicing key point Sk points total;

[0099] S6.2. Combine the key point database to obtain the splicing key point Sk points The whole image structure is compared with the replaced whole image structure to obtain the splicing key point Sk corresponding to the successfully compared whole image structure. points ;

[0100] S6.3, combined with the splicing key point Sk in S6.1 points Total number, calculate the successful splicing key points Sk points The ratio of , marked as the alignment ratio;

[0101] S6.4. Establish a comparison and registration rate threshold, mark the real-time images whose comparison and registration rate is not less than the comparison and registration rate threshold as registered images, and mark the real-time images whose comparison and registration rate is less than the comparison and registration rate threshold as abnormal images.

[0102] Furthermore, the threshold comparison algorithm is used to obtain the comparison and registration rate in S6, and the algorithm formula is as follows:

[0103]

[0104]

[0105] Among them, Cr rate is the alignment rate, Cot amount is the splicing key point Sk points Total, Cot amount Construct the corresponding stitching key point Sk for the successful overall image comparison points Total, F(Cr rate ) is the comparison function of the registration rate, is the alignment rate threshold.

[0106] In the process of obtaining the comparison and registration rate, we first need to obtain the replacement sub-image Res image With the adjacent clear sub-image Cls image The splicing key point Sk points Total Cot amount It is worth noting that the total number here is excluding redundant splicing key points Sk points The number of subsequent key points is then combined with the key point database to obtain the splicing key point Skpoints The overall image construction, such as replacing the sub-image Res image With the adjacent clear sub-image Cls image The overall structure of the bolts between them, after replacement, is in the replacement sub-image Res image The bolt part and the adjacent clear sub-image Cls image The bolt parts are spliced ​​to form the overall bolt structure, and the overall image structure after replacement is compared to obtain the splicing key points Sk corresponding to the overall image structure that is successfully compared points , which is the splicing key point Sk points Total Cot amount , then combined with the splicing key point Sk in S6.1 points Total number, calculate the successful splicing key points Sk points The ratio is marked as the alignment ratio Cr rate , and finally set the alignment rate threshold When the alignment rate Cr rate Not less than the alignment rate threshold When the comparison registration rate comparison function F(Cr rate ) output is 1, indicating that the current real-time image is a registered image. When the registration rate Cr rate Below the alignment threshold When the comparison registration rate comparison function F(Cr rate ) output is 0, indicating that the current real-time image is an abnormal image.

[0107] The second object of the present invention is to provide a water purification method for image registration, including an image segmentation module, a splicing key point acquisition module, a key point database, a sub-image clarity evaluation module, a fuzzy image replacement module and a registration state recognition module.

[0108] The image segmentation module is used to obtain standard image information, perform image segmentation processing on the standard image, and generate a standard segmented image.

[0109] The splicing key point acquisition module is used to acquire the splicing key points between adjacent sub-images, and the key point database is used to store the splicing key point information between adjacent sub-images.

[0110] The sub-image clarity evaluation module is used to evaluate the image clarity of the real-time segmented image and divide it into fuzzy sub-images and clear sub-images.

[0111] The fuzzy image replacement module is used to obtain the position points of each fuzzy sub-image in the real-time segmented image, replace the sub-image at the corresponding position point of the standard segmented image, and mark it as the replacement sub-image.

[0112] The registration state recognition module is used to obtain the splicing key points between the replacement sub-image and the adjacent clear sub-image, and obtain the comparison registration rate in combination with the key point database to determine the image registration state.

[0113] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for image registration, characterized in that: The steps include: S1. Obtain standard image information and perform image segmentation processing on the standard image to generate a standard segmented image Se image ; S2, combined with the standard segmentation image Se in S1 image , get the splicing key points Sk between adjacent sub-images points , establish a key point database; S3, according to the standard image information in S1, collect the real-time images corresponding to different time points, and synchronously perform image segmentation to generate the real-time segmented image Rt image ; S4, real-time segmentation image Rt image Evaluate image clarity and divide fuzzy sub-images Fus image And the clear sub-image Cls image ; S5. Obtain each fuzzy sub-image Fus image In real-time segmentation image Rt image Internal position points, replace the standard segmentation image Se image The sub-image corresponding to the position points is marked as the replacement sub-image Res image ; S6. Obtaining the replacement sub-image Res image With the adjacent clear sub-image Cls image The splicing key point Sk points , combined with the key point database in S2, obtain the comparison and registration rate and determine the image registration status.

2. The image registration method according to claim 1, characterized in that: The standard image information in S1 includes the image acquisition time point, the image shooting angle and the continuous image shooting time interval; The image acquisition time point is the time when the standard image is taken, such as the same time point on different dates; The image shooting angle is the shooting angle of the monitoring instrument at the time of image acquisition; The continuous image shooting time interval is the time interval between adjacent images when acquiring multiple groups of monitoring images.

3. The image registration method according to claim 2, characterized in that: The method for performing image segmentation processing on the standard image in S1 comprises the following steps: S1.1, loading the image to be processed into the memory and preprocessing it; S1.2, using edge detection algorithm to detect edge points of the image to be processed, setting edge threshold and gradient direction, and connecting edge points into contours; S1.3, refine the edge contour, eliminate redundant pixels, and obtain accurate contour; S1.4, segment the image into different regions according to the edge contour, and establish marking coordinate points to mark each sub-region; S1.

5. Save the segmented image.

4. The method for image registration according to claim 3, characterized in that: The step S2 of obtaining the stitching key points Sk between adjacent sub-images points The method comprises the following steps: S2.1, identifying the connected regions between adjacent sub-images; S2.

2. Mark the landmarks that span adjacent sub-images in the connection area as splicing key points Sk points ; S2.

3. Distinguishing each splicing key point Sk points The positional relationship of the connected area, and remove the redundant splicing key points Sk points ; S2.4, and locate each splicing key point Sk according to the marked coordinate points in S1.4 points Location.

5. The method for image registration according to claim 1, characterized in that: In S4, the real-time segmentation image Rt is image The method for evaluating image clarity includes the following steps: S4.

1. Use gradient algorithm to obtain the pixel gray value Pi of each sub-image value ; S4.

2. Setting pixel grayscale threshold Compare the pixel grayscale values ​​Pi of each sub-image value ; The pixel grayscale threshold will be lower than The sub-image is marked as fuzzy sub-image Fus image ; Will not be lower than the pixel grayscale threshold The sub-image is marked as the clear sub-image Cls image .

6. The method for image registration according to claim 1, characterized in that: The method for obtaining the comparison and registration rate in S6 comprises the following steps: S6.

1. Obtaining the replacement sub-image Res image With the adjacent clear sub-image Cls image The splicing key point Sk points total; S6.

2. Combine the key point database to obtain the splicing key point Sk points The whole image structure is compared with the replaced whole image structure to obtain the splicing key point Sk corresponding to the successfully compared whole image structure. points ; S6.3, combined with the splicing key point Sk in S6.1 points Total number, calculate the successful splicing key points Sk points The ratio of , marked as the alignment ratio; S6.

4. Establish a comparison and registration rate threshold, mark the real-time images whose comparison and registration rate is not less than the comparison and registration rate threshold as registered images, and mark the real-time images whose comparison and registration rate is less than the comparison and registration rate threshold as abnormal images.

7. The method for image registration according to claim 6, characterized in that: The threshold comparison algorithm is used to obtain the comparison and registration rate in S6, and the algorithm formula is as follows: Among them, Cr rate is the alignment rate, Cot amount is the splicing key point Sk points Total, Cot amount Construct the corresponding stitching key point Sk for the successful overall image comparison points Total, F(Cr rate ) is the comparison function of the registration rate, is the alignment rate threshold.

8. A system for implementing the image registration method of claim 1, characterized in that: It includes image segmentation module, splicing key point acquisition module, key point database, sub-image clarity evaluation module, fuzzy image replacement module and registration state recognition module; The image segmentation module is used to obtain standard image information and perform image segmentation processing on the standard image to generate a standard segmented image; The splicing key point acquisition module is used to acquire the splicing key points between adjacent sub-images, and the key point database is used to store the splicing key point information between adjacent sub-images; The sub-image clarity evaluation module is used to evaluate the image clarity of the real-time segmented image and divide it into a fuzzy sub-image and a clear sub-image; The fuzzy image replacement module is used to obtain the position points of each fuzzy sub-image in the real-time segmented image, replace the sub-image at the corresponding position point of the standard segmented image, and mark it as a replacement sub-image; The registration state recognition module is used to obtain the splicing key points between the replacement sub-image and the adjacent clear sub-image, and obtain the comparison registration rate in combination with the key point database to determine the image registration state.

Images