Registration method of visible light and thermal infrared remote sensing images
By constructing a spectral mapping model and feature matching, combined with local module registration, the problem of limited robustness and accuracy in the registration of visible light and thermal infrared remote sensing images was solved, achieving efficient and noise-resistant image registration results.
Patent Information
- Application Number
- CN202511397813.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-28
- Publication Date
- 2025-12-26
AI Technical Summary
Existing registration methods for visible light and thermal infrared remote sensing images are limited in robustness and accuracy when faced with problems such as significant differences in grayscale of remote sensing images, noise interference, large image size and local distortion. Furthermore, methods based on feature points and transform fitting are prone to distortion of global transform parameters.
By constructing a spectral mapping model, image registration is performed using the spectral information of different ground features in the visible and thermal infrared bands. Combined with feature matching and local module registration, high-precision image registration is achieved.
It improves the robustness and accuracy of image registration, solves the problems of limited overall robustness and accuracy and distortion of global transformation parameters in traditional methods, and achieves efficient and noise-resistant image registration.
Smart Images
Figure CN121213624A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present disclosure relates to the technical field of remote sensing data processing and analysis, and particularly relates to a registration method of visible light and thermal infrared remote sensing images. BACKGROUND
[0002] Visible light and thermal infrared collaborative remote sensing is a key means of modern remote sensing observation, which can rely on satellites, unmanned aerial vehicles, near-ground cameras and other carriers, and is widely used in crop growth monitoring, disaster assessment, target reconnaissance and other tasks. By carrying visible light cameras and thermal infrared imagers and with the help of high-precision positioning and attitude synchronization systems, complementary information of ground objects in visible and near-infrared bands and long-wave thermal infrared bands can be obtained synchronously, high spatial resolution and continuous day and night observation can be realized, and the advantages of rich texture details and thermal radiation sensitivity can be combined. Registration of visible light and thermal infrared images is a necessary preprocessing link for subsequent collaborative analysis.
[0003] Existing registration methods of visible light and thermal infrared remote sensing images include whole-image transformation methods based on gray level and phase correlation, and methods based on feature points and transformation fitting. However, the whole-image transformation methods based on gray level and phase correlation are limited in robustness and accuracy when facing problems such as significant gray level difference, noise interference, large image size and local distortion. The methods based on feature points and transformation fitting may have problems of global transformation parameter distortion when facing problems such as significant color difference, insufficient features and uneven distribution of feature points. SUMMARY
[0004] In view of the above problems, an embodiment of the present disclosure provides a registration method of visible light and thermal infrared remote sensing images.
[0005] One aspect of the present disclosure provides a registration method of visible light and thermal infrared remote sensing images, comprising: performing spectral mapping on a visible light band image of a target region by using a spectral mapping model to obtain a mapped thermal infrared band image, the spectral mapping model being constructed by visible light spectral information and thermal infrared band spectral information of a plurality of ground objects in the target region; performing coarse registration on a thermal infrared band image of the target region and the mapped thermal infrared band image to obtain a coarse registration thermal infrared band image; performing feature matching on the mapped thermal infrared band image and the coarse registration thermal infrared band image to determine a feature pair set of the target region, the feature pair set comprising a plurality of matched feature pairs, and each feature pair representing a pair of matched feature points; and performing registration on the visible light band image and the thermal infrared band image according to the feature pair set.
[0006] According to an embodiment of the present disclosure, before the visible light band image of the target region is subjected to spectral mapping by using the spectral mapping model to obtain the mapped thermal infrared band image, the method further includes: obtaining an initial visible light band image of the target region, performing ground feature classification on the initial visible light band image to obtain a ground feature classification result, the ground feature classification result representing differences in gray values between different types of ground features in different bands; determining a reference band according to the ground feature classification result, and taking the initial visible light band image corresponding to the reference band as the visible light band image.
[0007] According to an embodiment of the present disclosure, the construction process of the spectral mapping model includes: obtaining an initial spectral mapping model and visible light spectrum information and thermal infrared band spectrum information of a plurality of ground features; inputting the visible light spectrum information and the thermal infrared band spectrum information of each ground feature in the plurality of ground features into the initial spectral mapping model respectively to analyze mapping relationships between the visible light spectrum information and the thermal infrared band spectrum information in each ground feature; and adjusting internal parameters of the initial spectral mapping model according to the plurality of mapping relationships to obtain the spectral mapping model.
[0008] According to an embodiment of the present disclosure, before the thermal infrared band image of the target region is coarsely registered with the mapped thermal infrared band image, the method further includes: up-sampling the thermal infrared band image to make the spatial resolutions of the thermal infrared band image and the mapped thermal infrared band image consistent.
[0009] According to an embodiment of the present disclosure, after the feature pair set of the target region is determined by performing feature matching on the mapped thermal infrared band image and the coarsely registered thermal infrared band image, the method further includes: in a case where the number of feature pairs in the feature pair set is less than a first preset threshold or the uniformity of the feature pairs is less than a second preset threshold, constructing a first pseudo feature pair from two points respectively from the mapped thermal infrared band image and the coarsely registered thermal infrared band image at the same image coordinates; constructing a plurality of first pseudo feature pairs and performing local module registration on the first pseudo feature pairs to obtain registered first pseudo feature pairs; and adding at least one registered first pseudo feature pair to the feature pair set until the total number of the registered first pseudo feature pairs and the feature pairs in the feature pair set is greater than or equal to the first preset threshold and the total uniformity of the registered first pseudo feature pairs and the feature pairs is greater than or equal to the second preset threshold.
[0010] According to an embodiment of the present disclosure, after the feature pair set of the target region is determined by performing feature matching on the mapped thermal infrared band image and the coarsely registered thermal infrared band image, the method further includes: dividing the coarsely registered thermal infrared band image into a plurality of sub-images to divide the coarsely registered thermal infrared band image into the plurality of sub-images, the plurality of sub-images respectively corresponding to a plurality of sub-regions in the target region; and distributing the feature pair set according to the plurality of sub-images to determine sub-feature pair sets of the plurality of sub-regions.
[0011] According to an embodiment of the present disclosure, after the feature pair set is distributed according to the plurality of sub-images to determine the plurality of sub-region feature pair sets, the method further comprises: in a case where a number of any one of the plurality of sub-feature pair sets is less than a third preset threshold or uniformity between the plurality of sub-images is less than a fourth preset threshold, constructing two points at a same image coordinate from the mapping thermal infrared band image and the sub-image corresponding to the sub-feature pair set respectively as a second pseudo feature pair; constructing a plurality of second pseudo feature pairs and performing local module registration on the second pseudo feature pairs to obtain registered second pseudo feature pairs; and adding the registered second pseudo feature pairs to the sub-feature pair set until a number of the registered second pseudo feature pairs and the sub-feature pairs in the sub-feature pair set is greater than or equal to the third preset threshold and the uniformity between the plurality of sub-images is greater than or equal to the fourth preset threshold; wherein the uniformity between the plurality of sub-images is determined by a distance between a plurality of average center points of the plurality of sub-images, and the average center points are determined by a plurality of feature points in the sub-image.
[0012] According to an embodiment of the present disclosure, the method of performing block division on the coarsely registered thermal infrared band image comprises: determining a center point of the coarsely registered thermal infrared band image; and performing block division on the coarsely registered thermal infrared band image according to a distance between each of the plurality of sub-images and the center point, the size of the sub-image being inversely proportional to the distance between the sub-image and the center point.
[0013] According to an embodiment of the present disclosure, the method of registering the visible light band image and the thermal infrared band image according to the feature pair set comprises: performing target transformation matrix fitting by using a deviation of each of the feature pairs in the feature pair set to obtain a target transformation matrix; and registering the visible light band image and the thermal infrared band image according to the target transformation matrix.
[0014] According to an embodiment of the present disclosure, the method of coarsely registering the thermal infrared band image of the target region and the mapping thermal infrared band image comprises: coarsely registering the thermal infrared band image of the target region and the mapping thermal infrared band image by using a phase correlation method.
[0015] The present method constructs a spectral mapping model by using spectral information of different ground objects in the visible light and thermal infrared bands as prior knowledge, fully utilizes spectral characteristics of different types of ground objects in the visible light band and the thermal infrared band, and makes the constructed mapping model accurate and reliable, so that the effective conversion of the visible light reference band spectral feature to the thermal infrared band can be completed. In addition, in the present method, the spectral mapping model constructed in advance is used to map the thermal infrared band with a large color difference from the visible light color to a mapping thermal infrared band with a similar color and perform registration, without bearing the influence of high color difference, so that the registration effect is good. The present method can solve the problems of limited overall robustness and precision and global transformation parameter distortion in the traditional registration method. BRIEF DESCRIPTION OF DRAWINGS
[0016] The above and other objects, features and advantages of the present disclosure will become more apparent from the following description when taken in conjunction with the accompanying drawings, in which:
[0017] Figure 1 A flowchart of a method for registering visible light and thermal infrared remote sensing images according to an embodiment of the present disclosure is schematically shown;
[0018] Figure 2 A framework diagram of a method for registering visible light and thermal infrared remote sensing images according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION
[0019] To make the objects, technical solutions and advantages of the present disclosure clearer, the present disclosure will be further described in detail below with reference to specific embodiments and drawings.
[0020] It should be noted that similar or identical parts are denoted by the same reference numerals in the drawings or the description. The technical features in the example embodiments shown in the description can be freely combined as long as there is no conflict. In addition, each claim can stand alone as an embodiment or the technical features in each claim can be combined as a new embodiment, and in the drawings, the shape or thickness of the embodiments can be exaggerated and simplified for the purpose of convenience or brevity. Furthermore, elements or implementation manners not shown or described in the drawings are in the form known to one of ordinary skill in the art. In addition, although examples can provide parameters including specific values, it should be understood that the parameters do not need to be exactly equal to the corresponding values, but can be approximately equal to the corresponding values within an acceptable error tolerance or design constraint.
[0021] The various embodiments of the present disclosure described above can be freely combined to form further embodiments, which are all within the scope of the present disclosure, unless there is a technical obstacle or contradiction.
[0022] Although the present disclosure is described with reference to the drawings, the embodiments disclosed in the drawings are intended to exemplarily illustrate the preferred embodiments of the present disclosure and cannot be understood as a limitation of the present disclosure. The dimensional proportions in the drawings are merely schematic and cannot be understood as a limitation of the present disclosure.
[0023] Although some embodiments of the general concept of the present disclosure have been shown and described, those of ordinary skill in the art will understand that changes can be made in these embodiments without departing from the principles and spirit of the general concept of the present disclosure, and the scope of the present disclosure is defined by the claims and their equivalents.
[0024] Figure 1 A flowchart of a method for registering visible light and thermal infrared remote sensing images according to an embodiment of the present disclosure is schematically shown.
[0025] As Figure 1 shown, the embodiment of the present disclosure provides a registration method of visible light and thermal infrared remote sensing images, comprising operations S110-S140.
[0026] In operation S110, the visible light band image of the target area is spectrally mapped by using a spectral mapping model to obtain a mapped thermal infrared band image, and the spectral mapping model is constructed by using the visible light spectral information and the thermal infrared band spectral information of multiple ground objects in the target area.
[0027] In operation S120, the thermal infrared band image of the target area is coarsely registered with the mapped thermal infrared band image to obtain a coarsely registered thermal infrared band image.
[0028] In operation S130, the mapped thermal infrared band image and the coarsely registered thermal infrared band image are feature matched to determine a feature pair set of the target area, and the feature pair set comprises multiple matched feature pairs, and the feature pair represents a pair of matched feature points.
[0029] In operation S140, the visible light band image and the thermal infrared band image are registered according to the feature pair set.
[0030] In some embodiments, the registration method of visible light and thermal infrared remote sensing images in the embodiment can be used for the visible light band and the thermal infrared band with large color distance, and the spectral information between different ground objects is used to perform spectral mapping of the latter to the former and to realize registration.
[0031] Figure 2 A framework diagram of the registration method of visible light and thermal infrared remote sensing images according to the embodiment of the present disclosure is schematically shown.
[0032] As Figure 2 shown, in the embodiment, the initial visible light band image of the target area can be first acquired, and the ground object classification is performed on the initial visible light band image to obtain a ground object classification result. Then, the reference band is determined according to the ground object classification result, and the initial visible light band image corresponding to the reference band (i.e., the reference band classification map in Figure 2 ) is taken as the visible light band image.
[0033] Among them, the ground object classification represents the gray value difference between different types of ground objects under different bands, and the meaning of determining the reference band according to the ground object classification result is that the most suitable band in the current visible light band is selected as the reference band according to the ground object classification result, and the most suitable band may, for example, be the band with the largest gray value difference between different types of ground objects.
[0034] After the selection of the visible light band image is completed, the visible light band image of the target region is subjected to spectral mapping using a spectral mapping model, and a plurality of mapped thermal infrared band classification maps can be obtained. The plurality of mapped thermal infrared band classification maps are combined to obtain a mapped thermal infrared band image.
[0035] The spectral mapping model is obtained by the following operations: first, obtaining an initial spectral mapping model and visible light spectrum information and thermal infrared band spectrum information of N ground objects (N is a positive integer greater than 0). Second, the visible light spectrum information and the thermal infrared band spectrum information of each ground object in the N ground objects are input into the initial spectral mapping model as prior knowledge to analyze the mapping relationship between the visible light spectrum information and the thermal infrared band spectrum information of each ground object. Finally, the internal parameters of the initial spectral mapping model are adjusted according to the plurality of mapping relationships to obtain the spectral mapping model.
[0036] The registration method of the visible light and thermal infrared remote sensing images in the embodiment can accurately complete the construction of the spectral mapping model. By using the spectrum information of different ground objects in the visible light and thermal infrared bands as prior knowledge to construct the spectral mapping model, the spectral characteristics of different types of ground objects in the visible light band and the thermal infrared band are fully utilized, so that the constructed mapping model has accuracy and reliability, and can complete the effective conversion of the visible light reference band spectrum feature to the thermal infrared band.
[0037] After obtaining the mapped thermal infrared band image, first, the thermal infrared band image of the target region can be obtained, and the thermal infrared band image is up-sampled to make the spatial resolutions of the thermal infrared band image and the mapped thermal infrared band image consistent. Then, the thermal infrared band image and the mapped thermal infrared band image can be coarsely registered using a simple phase correlation method to obtain a coarsely registered thermal infrared band image. The registration between the visible light and the thermal infrared band can be completed.
[0038] The registration method of the visible light and thermal infrared remote sensing images in the embodiment uses the spectral mapping model constructed in advance to map the thermal infrared band with a large color difference from the visible light color to a mapped thermal infrared band with a similar color and perform registration, without needing to bear the influence of high color difference. The spectral characteristics of the images to be registered are guaranteed to be consistent, and the registration effect is good.
[0039] After obtaining the coarsely registered thermal infrared band image, first, the feature matching between the mapped thermal infrared band image and the coarsely registered thermal infrared band image can be performed using a feature method to determine a feature pair set of the target region (i.e. Figure 2The coarse registration thermal infrared band image can be divided into a plurality of sub-images by the multi-scale center gravity adaptive block module. The plurality of sub-images and the mapped thermal infrared band image are matched in features to determine a plurality of sub-regional sub-feature pair sets.
[0040] The coarse registration thermal infrared band image can be divided according to the distance between each of the plurality of sub-images and the center point. The size of the sub-image is inversely proportional to the distance between the sub-image and the center point.
[0041] The registration method of the visible light and thermal infrared remote sensing images in the embodiment uses the multi-scale center gravity adaptive block module to non-uniformly divide the images, thereby improving the registration effect of the edge region of the image. In the multi-scale center gravity adaptive block module, the size of the block is determined by the distance from the center of the image. The greater the distance, the smaller the size of the block. This means that the number and uniformity of the inner points in the edge region of the image have a finer granularity, so that the inner points in the edge region of the image have more and more uniform distribution, thereby providing a guarantee for the effective registration of the edge region of the image.
[0042] After the coarse registration thermal infrared band image is divided, the sub-feature pair set can be sequentially subjected to quantity judgment, and uniformity judgment between the plurality of sub-images. The uniformity between the plurality of sub-images is determined by the distance between a plurality of average center points of the plurality of sub-images. The average center point is determined by a plurality of feature points in the sub-image. The uniformity judgment can use the k-nearest neighbor method to calculate a k-nearest neighbor distance set, and compare it with a pre-set threshold value. When the quantity or uniformity does not reach the set threshold value, pseudo-feature pairs are randomly selected on the coarse registration thermal infrared band image and the mapped thermal infrared band image, and the pseudo-feature points are matched by using the local module registration. The pseudo-feature pairs formed by the matched pseudo-feature points are added to the previous feature pair set to obtain a new set.
[0043] The registration method of the visible light and thermal infrared remote sensing images in the embodiment supplements the inner points selected by the feature method with pseudo-feature points, thereby eliminating the distortion of the transformation parameters caused by using only the former. When the inner points selected by the feature method face the problems of insufficient number and uneven distribution, a sufficient number of pseudo-feature points with uniform distribution are first randomly selected, and then the deviation is solved by using the high-precision local module registration, thereby supplementing the original inner point set. Thus, the new set has good accuracy, and does not appear to be biased due to insufficient number and uneven distribution in the global and block scales, thereby improving the accuracy and full consistency of the transformation matrix fitting. Figure 1
[0044] In addition, the registration method for visible light and thermal infrared remote sensing images in the embodiment can realize high-precision registration between images and has strong anti-noise capability. The registration method for visible light and thermal infrared remote sensing images in the embodiment introduces the local module registration with strong anti-noise capability into the process, improves the anti-noise capability of the cross-correlation spectrum, and avoids the influence of excessive noise on the original image on the quality of the cross-correlation spectrum. Therefore, the least square deviation estimation step using the pseudo feature point pair can obtain higher-precision sub-pixel level solution, and the registration effect is improved.
[0045] Finally, when the number of feature pairs in the new set and the distribution uniformity between the plurality of sub-images both meet the requirements, the transformation matrix is fitted by using the deviations of the feature pairs in the new set, and the transformation matrix is obtained. Then, the visible light band image and the thermal infrared band image are registered according to the target transformation matrix.
[0046] The registration method for visible light and thermal infrared remote sensing images in the embodiment can realize efficient registration between large-size images. By using the simple phase correlation method to coarsely register the large-size images first, and then using the accurate inner point set to fit the transformation matrix, the large-size images do not need to be directly subjected to complex transformation and mathematical processing, the time burden caused by the high complexity step is avoided, and the problem of poor adaptability to complex transformation caused by using only the phase correlation method is also avoided. Through the reasonable pre-registration and the inner point transformation fitting step, the high inner point deviation calculation precision and the high calculation efficiency can be achieved.
[0047] It should be understood that the specific order or hierarchy of steps in the disclosed processes is an example of an example order or hierarchy of steps. Based on design preference, it should be understood that the specific order or hierarchy of steps in the processes can be rearranged without departing from the scope of the disclosure. The appended method claims recite the elements of the various steps in an example order, and are not intended to be limited to a particular order or hierarchy.
[0048] It should also be noted that the directional terms mentioned in the embodiments, such as "up", "down", "front", "back", "left", "right", etc., are only the directions of the drawings, and are not intended to limit the scope of the disclosure. Throughout the drawings, the same elements are represented by the same or similar reference numerals. When the understanding of the disclosure may be confused, the conventional structure or configuration will be omitted. And the shape, size, positional relationship of the components in the drawings do not reflect the true size, proportion and actual positional relationship.
[0049] In the detailed description above, various features are grouped together in single embodiments for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting a necessity to more features than are expressly recited in each claim. Thus, the claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment. It is the intention of the inventor to encompass all such variations and / or permutations of the disclosure within the purview of the appended claims, which follow.
[0050] Furthermore, the terms "first", "second", and the like, do not denote any quantity or order but are used as labels, names for referring to the features. The terms "comprises", "comprising", "includes", "including" and "having" are used in the detailed description and in the claims of this disclosure as open-ended terms. The terms "comprises", "comprising", "includes", "including" and "having" are used in the detailed description and in the claims of this disclosure as open-ended terms. The term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items, as well as the lack of combinations when interpreted in the alternative ("or").
[0051] The specific embodiments described above are examples of the disclosure and do not limit the disclosure. The scope of the disclosure is limited only by the claims.
Claims
1. A method for registering visible and thermal infrared remote sensing images, characterized in that, The method comprises the following steps: mapping the visible light band image of the target area by using a spectral mapping model to obtain a mapped thermal infrared band image, wherein the spectral mapping model is constructed by using the visible light spectral information and the thermal infrared band spectral information of multiple ground objects in the target area; coarsely registering the thermal infrared band image of the target area with the mapped thermal infrared band image to obtain a coarsely registered thermal infrared band image; performing feature matching on the mapped thermal infrared band image and the coarsely registered thermal infrared band image to determine a feature pair set of the target area, wherein the feature pair set comprises multiple matched feature pairs, and each feature pair represents a pair of matched feature points; registering the visible light band image and the thermal infrared band image according to the feature pair set.
2. The registration method of claim 1, wherein, Before the step of mapping the visible light band image of the target area by using a spectral mapping model to obtain a mapped thermal infrared band image, the method further comprises the following steps: obtaining an initial visible light band image of the target area, performing ground object classification on the initial visible light band image to obtain a ground object classification result, wherein the ground object classification result represents the gray value difference between different types of ground objects in different bands; determining a reference band according to the ground object classification result, and taking the initial visible light band image corresponding to the reference band as the visible light band image.
3. The registration method of claim 1, wherein, The construction process of the spectral mapping model comprises the following steps: obtaining an initial spectral mapping model and the visible light spectral information and the thermal infrared band spectral information of the multiple ground objects; inputting the visible light spectral information and the thermal infrared band spectral information of each ground object in the multiple ground objects into the initial spectral mapping model to analyze the mapping relationship between the visible light spectral information and the thermal infrared band spectral information of each ground object; adjusting the internal parameters of the initial spectral mapping model according to multiple mapping relationships to obtain the spectral mapping model.
4. The registration method of claim 1, wherein, Before the step of coarsely registering the thermal infrared band image of the target area with the mapped thermal infrared band image, the method further comprises the following step: upsampling the thermal infrared band image to make the spatial resolutions of the thermal infrared band image and the mapped thermal infrared band image consistent.
5. The registration method of claim 1, wherein, After the step of performing feature matching on the mapped thermal infrared band image and the coarsely registered thermal infrared band image to determine a feature pair set of the target area, the method further comprises the following steps: in a case where the number of feature pairs in the feature pair set is less than a first preset threshold or the uniformity of the feature pairs is less than a second preset threshold, constructing a first pseudo feature pair from two points in the same image coordinate respectively from the mapped thermal infrared band image and the coarsely registered thermal infrared band image; constructing multiple first pseudo feature pairs and performing local module registration on the first pseudo feature pairs to obtain registered first pseudo feature pairs; adding at least one registered first pseudo feature pair to the feature pair set until the total number of the registered first pseudo feature pairs and the feature pairs in the feature pair set is greater than or equal to the first preset threshold, and the total uniformity of the registered first pseudo feature pairs and the feature pairs is greater than or equal to the second preset threshold.
6. The registration method of claim 1, wherein, After the feature pair set of the target region is determined by performing feature matching on the mapped thermal infrared band image and the coarsely registered thermal infrared band image, the method further includes: performing block division on the coarsely registered thermal infrared band image to divide the coarsely registered thermal infrared band image into a plurality of sub-images, the plurality of sub-images respectively corresponding to a plurality of sub-regions in the target region; allocating the feature pair set to the plurality of sub-images to determine a plurality of sub-feature pair sets of the plurality of sub-regions.
7. The method of registration of claim 6, wherein, After the feature pair set of the target region is determined by performing feature matching on the mapped thermal infrared band image and the coarsely registered thermal infrared band image, the method further includes: in a case where a number of any one of the plurality of sub-feature pair sets is less than a third preset threshold or uniformity between the plurality of sub-images is less than a fourth preset threshold, constructing a second pseudo feature pair from two points at a same image coordinate in a same image respectively corresponding to the mapped thermal infrared band image and the sub-feature pair set; constructing a plurality of the second pseudo feature pairs and performing local module registration on the second pseudo feature pairs to obtain registered second pseudo feature pairs; adding the registered second pseudo feature pairs to the sub-feature pair set until a number of the registered second pseudo feature pairs and the sub-feature pairs in the sub-feature pair set is greater than or equal to the third preset threshold and the uniformity between the plurality of sub-images is greater than or equal to the fourth preset threshold; wherein the uniformity between the plurality of sub-images is determined by a distance between a plurality of average center points of the plurality of sub-images, and the average center points are determined by a plurality of feature points in the sub-image.
8. The method of registration of claim 6, wherein, The block division on the coarsely registered thermal infrared band image includes: determining a center point of the coarsely registered thermal infrared band image; performing block division on the coarsely registered thermal infrared band image according to a distance between each sub-image of the plurality of sub-images and the center point, the size of the sub-image being inversely proportional to the distance between the sub-image and the center point.
9. The registration method of claim 1, wherein, The registration of the visible light band image and the thermal infrared band image according to the feature pair set includes: performing transformation matrix fitting on a deviation of each feature pair in the feature pair set to obtain a target transformation matrix; performing registration on the visible light band image and the thermal infrared band image according to the target transformation matrix.
10. The registration method of claim 1, wherein, The coarsely registering the thermal infrared band image of the target region with the mapped thermal infrared band image includes: performing coarsely registration on the thermal infrared band image of the target region and the mapped thermal infrared band image by using a phase correlation method.