Image alignment method and device, equipment and storage medium
By integrating the SIFT algorithm into the NCC algorithm, the problem of sensitivity to scale and rotation changes during image alignment is solved, improving the success rate and stability of image alignment and achieving more efficient image alignment results.
Patent Information
- Application Number
- CN202410777389.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-14
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-06-14
AI Technical Summary
Existing NCC algorithms are sensitive to scale and rotation changes in images during the image alignment process, leading to alignment failures and making it difficult to effectively align electron microscope images.
By integrating the SIFT algorithm into the NCC algorithm, the NCC algorithm process is improved by determining SIFT feature point pairs and projecting them onto the image, thus reducing the probability of image alignment failure.
It improves the success rate and stability of image alignment, reduces the probability of alignment failure caused by scale and rotation changes, and enhances the cross-correlation of key points between images.
Smart Images

Figure CN118967755A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of image processing technology, and in particular to an image alignment method, device, equipment and storage medium. Background Art
[0002] After scanning the sample layer by layer using an electron microscope, a series of electron microscope images will be obtained, and these electron microscope images need to be aligned layer by layer.
[0003] The Normalized Cross-Correlation (NCC) algorithm is a common similarity measurement method in image processing. It can be used to calculate the similarity between two images to complete tasks such as template matching and target detection. It is used to describe the correlation between two vectors, windows or samples of the same dimension. The calculation of the NCC algorithm is relatively simple and can be implemented using the Fast Fourier Transform (FFT). It has the characteristics of fast calculation speed, easy implementation and simple and intuitive results. Therefore, it is widely used in image alignment work.
[0004] When the inventors were conducting an analysis and study on the electron microscope image stitching algorithm, they found that the NCC algorithm could not identify the similar features of the two images in some cases, resulting in unsuccessful alignment. For example, during electron microscope image acquisition, since it is necessary to scan the sample layer by layer, there may be slight movement or tilt of the sample, or slight mechanical vibration or displacement of the acquisition equipment, or the conductivity of the sample, etc., which may cause scale changes and rotational offsets in the electron microscope image during the layer-by-layer scanning process. When this happens, the NCC method is more sensitive to scale and rotation changes of the image, which makes it easy for image alignment to fail. Summary of the invention
[0005] The embodiments of the present invention provide an image alignment method, apparatus, device and storage medium, aiming to improve the success rate of image alignment.
[0006] In a first aspect, an embodiment of the present invention provides an image alignment method, comprising:
[0007] When the NCC mutual correlation coefficient between the source image and the image to be aligned meets the first preset condition, determining a scale-invariant feature transform (SIFT) feature point pair between the source image and the image to be aligned;
[0008] Based on the SIFT feature point pairs, according to the spatial transformation relationship between the source image and the image to be aligned, performing image projection on the image to be aligned to obtain a projected image;
[0009] When the NCC correlation coefficient between the source image and the projection image meets a second preset condition, the projection image is aligned with the source image based on an NCC algorithm.
[0010] In a second aspect, an embodiment of the present invention provides an image alignment device, comprising:
[0011] A feature point pair determination module, configured to determine a SIFT feature point pair between the source image and the image to be aligned when the NCC mutual correlation coefficient between the source image and the image to be aligned meets a first preset condition;
[0012] An image projection module, configured to project the image to be aligned based on the SIFT feature point pair and according to a spatial transformation relationship between the source image and the image to be aligned, to obtain a projected image;
[0013] The image alignment module is used to align the projection image with the source image based on the NCC algorithm when the NCC mutual correlation coefficient between the source image and the projection image meets the second preset condition.
[0014] In a third aspect, an embodiment of the present invention provides an electronic device, including:
[0015] one or more processors;
[0016] A memory for storing one or more programs;
[0017] When the one or more programs are executed by the one or more processors, the one or more processors implement the image alignment method provided by any embodiment of the present invention.
[0018] In a fourth aspect, an embodiment of the present invention provides a storage medium comprising computer executable instructions, which, when executed by a computer processor, are used to perform an image alignment method as provided in any embodiment of the present invention.
[0019] An image alignment method, apparatus, device and storage medium provided by an embodiment of the present invention integrate the SIFT algorithm into the NCC algorithm to improve the NCC algorithm process, thereby solving the problem of image alignment failure caused by the NCC algorithm being sensitive to scale changes and rotation changes of the image, improving the mutual correlation of key points between images, reducing the probability of image alignment failure caused by scale changes and rotation changes of the image, and improving the stability of the image alignment algorithm. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] Figure 1 A flowchart of an image alignment method provided in Embodiment 1 of the present invention;
[0021] Figure 2 A schematic diagram of the structure of an image alignment device provided in Embodiment 2 of the present invention;
[0022] Figure 3 A schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention;
[0023] Figure 4 An example of a two-layer electron microscope image of a sample in an embodiment of the present invention;
[0024] Figure 5a for Figure 4 Normalized cross-correlation of the two-layer electron microscope images;
[0025] Figure 5b for Figure 4 Normalized cross-correlation of the two layers of electron microscope images after NIFT processing;
[0026] Figure 6 for Figure 4 Schematic diagram of the feature points extracted by NIFT after the two layers of electron microscope images are randomly rotated;
[0027] Figure 7a for Figure 6 Normalized cross-correlation of two layers of EM images after random rotation;
[0028] Figure 7b for Figure 6 Normalized cross-correlation of two layers of electron microscope images after random rotation and NIFT processing;
[0029] Figure 8 for Figure 4 Schematic diagram of the feature points extracted by NIFT after random cropping of the image to be stitched;
[0030] Figure 9a for Figure 8 Normalized cross-correlation of two layers of electron microscope images after random cropping of the image to be stitched;
[0031] Figure 9b for Figure 8 Normalized cross-correlation of two layers of electron microscope images after random cropping and NIFT processing of the image to be stitched. DETAILED DESCRIPTION
[0032] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.
[0033] Embodiment 1
[0034] Figure 1 This is a flow chart of an image alignment method provided in Embodiment 1 of the present invention. This embodiment is applicable to the case where an electron microscope performs electron microscope image alignment and splicing after scanning a sample layer by layer. The method can be performed by an image alignment device, which can be implemented by hardware and / or software and can generally be integrated in an electronic device, such as a computer device. The method specifically includes:
[0035] Step 110 : When the NCC mutual correlation coefficient between the source image and the image to be aligned meets a first preset condition, determine a SIFT feature point pair between the source image and the image to be aligned.
[0036] Among them, scanning the sample layer by layer will successively obtain an image sequence consisting of multiple electron microscope images, and the splicing between electron microscope images requires image alignment, usually aligning adjacent electron microscope images. At the beginning of image alignment, the source image can be the first electron microscope image in the image sequence, of course, it can also be one of the electron microscope images selected manually, and the image to be aligned is the electron microscope image adjacent to the source image, generally, it can be the adjacent electron microscope image after the source image. If the alignment of some electron microscope images has been completed in the early stage, then the currently aligned image can be selected as the source image. The mutual correlation coefficient between the source image and the image to be aligned can be obtained by using the NCC algorithm, thereby reflecting the similarity between the source image and the image to be aligned. The purpose of this step is to determine whether the two images can find a region with greater similarity without any processing. If the correlation meets the conditions for alignment using the NCC algorithm, then the points found under the NCC algorithm can be directly used for image alignment. It is no longer necessary to extract other feature points, which reduces time complexity and improves efficiency. The above-mentioned NCC mutual correlation coefficient meets the first preset condition, which means that the correlation between the current source image and the image to be aligned cannot meet the conditions for alignment using the NCC algorithm. At this time, it is necessary to use the SIFT algorithm to extract other feature points to improve the NCC algorithm and complete the image alignment.
[0037] In an optional implementation, when the NCC mutual correlation coefficient between the source image and the image to be aligned meets the first preset condition, the SIFT feature point pair between the source image and the image to be aligned is determined, including: calculating the NCC mutual correlation coefficient between the source image and the image to be aligned; selecting the NCC mutual correlation coefficient with the highest numerical order and comparing it with the preset threshold; if the NCC mutual correlation coefficient with the highest numerical order is less than the preset threshold, calculating the SIFT feature points between the source image and the image to be aligned, and determining the matching SIFT feature point pair. Among them, the peak value of the NCC mutual correlation coefficient can be selected to compare with the preset threshold to determine whether its peak value is less than the set threshold. If it is less than the preset threshold, it means that the image difference is large, and there may be factors such as scale change or rotation change in the image that affect the peak value of NCC, then it is necessary to extract SIFT feature points to determine the matching SIFT feature point pair between the source image and the image to be aligned. For example, the preset threshold is 0.3. The SIFT algorithm is invariant to changes in images at different scales. Even if the image scale changes, SIFT can still stably detect the feature point pairs between images. The SIFT algorithm also has rotation invariance, which makes SIFT perform well when encountering image rotation. However, due to the high complexity of the SIFT algorithm, the efficiency is low when aligning a large number of images.
[0038] Step 120 : Based on the SIFT feature point pairs and according to the spatial transformation relationship between the source image and the image to be aligned, the image to be aligned is projected to obtain a projected image.
[0039] Among them, SIFT feature points can be used to estimate the geometric transformation between images, and the image to be aligned is projected to obtain a projected image. This step aims to obtain a new image that has almost no scale and rotation changes compared to the source image, so that NCC calculation can be performed again in the future to ensure the quality of alignment and reduce the probability of incorrect alignment.
[0040] Step 130: When the NCC correlation coefficient between the source image and the projection image meets a second preset condition, align the projection image with the source image based on an NCC algorithm.
[0041] The NCC correlation coefficient is calculated between the source image and the newly obtained projection image. This step is to reconfirm the similarity between the source image and the projection image, further ensure the relationship between the source image and the projection image, and reduce the probability of incorrect alignment. If the second preset condition is determined to be met based on the NCC correlation coefficient between the source image and the projection image, the two images can be aligned using the NCC algorithm.
[0042] The technical solution of this embodiment integrates the SIFT algorithm into the NCC algorithm, improves the NCC algorithm process, solves the problem of image alignment failure caused by the sensitivity of the NCC algorithm to scale changes and rotation changes of images, realizes the improvement of the cross-correlation of key points between images, reduces the probability of alignment failure caused by image scale changes and rotation changes, and improves the stability of the image alignment algorithm.
[0043] In an alternative implementation, before determining the SIFT feature point pairs between the source image and the image to be aligned when the NCC cross-correlation coefficient between the source image and the image to be aligned meets the preset conditions, it further includes:
[0044] Obtain a series of electron microscope images obtained by scanning the sample layer by layer with an electron microscope;
[0045] Use the previous electron microscope image as the source image and the adjacent subsequent electron microscope image as the image to be aligned.
[0046] For example, for a sample, after layer-by-layer scanning, n layers of electron microscope images are obtained. Layer-by-layer alignment is required. If m images (1 <= m < n) have been aligned previously, then the m-th image is the source image and the (m + 1)-th image is the image to be aligned, and so on. It can be understood that both n and m are integers greater than 1.
[0047] In an alternative implementation, based on the SIFT feature point pairs, according to the spatial transformation relationship between the source image and the image to be aligned, project the image to be aligned to obtain a projected image, including:
[0048] According to the SIFT feature point pairs, use the Random Sample Consensus (RANSAC) algorithm to determine the geometric transformation model between the source image and the image to be aligned; among them, common models include translation, affine transformation, etc.
[0049] Use the geometric transformation model to project the image to be aligned to obtain the corresponding projected image.
[0050] In an alternative implementation, when the NCC cross-correlation coefficient between the source image and the projected image meets the second preset condition, align the projected image with the source image based on the NCC algorithm, including:
[0051] Calculate the NCC cross-correlation coefficient between the source image and the projected image;
[0052] Select the NCC cross-correlation coefficients with higher numerical rankings and compare them with the preset threshold;
[0053] If the NCC cross-correlation coefficients with higher numerical rankings are greater than or equal to the preset threshold, align the projected image with the source image based on the NCC algorithm.
[0054] The peak value of the NCC correlation coefficient may be selected to be compared with a preset threshold value. If the peak value is greater than or equal to the preset threshold value, it indicates that the projection image and the source image may be aligned based on the NCC algorithm.
[0055] In an optional implementation, the image alignment method further includes:
[0056] If the NCC correlation coefficient with the highest value between the source image and the projected image is less than the preset threshold, the alignment failure information is returned. The peak value of the NCC correlation coefficient between the source image and the projected image can be selected and compared with the preset threshold. If the peak value of the NCC correlation coefficient between the source image and the projected image is less than the preset threshold, it means that the source image and the projected image cannot be successfully aligned.
[0057] The inventors verified the effect of the image alignment method in the embodiment of the present invention through experiments, including the comparative experimental quantification results of NCC+SIFT, SIFT, and NCC, the case where NCC / SIFT alignment failed but NCC+SIFT (NIFT) successfully aligned, rotation invariance verification, and scale invariance verification.
[0058] First, a comparative experiment of NIFT, SIFT and NCC was conducted to quantify the overall advantages and disadvantages of the three methods by comparing the success rate and time consumption of stitching images. The results are shown in Table 1. It can be seen that NIFT combines the advantages of SIFT and NCC. While maintaining the high success rate of SIFT, the average time consumption is only 40% of SIFT, and the success rate is 9.1% higher than NCC. It can be seen that the proposed NIFT method is superior.
[0059] Table 1
[0060] method Success rate time consuming NIFT 97.6% 194.07s SIFT 97.6% 490.22s NCC 88.5% 160.46s
[0061] In order to verify that NIFT can align cases where NCC alignment fails, the experimental data is tested on 166 groups of images, with two images in each group. Sample data of two adjacent layers where only NCC alignment fails are selected, and NCC and NIFT are calculated respectively. Figure 4 are the two layers of samples where NCC alignment fails, and the normalized cross-correlation between them is as follows: Figure 5a As shown, the maximum value is 0.0955, which is lower than the set threshold of 0.3. Figure 5b This is an image aligned by NIFT. The maximum NCC value is 0.4607, which is greater than the threshold value of 0.3, and the peak is prominent, which shows the optimization effect of NIFT.
[0062] In order to verify that NIFT is rotation invariant, the stitched images are randomly rotated. Figure 6 After random rotation, NIFT can effectively extract feature points. After the image is randomly rotated, NCC decreases significantly, such as Figure 7a After NIFT processing, a higher NCC can still be obtained, as shown in Figure 7b As shown in Figure 2, NIFT inherits the advantages of SIFT and has rotation invariance.
[0063] In order to verify that NIFT is scale invariant, the stitched images are randomly cropped. Figure 8 After random cropping, NIFT can effectively extract feature points. The NCC of the image after random cropping is as follows Figure 9a As shown in Figure 2, the peak is low and there are interfering peaks. Figure 9b As shown in Figure 2, the NCC peak after NIFT processing is high and unique. Therefore, NIFT is scale invariant.
[0064] Experiments show that this method solves the problems of the original NCC algorithm, has scale invariance and rotation invariance, and is robust to scale and rotation transformations of images.
[0065] Embodiment 2
[0066] Figure 2 A schematic diagram of the structure of an image alignment device provided in Embodiment 2 of the present invention is shown in FIG. Figure 2 As shown, the image alignment device includes: a feature point pair determination module 210, an image projection module 220 and an image alignment module 230, wherein:
[0067] A feature point pair determination module 210, configured to determine a SIFT feature point pair between the source image and the image to be aligned when the NCC mutual correlation coefficient between the source image and the image to be aligned meets a first preset condition;
[0068] An image projection module 220 is used to project the image to be aligned based on the SIFT feature point pairs and according to the spatial transformation relationship between the source image and the image to be aligned to obtain a projected image;
[0069] The image alignment module 230 is configured to align the projection image with the source image based on an NCC algorithm when the NCC correlation coefficient between the source image and the projection image meets a second preset condition.
[0070] Optionally, the feature point pair determination module 210 includes:
[0071] A first coefficient calculation unit, used for calculating the NCC mutual correlation coefficient between the source image and the image to be aligned;
[0072] A first coefficient comparison unit is used to select a NCC mutual correlation coefficient with a top numerical value and compare it with a preset threshold;
[0073] The feature point calculation unit is used to calculate the SIFT feature points between the source image and the image to be aligned, and determine the matching SIFT feature point pairs if the NCC mutual correlation coefficient with the top numerical ranking is less than a preset threshold.
[0074] Optionally, the image alignment device further includes:
[0075] An image acquisition module, used for acquiring a series of electron microscope images obtained by scanning the sample layer by layer through an electron microscope before determining the SIFT feature point pairs between the source image and the image to be aligned when the NCC mutual correlation coefficient between the source image and the image to be aligned meets a preset condition;
[0076] The image determination module is used to use the preceding electron microscope image as the source image and the adjacent succeeding electron microscope image as the image to be aligned.
[0077] Optionally, the image projection module 220 includes:
[0078] A model determination unit, used for determining a geometric transformation model between a source image and an image to be aligned using a random sampling consensus RANSAC algorithm according to a pair of SIFT feature points;
[0079] The projection unit is used to project the image to be aligned using a geometric transformation model to obtain a corresponding projection image.
[0080] Optionally, the image alignment module 230 includes:
[0081] A second coefficient calculation unit, used for calculating the NCC mutual correlation coefficient between the source image and the projection image;
[0082] The second coefficient comparison unit is used to select the NCC mutual correlation coefficient with the highest value ranking and compare it with the preset threshold;
[0083] The second feature point calculation unit is configured to align the projection image with the source image based on the NCC algorithm if the NCC correlation coefficient with the highest numerical value is greater than or equal to a preset threshold.
[0084] Optionally, the image alignment device further includes:
[0085] The failure information returning module is used to return the alignment failure information if the NCC mutual correlation coefficient with the highest numerical value between the source image and the projection image is less than a preset threshold.
[0086] The image alignment device provided in the embodiment of the present invention can execute the image alignment method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.
[0087] Embodiment 3
[0088] Figure 3 A schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention is shown in FIG. Figure 3 As shown, the electronic device includes a processor 310, a memory 320, an input device 330 and an output device 340; the number of the processor 310 in the electronic device can be one or more. Figure 3 A processor 310 is taken as an example; the processor 310, the memory 320, the input device 330 and the output device 340 in the electronic device can be connected via a bus or other means. Figure 3 The example of connecting through bus is taken in the following.
[0089] The memory 320, as a computer-readable storage medium, can be used to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the image alignment method in the embodiment of the present invention (for example, the feature point pair determination module 210, the image projection module 220 and the image alignment module 230 in the image alignment device). The processor 310 executes various functional applications and data processing of the electronic device by running the software programs, instructions and modules stored in the memory 320, that is, implements the above-mentioned image alignment method.
[0090] The memory 320 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and at least one application required for a function; the data storage area may store data created according to the use of the terminal, etc. In addition, the memory 320 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 320 may further include a memory remotely arranged relative to the processor 310, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0091] The input device 330 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the electronic device. The output device 340 may include a display device such as a display screen.
[0092] Embodiment 4
[0093] Embodiment 4 of the present invention further provides a storage medium containing computer executable instructions, wherein the computer executable instructions are used to perform an image alignment method when executed by a computer processor, including:
[0094] When the NCC mutual correlation coefficient between the source image and the image to be aligned meets the first preset condition, determining a SIFT feature point pair between the source image and the image to be aligned;
[0095] Based on the SIFT feature point pairs, according to the spatial transformation relationship between the source image and the image to be aligned, performing image projection on the image to be aligned to obtain a projected image;
[0096] When the NCC correlation coefficient between the source image and the projection image meets a second preset condition, the projection image is aligned with the source image based on an NCC algorithm.
[0097] Of course, the computer executable instructions of a storage medium including computer executable instructions provided by an embodiment of the present invention are not limited to the operations of the method described above, and can also execute related operations in the image alignment method provided by any embodiment of the present invention.
[0098] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.
[0099] It is worth noting that in the embodiment of the above-mentioned image alignment device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.
[0100] Although the present invention has been described in detail above by means of general description, specific implementation methods and tests, it is obvious to those skilled in the art that some modifications or improvements may be made to the present invention. Therefore, these modifications or improvements made without departing from the spirit of the present invention all fall within the scope of protection claimed by the present invention.
Claims
1. An image alignment method, characterized in that: include: When the NCC mutual correlation coefficient between the source image and the image to be aligned meets the first preset condition, determining a SIFT feature point pair between the source image and the image to be aligned; Based on the SIFT feature point pairs, according to the spatial transformation relationship between the source image and the image to be aligned, performing image projection on the image to be aligned to obtain a projected image; When the NCC correlation coefficient between the source image and the projection image meets a second preset condition, the projection image is aligned with the source image based on an NCC algorithm.
2. The method according to claim 1, characterized in that The determining of the SIFT feature point pairs between the source image and the image to be aligned when the NCC mutual correlation coefficient between the source image and the image to be aligned meets the first preset condition comprises: Calculating the NCC mutual correlation coefficient between the source image and the image to be aligned; Select the NCC correlation coefficient with the highest value to compare with the preset threshold; If the NCC mutual correlation coefficient with the highest numerical ranking is less than the preset threshold, the SIFT feature points between the source image and the image to be aligned are calculated to determine a matching pair of SIFT feature points.
3. The method according to claim 2, characterized in that In the case that the NCC mutual correlation coefficient between the source image and the image to be aligned meets the first preset condition, before determining the SIFT feature point pair between the source image and the image to be aligned, the method further includes: Obtain a series of electron microscope images by scanning the sample layer by layer through an electron microscope; The preceding electron microscope image is used as the source image, and the adjacent subsequent electron microscope image is used as the image to be aligned.
4. The method according to claim 1, characterized in that: The step of projecting the image to be aligned based on the SIFT feature point pair and according to the spatial transformation relationship between the source image and the image to be aligned to obtain a projected image includes: Determine a geometric transformation model between the source image and the image to be aligned using a random sampling consensus RANSAC algorithm according to the SIFT feature point pairs; The geometric transformation model is used to perform image projection on the image to be aligned to obtain the corresponding projection image.
5. The method according to claim 1, characterized in that When the NCC correlation coefficient between the source image and the projection image meets a second preset condition, aligning the projection image with the source image based on an NCC algorithm includes: Calculating a NCC mutual correlation coefficient between the source image and the projection image; Select the NCC correlation coefficient with the highest value to compare with the preset threshold; If the NCC correlation coefficient with the highest numerical value is greater than or equal to the preset threshold, the projection image is aligned with the source image based on the NCC algorithm.
6. The method according to claim 5, characterized in that Also includes: If the NCC correlation coefficient with the highest numerical value between the source image and the projection image is less than the preset threshold, alignment failure information is returned.
7. An image alignment device, characterized in that: include: A feature point pair determination module, configured to determine a SIFT feature point pair between the source image and the image to be aligned when the NCC mutual correlation coefficient between the source image and the image to be aligned meets a first preset condition; An image projection module, configured to project the image to be aligned based on the SIFT feature point pair and according to a spatial transformation relationship between the source image and the image to be aligned, to obtain a projected image; The image alignment module is used to align the projection image with the source image based on the NCC algorithm when the NCC mutual correlation coefficient between the source image and the projection image meets the second preset condition.
8. The device according to claim 7, characterized in that The feature point pair determination module includes: A first coefficient calculation unit, used to calculate the NCC mutual correlation coefficient between the source image and the image to be aligned; A first coefficient comparison unit is used to select the NCC mutual correlation coefficient with the highest value ranking to compare with the preset threshold; The feature point calculation unit is used to calculate the SIFT feature points between the source image and the image to be aligned, and determine the matching SIFT feature point pairs if the NCC mutual correlation coefficient with the highest numerical ranking is less than the preset threshold.
9. An electronic device, characterized in that: include: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the image alignment method as described in any one of claims 1 to 6.
10. A storage medium containing computer executable instructions, characterized in that: When the computer executable instructions are executed by a computer processor, they are used to perform the image alignment method as described in any one of claims 1-6.
Citation Information
Patent Citations
Image alignment method, device and equipment
CN110059711A
Data processing method and device, equipment and medium
CN116758205A
Method for stitching images of capsule endoscope, electronic device and readable storage medium
US20230123664A1
