An image stitching method, apparatus, storage medium, and electronic device

By acquiring vehicle motion center segmentation and stitching vehicle images, the problem of incomplete vehicle image stitching is solved, achieving the integrity of target vehicle images and improving detection performance.

CN116433479BActive Publication Date: 2026-11-17ZHEJIANG UNIVIEW TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202111648667.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-30
Publication Date
2026-11-17
Estimated Expiration
2041-12-30

AI Technical Summary

Technical Problem

In video surveillance scenarios, existing technologies struggle to effectively stitch vehicle images together, causing large and medium-sized vehicles to easily extend beyond the frame and become unrecognizable by the machine, resulting in missed vehicles and unclear deduplication methods.

Method used

By obtaining the vehicle motion center, the original vehicle image is segmented, and the target vehicle image is stitched based on the segmented local feature images. The vehicle motion center is used to perform feature segmentation and stitching on the candidate vehicle images to improve the integrity of the target vehicle.

Benefits of technology

It improves the integrity and detection effect of target vehicle images, and can more accurately identify information such as vehicle model, face and vehicle interior, solving the problem of incomplete vehicle image stitching in the existing technology.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116433479B_ABST
    Figure CN116433479B_ABST
Patent Text Reader

Abstract

This application discloses an image stitching method, apparatus, storage medium, and electronic device. The method includes: acquiring at least two original vehicle images and the vehicle motion center of each original vehicle image; determining at least two candidate vehicle images from each original vehicle image based on the vehicle motion center; performing feature segmentation on each candidate vehicle image based on the vehicle motion center to obtain local feature images corresponding to each candidate vehicle image; and stitching the local feature images together to obtain a target vehicle image. This technical solution can segment candidate vehicle images using vehicle motion centers, thereby performing target vehicle image stitching based on the segmented local feature images, which can improve the integrity of the target vehicle image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to an image stitching method, apparatus, storage medium and electronic device. Background Technology

[0002] In video surveillance scenarios, intelligent cameras have generally achieved high-definition resolution, and the performance of intelligent analysis chips is constantly improving, enabling them to detect targets and trigger image capture in a very short time. With stable camera images and ample detection capabilities, user demands are also increasing. For a single target, more pixels are beneficial for detecting target details, such as interior objects in a vehicle or people in the back seat. Therefore, in engineering survey scenarios, a single target should fill the entire frame as much as possible to achieve the requirement of maximizing pixel features. However, this condition currently has certain limitations. Due to the uncertainty of target types and the impact of target movement, if all targets need to be included in an engineering survey scenario, the effect on common targets is not ideal.

[0003] Taking intelligent transportation vehicle target detection as an example, if the surveying scenario mainly includes small vehicles, then large and medium-sized vehicles are likely to go out of the frame, and large vehicles may not even be recognized by the machine, resulting in missed vehicles. Currently, the movement trajectory is obtained by a planar sensor, and the images of two consecutive frames captured by the camera are deduplicated and stitched together. The stitched image is then used for intelligent detection and recognition.

[0004] However, this method cannot confirm the integrity of the target in two adjacent frames of the image, and the deduplication method is unclear. Summary of the Invention

[0005] This application provides an image stitching method, apparatus, storage medium, and electronic device, which can segment the original vehicle image through the vehicle motion center, and then stitch the target vehicle image based on the segmented local feature images, thereby improving the integrity of the target vehicle.

[0006] In a first aspect, embodiments of this application provide an image stitching method, the method comprising:

[0007] Acquire at least two original vehicle images, and the vehicle motion center of each original vehicle image;

[0008] Based on the vehicle motion center, at least two candidate vehicle images are determined from each original vehicle image;

[0009] Based on the vehicle motion center, feature segmentation is performed on each candidate vehicle image to obtain a local feature image corresponding to each candidate vehicle image.

[0010] The local feature images are stitched together to obtain the target vehicle image.

[0011] Secondly, embodiments of this application provide an image stitching device, the device comprising:

[0012] The vehicle motion center acquisition module is used to acquire at least two original vehicle images and the vehicle motion center of each original vehicle image.

[0013] The candidate vehicle image determination module is used to determine at least two candidate vehicle images from each original vehicle image based on the vehicle's center of motion.

[0014] The local feature image acquisition module is used to perform feature segmentation on each candidate vehicle image based on the vehicle motion center to obtain a local feature image corresponding to each candidate vehicle image.

[0015] The target vehicle image acquisition module is used to stitch together the local feature images to obtain the target vehicle image.

[0016] Thirdly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the image stitching method as described in embodiments of this application.

[0017] Fourthly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the image stitching method as described in the embodiments of this application.

[0018] The technical solution provided in this application involves acquiring at least two original vehicle images and the vehicle motion center of each original vehicle image. Based on the vehicle motion center, at least two candidate vehicle images are determined from each original vehicle image. Then, based on the vehicle motion center, feature segmentation is performed on each candidate vehicle image to obtain local feature images corresponding to each candidate vehicle image. Finally, the local feature images are stitched together to obtain the target vehicle image. This technical solution can segment candidate vehicle images using the vehicle motion center, thereby improving the integrity of the target vehicle image by stitching together the segmented local feature images. Attached Figure Description

[0019] Figure 1 This is a flowchart of the image stitching method provided in Embodiment 1 of this application;

[0020] Figure 2 This is a schematic diagram of the movement of the splicing area provided in Embodiment 1 of this application;

[0021] Figure 3 This is a schematic diagram of the target vehicle image provided in Embodiment 1 of this application;

[0022] Figure 4 This is a flowchart of the target image information output provided in Embodiment 1 of this application;

[0023] Figure 5 This is a schematic diagram of the image stitching process provided in Embodiment 2 of this application;

[0024] Figure 6 This is a schematic diagram of a local feature image provided in Embodiment 2 of this application;

[0025] Figure 7 This is a schematic diagram of another partial feature image provided in Embodiment 2 of this application;

[0026] Figure 8 This is a schematic diagram of the image stitching device provided in Embodiment 3 of this application;

[0027] Figure 9 This is a schematic diagram of the structure of an electronic device provided in Embodiment 5 of this application. Detailed Implementation

[0028] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the application and not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present application, not the entire structure.

[0029] Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of these steps can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the steps can be rearranged. The process can be terminated when its operation is complete, but may also have additional steps not included in the figures. The process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0030] Example 1

[0031] Figure 1 This is a flowchart of the image stitching method provided in Embodiment 1 of this application. This embodiment is applicable to the segmentation and stitching of original vehicle images. The method can be executed by the image stitching device provided in this embodiment. The device can be implemented by software and / or hardware and can be integrated into devices such as smart terminals used for image processing.

[0032] like Figure 1 As shown, the image stitching method includes:

[0033] S110. Acquire at least two original vehicle images and the vehicle motion center of each original vehicle image;

[0034] In this scheme, the original vehicle image can refer to multiple frames of vehicle images captured by the shooting device when the target vehicle passes through the preset capture position.

[0035] The vehicle motion center can refer to the center of motion of the vehicle in the original vehicle image. The vehicle motion center may differ between different original vehicle images; an original vehicle image may contain one or multiple vehicle motion centers. This can be acquired based on a position sensor. The vehicle motion center can be represented in coordinate form.

[0036] In this technical solution, optionally, the vehicle motion center of each original vehicle image is obtained, including:

[0037] Vehicle clustering motion centers based on original vehicle images acquired from position sensors;

[0038] The coordinate transformation of the vehicle cluster motion center yields the vehicle motion center.

[0039] In this embodiment, the raw data processing of the position sensor is based on the Doppler effect to obtain target motion information, that is, detecting the change in the relative motion frequency between the electromagnetic wave and the observer to confirm the target's position and velocity information. The data collected by the sensor reflects the area of ​​the moving object, and the signal is stronger closer to the center of the target. Typically, the sensor selects the point with the strongest signal as the vehicle clustering motion center.

[0040] In this scheme, the world coordinate system and the video 2D coordinate system can be transformed by using the calibration intrinsic and extrinsic parameters of the road surface, rotation coefficient R, and translation coefficient t. The video coordinates (x, y, t) are obtained with the upper left corner as the origin. c ,y c It should conform to P. c =RP w +t, represented in three-dimensional coordinates as:

[0041]

[0042] Among them, P c Where R is the video coordinate, P is the rotation coefficient, and R is the rotation coefficient. w Let z be the image coordinates, and t be the translation coefficient. The z-coordinate is calculated. c The value is 0, meaning the video coordinates are two-dimensional coordinates.

[0043] Based on the above relationship, the rotation coefficient R and translation coefficient t of the parameter calibration matrix are calculated for the scene and stored. Then, based on one or more vehicle cluster motion centers obtained from the position sensor, the coordinates of the vehicle cluster motion centers are transformed to obtain one or more vehicle motion centers.

[0044] By obtaining the vehicle's center of motion through coordinate transformation, the vehicle image can be segmented based on the center of motion. Then, the target vehicle image can be stitched together based on the segmented local feature images, which can improve the integrity of the target vehicle.

[0045] S120. Based on the vehicle motion center, determine at least two candidate vehicle images from each original vehicle image;

[0046] In this scheme, the original vehicle images can be filtered based on the vehicle motion center, and a portion of the original vehicle images can be selected as candidate vehicle images.

[0047] Specifically, at least two candidate vehicle images can be selected from each original vehicle image by determining the positional relationship between the vehicle's motion center and the effective region. The effective region is determined based on the number of vehicle motion centers and the size of the original vehicle images.

[0048] S130. Based on the vehicle motion center, perform feature segmentation on each candidate vehicle image to obtain a local feature image corresponding to each candidate vehicle image.

[0049] The local feature image can refer to an image composed of a portion of the pixel region in the candidate vehicle image. For example, the local feature image can be an image composed of the pixel region from the vehicle's center of motion to the bottom edge of the candidate vehicle image; or it can be an image composed of the pixel region from the vehicle's center of motion to the top edge of the candidate vehicle image.

[0050] In this scheme, the original vehicle images can be filtered based on the vehicle motion center, candidate vehicle images can be selected from the original vehicle images, and the selected candidate vehicle images can be segmented based on the vehicle motion center to obtain local feature images corresponding to the candidate vehicle images.

[0051] S140. The local feature images are stitched together to obtain the target vehicle image.

[0052] In this embodiment, the vehicle motion center in each local feature image can be moved to a preset image position by linearly moving or scaling the local feature images, and the moved local feature images can be stitched together to obtain a complete target vehicle image.

[0053] Optionally, in this technical solution, the local feature images are stitched together to obtain the target vehicle image, including:

[0054] The local feature image is divided into regions to obtain the stitched region;

[0055] Move the stitching area where the vehicle's motion center is located to the pre-set image center to obtain at least two stitched images after the movement;

[0056] The at least two stitched images are stitched together to obtain the target vehicle image.

[0057] This involves dividing the local feature image into several checkerboard grids to obtain the stitching region. The division scale can be set according to the stitching requirements. For example, the local feature region can be divided into a nine-square grid stitching region.

[0058] In this embodiment, before dividing the local feature image into regions, the local feature image is further processed using conventional deduplication methods. For example, the local feature image is converted to grayscale, the repetition between local feature images is calculated using the binary features of the image, and it is determined whether there are duplicate regions in the local feature image. If so, the high-quality duplicate regions are selected for splicing. The specific deduplication method used is not limited in this embodiment.

[0059] In this solution, the stitching area where the vehicle's motion center is located is calibrated and stitched. Since the main target in the large scene is concentrated in the center of the video image, the stitching area where the vehicle's motion center is located can be moved to the image center based on the image center of the pre-set device, so as to obtain at least two stitched images after the movement.

[0060] For example, Figure 2 This is a schematic diagram illustrating the movement of the splicing area provided in Embodiment 1 of this application. For example... Figure 2 As shown, a1 represents the stitching area where the vehicle's motion center is located, and a2 represents the image center. The image at position a1 can be transformed to position a2 through linear translation or image scaling. Using the reference frame image as the origin, subsequent supplementary frame data are shifted. The calibration parameters should conform to a certain relationship. Taking linear calibration as an example, the output a2 = Z(Zoom) * a1 + O(Offset). Here, Z(Zoom) represents image scaling, and O(Offset) represents linear translation.

[0061] In this embodiment, since the vehicle motion center of each stitched image is in the same position, the stitched images can be stitched together with reference to the original vehicle image to obtain the target vehicle image.

[0062] For example, Figure 3 This is a schematic diagram of the target vehicle image provided in Embodiment 1 of this application, as shown below. Figure 3 As shown, by stitching together the local features of the candidate vehicle images, a complete target vehicle image can be obtained.

[0063] By stitching together local feature images, a target vehicle image with better image quality can be obtained, which is beneficial to improving the detection effect of target image information.

[0064] Optionally, after stitching the local feature images to obtain the target vehicle image, the method further includes:

[0065] The target vehicle image is used as input, and target image information is obtained based on a pre-configured feature detection model; wherein, the target image information includes vehicle model, face, and vehicle interior.

[0066] The feature detection model can be a neural network model, used to identify the content of an image.

[0067] In this embodiment, the target vehicle image is input into the feature detection model, and the features that can be extracted include: vehicle model, face, vehicle interior, etc. Because positional deviations are introduced during the stitching process, some distortion information is generated, leading to incorrect feature output, and the stitched image cannot be directly output. Therefore, the logic layer needs to make a final decision on the structure and select a suitable frame containing features of interest, such as license plate information, to complete the evidence map output. By calibrating the target position in the scene and increasing the training samples for the features, the problems caused by stitching can be largely offset.

[0068] For example, Figure 4 This is a flowchart of the target image information output provided in Embodiment 1 of this application, as follows: Figure 4 As shown, the vehicle motion center is obtained based on the position sensor. Based on the vehicle motion center, the original vehicle images are segmented to obtain at least two local feature images, namely feature information a and feature information b. Then, information a and feature information b are concatenated and deduplicated to obtain the target vehicle image, i.e., the concatenated feature target. Then, the concatenated feature target is used as input, and the feature output is based on a pre-configured feature detection model for user decision-making.

[0069] By clearly defining the target location for detection control, the original vehicle image is modeled, and its features are stitched together. A corresponding network is then trained to recognize image content, thereby improving the accuracy of the image feature algorithm. Through software logic decision-making, more effective key evidence is generated, thus addressing the shortcomings of feature detection.

[0070] The technical solution provided in this application involves acquiring at least two original vehicle images and the vehicle motion center of each original vehicle image. Based on the vehicle motion center, at least two candidate vehicle images are determined from each original vehicle image. Then, based on the vehicle motion center, feature segmentation is performed on each candidate vehicle image to obtain local feature images corresponding to each candidate vehicle image. These local feature images are then stitched together to obtain the target vehicle image. By implementing this technical solution, candidate vehicle images can be segmented using the vehicle motion center, and the target vehicle image can be stitched together based on the segmented local feature images, thereby improving the integrity of the target vehicle image.

[0071] Example 2

[0072] Figure 5 This is a schematic diagram of the image stitching process provided in Embodiment 2 of this application. Embodiment 2 is a further optimization based on Embodiment 1. Specifically, the optimization involves determining at least two candidate vehicle images from each original vehicle image based on the vehicle motion center, including: using the first frame of the original vehicle image where the vehicle motion center is located within a first effective region as the first candidate vehicle image; and using the last frame of the original vehicle image where the vehicle motion center is located within a second effective region as the second candidate vehicle image; wherein the first and second effective regions are determined based on the number of vehicle motion centers and the size of the original vehicle images, respectively. Details not described in this embodiment can be found in Embodiment 1. Figure 5 As shown, the method includes the following steps:

[0073] S510. Acquire at least two original vehicle images and the vehicle motion center of each original vehicle image;

[0074] S520, the first frame of the original vehicle image in which the vehicle's motion center is located within the first effective area is taken as the first candidate vehicle image; and the last frame of the original vehicle image in which the vehicle's motion center is located within the second effective area is taken as the second candidate vehicle image; wherein, the first effective area and the second effective area are determined according to the number of vehicle motion centers and the size of the original vehicle image, respectively.

[0075] In this scheme, when there is only one vehicle motion center, the distance from the top edge of the original vehicle image can be considered. The region within the pixel range is taken as the first effective region; the region is located at the lower edge of the original vehicle image. The region within the pixel area is designated as the second effective region, i.e. The region within the pixel range is considered the second effective region. Here, h is the maximum pixel height of the vehicle, and H is the pixel height of the original vehicle image. With n vehicle motion centers, the region can be defined as the region furthest from the top edge of the original vehicle image. The region within the pixel range is designated as the first effective region; and the region located at the lower edge of the original vehicle image is considered as the first effective region. The area within the pixel range is considered the second effective area.

[0076] In this embodiment, the vehicle motion center is used as the determination criterion. If there is only one vehicle motion center, and it is located within the first effective area, the first frame of the original vehicle image is selected as the first candidate vehicle image. In this case, the lower half of the target vehicle in the first frame of the original vehicle image is relatively complete. If the vehicle motion center leaves the first effective area and enters the second effective area, the last frame of the original vehicle image that enters the second effective area is selected as the second candidate vehicle image. In this case, the upper half of the target vehicle in the last frame of the original vehicle image is relatively complete. The number of frames in the original vehicle image can be determined during the acquisition process of the capturing device.

[0077] In this scheme, when there are n vehicle motion centers, they can be sorted from bottom to top based on the measurement thresholds of the position sensors to obtain the first vehicle motion center, the second vehicle motion center, and the nth vehicle motion center. If the first vehicle motion center is located within the first effective area, the first frame of the original vehicle image is selected as the first candidate vehicle image, in which case the lower half of the target vehicle in the first frame of the original vehicle image is relatively complete. If the nth vehicle motion center enters the second effective area, the last frame of the original vehicle image that enters the second effective area is selected as the second candidate vehicle image, in which case the upper half of the target vehicle in the last frame of the original vehicle image is relatively complete. Here, n is greater than or equal to 2.

[0078] In this technical solution, optionally, determining at least two candidate vehicle images from each original vehicle image based on the vehicle motion center further includes:

[0079] Determine whether the number of vehicle motion centers in the original vehicle image meets the quantity condition;

[0080] If the conditions are met, the original vehicle image will be used as the third candidate vehicle image.

[0081] In this scheme, the quantity condition is greater than or equal to 2. That is, if there is only 1 vehicle motion center in the original vehicle image, the quantity condition is not met; if there are 2 or more vehicle motion centers in the original vehicle image, the quantity condition is met.

[0082] In this embodiment, if the original vehicle image contains multiple vehicle motion centers, then the original vehicle image is used as the third candidate vehicle image.

[0083] By identifying candidate vehicle images, the target vehicle image can be stitched together based on these candidate images, thus improving the integrity of the target vehicle image.

[0084] S530. Based on the vehicle motion center, perform feature segmentation on each candidate vehicle image to obtain a local feature image corresponding to each candidate vehicle image.

[0085] In this embodiment, the candidate vehicle image can be divided into several feature regions based on the location of the vehicle's center of motion, and one of these feature regions can be selected as the local feature image corresponding to the candidate vehicle image.

[0086] In this technical solution, optionally, feature segmentation is performed on each candidate vehicle image based on the vehicle motion center to obtain a local feature image corresponding to each candidate vehicle image, including:

[0087] If the candidate vehicle image is the first candidate vehicle image, then the pixel region containing the vehicle's center of motion and the first edge of the first candidate vehicle image will be used as the local feature image.

[0088] If the candidate vehicle image is the second candidate vehicle image, then the pixel region containing the vehicle's center of motion and the second edge of the second candidate vehicle image will be used as the local feature image.

[0089] The first edge is used to characterize the edge of the first candidate vehicle image. It can be the bottom edge, left edge, or right edge of the first candidate vehicle image. Preferably, the first edge can refer to the bottom edge of the first candidate vehicle image.

[0090] In this embodiment, the second edge is used to characterize the edge of the second candidate vehicle image. It can be the top edge, right edge, and left edge of the second candidate vehicle image. Preferably, the second edge can refer to the top edge of the second candidate vehicle image.

[0091] In this scheme, when there is only one vehicle motion center, the pixel region from the vehicle motion center to the lower edge of the first candidate vehicle image can be used as a local feature image; and the pixel region from the vehicle motion center to the upper edge of the second candidate vehicle image can also be used as a local feature image. Alternatively, the pixel region containing both the vehicle motion center and the lower edge of the first candidate vehicle image can be used as a local feature image; and the pixel region containing both the vehicle motion center and the upper edge of the second candidate vehicle image can also be used as a local feature image. When stitching the local feature images, deduplication processing is required.

[0092] In this embodiment, when there are n vehicle motion centers, the pixel region from the lower edge of the first vehicle motion center to the first candidate vehicle image can be used as a local feature image; and the pixel region from the nth vehicle motion center to the upper edge of the second candidate vehicle image can be used as a local feature image. Alternatively, the pixel region containing the lower edge of the first vehicle motion center and the first candidate vehicle image can be used as a local feature image; and the pixel region containing the upper edge of the nth vehicle motion center and the second candidate vehicle image can be used as a local feature image. When stitching the local feature images, deduplication processing is required.

[0093] For example, Figure 6 This is a schematic diagram of a local feature image provided in Embodiment 2 of this application, as shown below. Figure 6 As shown in (a), the vehicle's center of motion is located within the first effective region, and the pixel region from the vehicle's center of motion to the lower edge of the first candidate vehicle image is taken as the local feature region. Figure 6 As shown in (b), the vehicle's motion center is located within the second effective region, and the pixel region from the vehicle's motion center to the upper edge of the second candidate vehicle image is taken as the local feature region. By determining the local feature image, the target vehicle image can be stitched based on the segmented local feature image, thus improving the integrity of the target vehicle.

[0094] In this technical solution, optionally, feature segmentation is performed on each candidate vehicle image based on the vehicle motion center to obtain a local feature image corresponding to each candidate vehicle image, and the solution further includes:

[0095] If the candidate vehicle image is the third candidate vehicle image, then the pixel region containing the center of vehicle motion between the center of vehicle motion is used as the local feature image.

[0096] In this embodiment, if there are n vehicle motion centers, it indicates a relatively large target. Therefore, the candidate vehicle image needs to be segmented into n+1 frames. Similar to the vehicle motion center processing process, the vehicle motion centers need to be determined sequentially, and the first and last frames should conform to... The pixel determination range is such that intermediate frames can be extracted as long as they remain in the frame.

[0097] For example, Figure 7 This is a schematic diagram of another partial feature image provided in Embodiment 2 of this application, such as... Figure 7 As shown in (a), the center of motion of the first vehicle is located within the first effective area, and the pixel region from the center of motion of the first vehicle to the lower edge of the first candidate vehicle image is used as the local feature image. Figure 7 As shown in (b), if the candidate vehicle image contains at least two vehicle motion centers, then the pixel region between the vehicle motion centers can be used as a local feature region. Figure 7As shown in (c), the motion center of the nth vehicle is located within the second effective region, and the pixel region from the motion center of the nth vehicle to the upper edge of the second candidate vehicle image is taken as the local feature region.

[0098] By identifying local feature images, target vehicle images can be stitched together based on the segmented local feature images, thereby improving the integrity of the target vehicle.

[0099] S540. The local feature images are stitched together to obtain the target vehicle image.

[0100] The technical solution provided in this application acquires at least two original vehicle images and the vehicle motion center of each original vehicle image. The first original vehicle image whose vehicle motion center is located within a first effective region is selected as a first candidate vehicle image, and the last original vehicle image whose vehicle motion center is located within a second effective region is selected as a second candidate vehicle image. Using the vehicle motion center, feature segmentation is performed on the at least two candidate vehicle images to obtain local feature images corresponding to each candidate vehicle image. These local feature images are then stitched together to obtain the target vehicle image. By implementing this technical solution, candidate vehicle images can be segmented using the vehicle motion center, and the target vehicle image can be stitched together based on the segmented local feature images, thereby improving the integrity of the target vehicle image.

[0101] Example 3

[0102] Figure 8 This is a schematic diagram of the image stitching device provided in Embodiment 3 of this application, as shown below. Figure 8 As shown, the image stitching device includes:

[0103] The vehicle motion center acquisition module 810 is used to acquire at least two original vehicle images and the vehicle motion center of each original vehicle image.

[0104] The candidate vehicle image determination module 820 is used to determine at least two candidate vehicle images from each original vehicle image based on the vehicle motion center.

[0105] The local feature image acquisition module 830 is used to perform feature segmentation on each candidate vehicle image based on the vehicle motion center to obtain a local feature image corresponding to each candidate vehicle image.

[0106] The target vehicle image acquisition module 840 is used to stitch together the local feature images to obtain the target vehicle image.

[0107] In this technical solution, optionally, the candidate vehicle image determination module 820 is specifically used for:

[0108] The first frame of the original vehicle image in which the center of motion of the vehicle is located within the first effective area is used as the first candidate vehicle image; and the last frame of the original vehicle image in which the center of motion of the vehicle is located within the second effective area is used as the second candidate vehicle image.

[0109] The first effective region and the second effective region are determined based on the number of vehicle motion centers and the size of the original vehicle image, respectively.

[0110] In this technical solution, optionally, the local feature image acquisition module 830 is specifically used for:

[0111] If the candidate vehicle image is the first candidate vehicle image, then the pixel region containing the vehicle's center of motion and the first edge of the first candidate vehicle image will be used as the local feature image.

[0112] If the candidate vehicle image is the second candidate vehicle image, then the pixel region containing the vehicle's center of motion and the second edge of the second candidate vehicle image will be used as the local feature image.

[0113] In this technical solution, optionally, the candidate vehicle image determination module 820 is also used for:

[0114] Determine whether the number of vehicle motion centers in the original vehicle image meets the quantity condition;

[0115] If the conditions are met, the original vehicle image will be used as the third candidate vehicle image.

[0116] In this technical solution, optionally, the local feature image acquisition module 830 is also used for:

[0117] If the candidate vehicle image is the third candidate vehicle image, then the pixel region containing the center of vehicle motion between the center of vehicle motion is used as the local feature image.

[0118] In this technical solution, optionally, the target vehicle image acquisition module 840 includes:

[0119] The splicing region is obtained by unit, which is used to divide the local feature image into regions to obtain the splicing region;

[0120] The image stitching unit is used to move the stitching area where the vehicle's motion center is located to a pre-set image center to obtain at least two stitched images after the movement.

[0121] The target vehicle image acquisition unit is used to stitch together the at least two stitched images to obtain the target vehicle image.

[0122] In this technical solution, optionally, the vehicle motion center acquisition module 810 includes:

[0123] The vehicle distance motion center acquisition unit is used to acquire the vehicle clustering motion center based on the position sensor of each original vehicle image;

[0124] The vehicle motion center is obtained by performing coordinate transformation on the vehicle cluster motion center to obtain the vehicle motion center.

[0125] The above-mentioned products can perform the methods provided in the embodiments of this application, and have the corresponding functional modules and beneficial effects of performing the methods.

[0126] Example 4

[0127] This application embodiment also provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform an image stitching method, the method comprising:

[0128] Acquire at least two original vehicle images, and the vehicle motion center of each original vehicle image;

[0129] Based on the vehicle motion center, at least two candidate vehicle images are determined from each original vehicle image;

[0130] Based on the vehicle motion center, feature segmentation is performed on each candidate vehicle image to obtain a local feature image corresponding to each candidate vehicle image.

[0131] The local feature images are stitched together to obtain the target vehicle image.

[0132] Storage medium – any type of memory device or storage device. The term “storage medium” is intended to include: mounting media, such as CD-ROM, floppy disk, or magnetic tape devices; computer system memory or random access memory, such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; non-volatile memory, such as flash memory, magnetic media (e.g., hard disk or optical storage); registers or other similar types of memory elements, etc. Storage medium may also include other types of memory or combinations thereof. Furthermore, storage medium may reside in a computer system in which a program is executed, or it may reside in a different second computer system connected to the computer system via a network (such as the Internet). The second computer system can provide program instructions to the computer for execution. The term “storage medium” can include two or more storage media that may reside in different locations (e.g., in different computer systems connected via a network). Storage medium may store program instructions (e.g., specifically implemented as a computer program) that can be executed by one or more processors.

[0133] Of course, the computer-executable instructions provided in the embodiments of this application are not limited to the image stitching operation described above, but can also perform related operations in the image stitching method provided in any embodiment of this application.

[0134] Example 5

[0135] This application provides an electronic device that can integrate the image stitching device provided in this application. Figure 9 This is a schematic diagram of the structure of an electronic device provided in Embodiment 5 of this application. Figure 9 As shown, this embodiment provides an electronic device 900, which includes: one or more processors 920; and a storage device 910 for storing one or more programs. When the one or more programs are executed by the one or more processors 920, the one or more processors 920 implement the image stitching method provided in this embodiment, the method including:

[0136] Acquire at least two original vehicle images, and the vehicle motion center of each original vehicle image;

[0137] Based on the vehicle motion center, at least two candidate vehicle images are determined from each original vehicle image;

[0138] Based on the vehicle motion center, feature segmentation is performed on each candidate vehicle image to obtain a local feature image corresponding to each candidate vehicle image.

[0139] The local feature images are stitched together to obtain the target vehicle image.

[0140] Of course, those skilled in the art will understand that the processor 920 also implements the technical solutions of the image stitching method provided in any embodiment of this application.

[0141] Figure 9 The electronic device 900 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0142] like Figure 9 As shown, the electronic device 900 includes a processor 920, a storage device 910, an input device 930, and an output device 940; the number of processors 920 in the electronic device can be one or more. Figure 9 Taking a processor 920 as an example; the processor 920, storage device 910, input device 930, and output device 940 in the electronic device can be connected via a bus or other means. Figure 9 Taking the connection between China and Israel via bus 950 as an example.

[0143] The storage device 910, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and module units, such as the program instructions corresponding to the image stitching method in the embodiments of this application.

[0144] Storage device 910 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a given function; the data storage area may store data created based on terminal usage. Furthermore, storage device 910 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory, or other non-volatile solid-state storage device. In some instances, storage device 910 may further include memory remotely located relative to processor 920, which can be connected via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0145] Input device 930 can be used to receive input digital, character, or voice information, and to generate key signal inputs related to user settings and function control of the electronic device. Output device 940 may include electronic devices such as a display screen and a speaker.

[0146] The electronic device provided in this application embodiment can achieve the purpose of improving the integrity of the target vehicle.

[0147] The image stitching apparatus, storage medium, and electronic device provided in the above embodiments can execute the image stitching method provided in any embodiment of this application, and have the corresponding functional modules and beneficial effects of executing the method. Technical details not described in detail in the above embodiments can be found in the image stitching method provided in any embodiment of this application.

[0148] Note that the above are merely preferred embodiments and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application, the scope of which is determined by the scope of the appended claims.

Claims

1. An image stitching method, characterized by, include: Acquire at least two original vehicle images, and the vehicle motion center of each original vehicle image; Based on the vehicle motion center, at least two candidate vehicle images are determined from each original vehicle image; Based on the vehicle motion center, feature segmentation is performed on each candidate vehicle image to obtain a local feature image corresponding to each candidate vehicle image. The local feature images are stitched together to obtain the target vehicle image; The step of determining at least two candidate vehicle images from each original vehicle image based on the vehicle's motion center includes: The first frame of the original vehicle image in which the center of motion of the vehicle is located within the first effective area is used as the first candidate vehicle image; and the last frame of the original vehicle image in which the center of motion of the vehicle is located within the second effective area is used as the second candidate vehicle image. The first effective region and the second effective region are determined based on the number of vehicle motion centers and the size of the original vehicle image, respectively.

2. The method of claim 1, wherein, Based on the vehicle motion center, feature segmentation is performed on each candidate vehicle image to obtain a local feature image corresponding to each candidate vehicle image, including: If the candidate vehicle image is the first candidate vehicle image, then the pixel region containing the vehicle's center of motion and the first edge of the first candidate vehicle image will be used as the local feature image. If the candidate vehicle image is the second candidate vehicle image, then the pixel region containing the vehicle's center of motion and the second edge of the second candidate vehicle image will be used as the local feature image.

3. The method of claim 1, wherein, Based on the vehicle motion center, at least two candidate vehicle images are determined from each original vehicle image, further comprising: Determine whether the number of vehicle motion centers in the original vehicle image meets the quantity condition; If the conditions are met, the original vehicle image will be used as the third candidate vehicle image.

4. The method of claim 1, wherein, Based on the vehicle motion center, feature segmentation is performed on each candidate vehicle image to obtain a local feature image corresponding to each candidate vehicle image, and the method further includes: If the candidate vehicle image is the third candidate vehicle image, then the pixel region containing the center of vehicle motion between the center of vehicle motion is used as the local feature image.

5. The method of claim 1, wherein, The local feature images are stitched together to obtain the target vehicle image, including: The local feature image is divided into regions to obtain the stitched region; Move the stitching area where the vehicle's motion center is located to the pre-set image center to obtain at least two stitched images after the movement; The at least two stitched images are stitched together to obtain the target vehicle image.

6. The method according to claim 1, characterized in that, Obtain the vehicle motion center of each original vehicle image, including: Vehicle clustering motion centers based on original vehicle images acquired from position sensors; The coordinate transformation of the vehicle cluster motion center yields the vehicle motion center.

7. An image stitching device, characterized in that, include: The vehicle motion center acquisition module is used to acquire at least two original vehicle images and the vehicle motion center of each original vehicle image. The candidate vehicle image determination module is used to determine at least two candidate vehicle images from each original vehicle image based on the vehicle's center of motion. The local feature image acquisition module is used to perform feature segmentation on each candidate vehicle image based on the vehicle motion center to obtain a local feature image corresponding to each candidate vehicle image. The target vehicle image acquisition module is used to stitch together the local feature images to obtain the target vehicle image; The candidate vehicle image determination module is specifically used for: The first frame of the original vehicle image in which the center of motion of the vehicle is located within the first effective area is used as the first candidate vehicle image; and the last frame of the original vehicle image in which the center of motion of the vehicle is located within the second effective area is used as the second candidate vehicle image. The first effective region and the second effective region are determined based on the number of vehicle motion centers and the size of the original vehicle image, respectively.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the image stitching method as described in any one of claims 1-6.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the image stitching method as described in any one of claims 1-6.

Citation Information

Patent Citations

  • Over-limit information acquisition method, device and system and over-limit detection equipment

    CN111192462A

  • Vehicle side view splicing device and method

    CN111862623A