CN114554158A - Panoramic video stitching method and system based on road traffic scene - Google Patents
Panoramic video stitching method and system based on road traffic scene Download PDFInfo
- Publication number
- CN114554158A CN114554158A CN202210188383.4A CN202210188383A CN114554158A CN 114554158 A CN114554158 A CN 114554158A CN 202210188383 A CN202210188383 A CN 202210188383A CN 114554158 A CN114554158 A CN 114554158A
- Authority
- CN
- China
- Prior art keywords
- image
- matching
- panoramic video
- images
- road traffic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 230000003068 static effect Effects 0.000 claims abstract description 20
- 230000004927 fusion Effects 0.000 claims abstract description 18
- 238000001914 filtration Methods 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims abstract description 10
- 230000009466 transformation Effects 0.000 claims abstract description 7
- 238000013135 deep learning Methods 0.000 claims abstract description 5
- 238000012937 correction Methods 0.000 claims abstract description 3
- 238000004422 calculation algorithm Methods 0.000 claims description 18
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 238000000605 extraction Methods 0.000 abstract description 15
- 238000012544 monitoring process Methods 0.000 abstract description 9
- 238000005267 amalgamation Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 4
- 230000001360 synchronised effect Effects 0.000 description 4
- 238000009499 grossing Methods 0.000 description 3
- 229920001651 Cyanoacrylate Polymers 0.000 description 1
- 239000004830 Super Glue Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/181—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a plurality of remote sources
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/60—Control of cameras or camera modules
- H04N23/698—Control of cameras or camera modules for achieving an enlarged field of view, e.g. panoramic image capture
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/222—Studio circuitry; Studio devices; Studio equipment
- H04N5/262—Studio circuits, e.g. for mixing, switching-over, change of character of image, other special effects ; Cameras specially adapted for the electronic generation of special effects
- H04N5/265—Mixing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/222—Studio circuitry; Studio devices; Studio equipment
- H04N5/262—Studio circuits, e.g. for mixing, switching-over, change of character of image, other special effects ; Cameras specially adapted for the electronic generation of special effects
- H04N5/272—Means for inserting a foreground image in a background image, i.e. inlay, outlay
Abstract
The invention discloses a panoramic video stitching method and a panoramic video stitching system based on a road traffic scene, which comprise image acquisition, image distortion correction and pretreatment, image characteristic point extraction, image matching, matched characteristic point pair filtering, acquisition of a homography matrix of projection transformation and image fusion. The extraction and the feature point matching of the feature points are realized through a deep learning network, when the image feature points are extracted, the feature points of a dynamic foreground and a static background are respectively extracted, the relative position relation between different cameras is dynamically obtained by extracting the feature points of the background and the foreground in an overlapping area between the different cameras and matching the feature points, and the real-time fusion and splicing of images among the cameras in a parcel road traffic system are realized; the problem of traditional single camera monitoring system, need relevant personnel's manual work to call out corresponding a plurality of camera images, compare the amalgamation, just can acquire useful information scheduling is solved, perfect the target is simultaneously followed tracks the problem across the camera.
Description
Technical Field
The invention relates to the technical field of traffic management and video monitoring, in particular to a panoramic video stitching method and a panoramic video stitching system based on a road traffic scene.
Background
With the gradual flourishing of economy and the continuous development of science and technology, the comprehensive real-time monitoring of urban road traffic becomes more important, and the establishment of a comprehensive road traffic image management platform by scientific and technical means becomes the target of traffic management departments in various regions. At present, a road traffic monitoring system basically realizes a digital network system covering main roads, auxiliary roads and intersections of cities and towns, is provided with corresponding image monitoring equipment and operating software, and can transmit the running condition of road traffic to a corresponding dispatching center in real time to monitor the running condition of the road traffic. Once a problem occurs, feedback is made at a first time. However, the existing monitoring equipment is mainly the existing security cameras, there is no intuitive position relation between the cameras, once the condition of crossing the cameras is found, the images of a plurality of corresponding cameras need to be called manually by related personnel, and the images are compared one by one to be spliced, so that the useful information can be acquired, which is time-consuming and labor-consuming.
CN113055613A discloses a "method and apparatus for panoramic video stitching based on mine scene", which adopts a traditional SURF algorithm to extract feature points and generate feature descriptors, where the SURF algorithm cannot implement real-time processing of video frames in practical application, and basically can exclude engineering application; moreover, the feature matching algorithm adopts a K nearest neighbor classification algorithm, and the matching error is large. For another example, CN103516995A discloses a method adopted by "a real-time panoramic video stitching method and apparatus based on ORB features" is as follows: starting to collect multi-path synchronous video data; performing feature point extraction on each image at the same moment by adopting an ORB feature extraction algorithm and calculating an ORB feature vector of each feature point; solving a homography matrix between corresponding frames of the synchronous video by adopting a nearest neighbor matching method and a RANSAC (random sample consensus) matching algorithm; splicing video frame scenes according to the homography matrix; and finally, outputting the spliced video. However, the comparison files are all extracted and matched by using a traditional image feature point extraction and matching method, such as sift or surf, and the extraction precision of the feature points is not high, the extraction time is too long, and real-time splicing cannot be realized.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a panoramic video stitching method and a panoramic video stitching system based on a road traffic scene, and solves the problems that the prior art is low in extraction precision, time-consuming and labor-consuming in searching and incapable of realizing real-time stitching.
In order to achieve the purpose, the invention adopts the following technical scheme:
a panoramic video splicing method based on a road traffic scene comprises the following steps:
s1, image acquisition: synchronously acquiring images of each camera in a test road section by a plurality of cameras, and requiring that the time stamps of the cameras are strictly consistent;
s2, image distortion correction and preprocessing: correcting and preprocessing a plurality of collected images by adopting a Thykes bright-dark channel defogging algorithm;
s3, extracting image feature points: extracting characteristic points of the image by adopting a deep learning network super; extracting characteristic points of an overlapping area between cameras, including angular points and points with obvious gradient change, and generating a characteristic descriptor;
s4, image matching: performing feature matching on the images by using a feature matching algorithm supervalue based on a graph convolution neural network, wherein feature points and descriptors extracted by a superpoint network in the two images are input, and the output is a matching relation between image features;
s5, filtering the matched feature point pairs: filtering the matched feature point pairs by using a K nearest neighbor algorithm;
s6, acquiring a homography matrix of projection transformation, and completing the projection transformation: and circularly traversing the residual matching point pairs after filtering by using a Randac algorithm until the point pairs required by the projection between the images are obtained through calculationOptimization ofThe homography matrix is used for completing the splicing among the images;
s7, image fusion: and searching the optimal splicing line by adopting dynamic programming to perform image fusion.
Further, the overlapping area in S3 is divided into a static background and a dynamic foreground, and the background and foreground feature points are extracted respectively.
In S4, the image feature matching includes static matching and dynamic matching; the static matching comprises matching of static objects such as roads, traffic signs, trees and the like in a road scene; dynamic matching includes matching that refers to pedestrians, vehicles, etc. The static matching is basically equivalent to a background for an image, and the feature points of the background can be matched independently; the dynamic matching is equivalent to the foreground for the image, and the characteristic points of the foreground can be matched independently.
Further, the image fusion uses a fast color interpolation model to search an optimal splicing line by combining with an adjacent shortest distance method to carry out image fusion.
The invention also provides a panoramic video stitching system based on the road traffic scene, which comprises a camera and a processor, wherein the processor executes the panoramic video stitching method.
Compared with the prior art, the invention has the following beneficial effects:
1. the invention realizes the extraction of the characteristic points and the matching of the characteristic points through a deep learning network, when the characteristic points of the image are extracted, the characteristic points of a dynamic foreground and a static background are respectively extracted, the relative position relation between different cameras, namely a homography matrix, is dynamically obtained by extracting the characteristic points of the background and the foreground in an overlapping area between different cameras and matching the characteristic points, and the real-time fusion and splicing of the images between the cameras in a film area road traffic system is realized.
2. The invention solves the problems that in a traditional single-camera monitoring system, relevant personnel need to manually call out corresponding multiple camera images and compare and combine the camera images one by one to obtain useful information, which is troublesome and labor-consuming, and simultaneously perfectly solves the problem of target cross-camera tracking.
3. According to the invention, the camera images in the overlapped area of the same area are subjected to panoramic stitching and fusion according to the relative position relationship, and then are directly projected to the corresponding large monitoring screen, so that the traffic condition of the whole area can be monitored more intuitively and clearly.
Drawings
FIG. 1 is a flow chart of a panoramic video stitching method based on a road traffic scene according to the present invention;
FIG. 2 is an example of an original image acquired using a wide-angle camera;
FIG. 3 is an example of a graph of the effect of an image after rectification;
FIG. 4 is an example of an image stitching effect graph;
fig. 5 is an image fusion effect illustration.
Detailed Description
In order to make those skilled in the art better understand the technical solution of the present invention, the following detailed description of the present invention is made with reference to the accompanying drawings, but the embodiments of the present invention are not limited thereto.
Referring to fig. 1, a flowchart of a panoramic video stitching method based on a road traffic scene is shown.
Firstly, image acquisition:
the method has the advantages that the multiple cameras are synchronized to simultaneously acquire images of each camera in a test road section, timestamps of each camera are required to be strictly consistent, each camera is subjected to image acquisition by adopting an independent process in the actual test process, the time difference between the fastest process and the slowest process is 0.5ms basically, and 25 frames per second synchronous acquisition can be met. Referring to fig. 2, an example of acquiring a cross-camera image using a Haikang Wide-Angle Camera is shown.
Secondly, correcting and preprocessing image distortion:
(2a) because the road camera mostly adopts the wide-angle camera, the image has serious radial distortion, which is shown as barrel-shaped distortion (as shown in figure 2), the original images of the three cameras are selected to cause image characteristic distortion, so that the camera parameters need to be corrected to eliminate distortion influence, and the processor adopts a Thank bright-dark channel defogging algorithm to correct and preprocess a plurality of collected images. An example of the effect of the rectified image is shown in fig. 3.
(2b) And preprocessing the image by adopting a Thykes bright-dark channel defogging algorithm. The camera is in open air environment for a long time, and the camera lens has the dust, and rainy day and fog day etc. all can cause the image unclear, need carry out corresponding preliminary treatment, promote image quality.
Thirdly, extracting image characteristic points:
the method adopts the deep learning network superpoint to extract the characteristic points of the image. The output of SuperPoint is not only the feature points but also descriptors of the feature points. And extracting characteristic points of an overlapping area between the cameras, including the angular points and the points with obvious gradient change, and generating a characteristic descriptor. Because the image information outside the overlapping region easily causes the mismatching of the feature points, the feature points can be extracted only in the overlapping region, so that the extraction speed of the feature points can be increased, and the matching accuracy can be improved. The overlapped area can be divided into a static background and a dynamic foreground, and characteristic points can be respectively extracted from the background and the foreground.
Fourthly, image matching:
the image matching includes static matching and dynamic matching, the static matching mainly refers to matching of static objects such as roads, traffic signs and trees in a road scene, and the dynamic matching refers to matching of pedestrians and vehicles. And respectively carrying out feature matching on the static and dynamic objects of the image according to the extracted feature descriptors. The invention adopts a character matching algorithm superglue based on a graph convolution neural network to carry out character matching on images, wherein the input is character points and descriptors extracted by the superpoint network in two images, and the output is the matching relation between image characters.
(4a) The static matching is basically equivalent to the background for the image, and the characteristic points of the background can be matched independently;
(4b) the dynamic matching is equivalent to the foreground for the image, and the characteristic points of the foreground can be matched independently.
Fifthly, matching characteristic point pairs and filtering:
the feature points extracted in the feature fuzzy region may become correctly matched interference points, resulting in mismatch, so that the matched feature points need to be filtered according to a certain rule. And filtering the matched characteristic point pairs by using a K nearest neighbor algorithm.
And sixthly, acquiring a homography matrix of projection transformation to finish the projection transformation:
and circularly traversing the residual matching point pairs after filtering by using a Randac algorithm until an optimal homography matrix required by projection between the images is obtained through calculation, and completing splicing between the images. Fig. 4 is an example of an image stitching effect map.
Seventhly, fusing images:
because of reasons such as installation angle between the different cameras for the colour and the luminance that the image demonstrates are not unified, cause the concatenation department very hard, and the effect is not good, still need treat the image of concatenation under this kind of condition and carry out image fusion operation. The invention adopts dynamic programming, namely, a fast color interpolation model is combined with an adjacent shortest distance method to search the optimal splicing line for image fusion. Fig. 5 is an image fusion effect illustration. The fast color interpolation model is mainly used for smoothing color difference values at the joint of two images by using the color gradient difference of the images in two adjacent cameras. The smoothing rule is to use the adjacent shortest distance method, that is, the color value of each pixel is gaussian filtered according to the color value of its eight neighborhoods, so as to achieve the purpose of smoothing color difference.
Therefore, the method has high extraction precision, realizes real-time panoramic stitching and fusion, and then directly projects the panoramic stitching and fusion to the corresponding large monitoring screen, so that the condition of the whole area can be monitored more intuitively and clearly. The innovation of the method is that the method for extracting the characteristic points adopts superpoint and the characteristic matching method adopts superslue; when the characteristics are extracted, the characteristics of the dynamic foreground and the static background are respectively extracted, so that the extracted characteristics have higher precision, and the similar points in the foreground information with wrong background information cannot be subjected to wrong matching; and the problem that the characteristic point extraction time of the traditional sift or surf is too long, so that real-time splicing cannot be realized is solved. Because the static background ratio in each frame of image is very high, but the change rate of the front frame and the rear frame is very low, the sift method is that each frame can extract the information of the whole image, which wastes calculation time, while the superpoint stores the background information of the image of the previous frame, each subsequent frame only needs to be differenced with the previous frame, and then only the feature extraction is carried out on the changed differential image, and only the feature extraction is carried out on the dynamic foreground, so that the time cost required by the feature extraction is greatly reduced.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present invention and not for limiting the technical solutions, and those skilled in the art should understand that modifications or equivalent substitutions can be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions, and all that should be covered by the claims of the present invention.
Claims (6)
1. A panoramic video stitching method based on a road traffic scene is characterized by comprising the following steps:
s1, image acquisition: synchronously acquiring images of each camera in a test road section by a plurality of cameras, and requiring that the time stamps of each camera are strictly consistent;
s2, image distortion correction and preprocessing: correcting and preprocessing a plurality of collected images by adopting a Thykes bright-dark channel defogging algorithm;
s3, extracting image feature points: extracting characteristic points of the image by adopting a deep learning network super; extracting characteristic points of an overlapping area between cameras, including angular points and points with obvious gradient change, and generating a characteristic descriptor;
s4, image matching: performing feature matching on the images by using a feature matching algorithm supervalue based on a graph convolution neural network, wherein feature points and descriptors extracted by a superpoint network in the two images are input, and the output is a matching relation between image features;
s5, filtering the matched feature point pairs: filtering the matched feature point pairs by using a K nearest neighbor algorithm;
s6, acquiring a homography matrix of projection transformation, and completing the projection transformation: and circularly traversing the residual matching point pairs after filtering by using a Randac algorithm until the point pairs required by the projection between the images are obtained through calculationOptimization ofThe homography matrix is used for completing the splicing among the images;
s7, image fusion: and searching the optimal splicing line by adopting dynamic programming to perform image fusion.
2. The method for stitching the panoramic video based on the road traffic scene as claimed in claim 1, wherein in the step S3, the overlapped area is divided into a static background and a dynamic foreground, and the background and foreground feature points are extracted respectively.
3. The method for stitching the panoramic video based on the road traffic scene as claimed in claim 1, wherein in the step S4, the image feature matching comprises static matching and dynamic matching; the static matching comprises matching of static objects such as roads, traffic signs, trees and the like in a road scene; dynamic matching includes matching that refers to pedestrians, vehicles, etc.
4. The method for stitching the panoramic video based on the road traffic scene as claimed in claim 1, wherein the static matching is substantially equivalent to the background for the image, and the feature points of the background can be matched separately; the dynamic matching is equivalent to the foreground for the image, and the characteristic points of the foreground can be matched independently.
5. The panoramic video stitching method based on the road traffic scene as claimed in claim 1, wherein the image fusion uses a fast color interpolation model in combination with a neighboring shortest distance method to find an optimal stitching line for image fusion.
6. A panoramic video stitching system based on a road traffic scene is characterized by comprising a camera and a processor, wherein the processor executes the panoramic video stitching method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210188383.4A CN114554158A (en) | 2022-02-28 | 2022-02-28 | Panoramic video stitching method and system based on road traffic scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210188383.4A CN114554158A (en) | 2022-02-28 | 2022-02-28 | Panoramic video stitching method and system based on road traffic scene |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114554158A true CN114554158A (en) | 2022-05-27 |
Family
ID=81680185
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210188383.4A Pending CN114554158A (en) | 2022-02-28 | 2022-02-28 | Panoramic video stitching method and system based on road traffic scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114554158A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117011553A (en) * | 2023-07-19 | 2023-11-07 | 苏州旭智设计营造有限公司 | Multi-scene conference free conversion control system |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102572450A (en) * | 2012-01-10 | 2012-07-11 | 中国传媒大学 | Three-dimensional video color calibration method based on scale invariant feature transform (SIFT) characteristics and generalized regression neural networks (GRNN) |
| CN108769578A (en) * | 2018-05-17 | 2018-11-06 | 南京理工大学 | A kind of real-time omnidirectional imaging system and method based on multi-path camera |
| CN110660023A (en) * | 2019-09-12 | 2020-01-07 | 中国测绘科学研究院 | Video stitching method based on image semantic segmentation |
| CN111599007A (en) * | 2020-05-26 | 2020-08-28 | 张仲靖 | Smart city CIM road mapping method based on unmanned aerial vehicle aerial photography |
| CN113159043A (en) * | 2021-04-01 | 2021-07-23 | 北京大学 | Feature point matching method and system based on semantic information |
| CN113793362A (en) * | 2021-09-22 | 2021-12-14 | 清华大学 | Pedestrian track extraction method and device based on multi-lens video |
-
2022
- 2022-02-28 CN CN202210188383.4A patent/CN114554158A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102572450A (en) * | 2012-01-10 | 2012-07-11 | 中国传媒大学 | Three-dimensional video color calibration method based on scale invariant feature transform (SIFT) characteristics and generalized regression neural networks (GRNN) |
| CN108769578A (en) * | 2018-05-17 | 2018-11-06 | 南京理工大学 | A kind of real-time omnidirectional imaging system and method based on multi-path camera |
| CN110660023A (en) * | 2019-09-12 | 2020-01-07 | 中国测绘科学研究院 | Video stitching method based on image semantic segmentation |
| CN111599007A (en) * | 2020-05-26 | 2020-08-28 | 张仲靖 | Smart city CIM road mapping method based on unmanned aerial vehicle aerial photography |
| CN113159043A (en) * | 2021-04-01 | 2021-07-23 | 北京大学 | Feature point matching method and system based on semantic information |
| CN113793362A (en) * | 2021-09-22 | 2021-12-14 | 清华大学 | Pedestrian track extraction method and device based on multi-lens video |
Non-Patent Citations (2)
| Title |
|---|
| 刘嗣超;武鹏达;赵占杰;李成名;: "交通监控视频图像语义分割及其拼接方法", 测绘学报, no. 04, 15 April 2020 (2020-04-15) * |
| 贾迪;朱宁丹;杨宁华;吴思;李玉秀;赵明远;: "图像匹配方法研究综述", 中国图象图形学报, no. 05, 16 May 2019 (2019-05-16) * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117011553A (en) * | 2023-07-19 | 2023-11-07 | 苏州旭智设计营造有限公司 | Multi-scene conference free conversion control system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107808389B (en) | Unsupervised video segmentation method based on deep learning | |
| CN111047510B (en) | Large-field-angle image real-time splicing method based on calibration | |
| CN108769578B (en) | Real-time panoramic imaging system and method based on multiple cameras | |
| CN111968129A (en) | Instant positioning and map construction system and method with semantic perception | |
| CN105184253B (en) | Face recognition method and face recognition system | |
| CN107274346A (en) | Real-time panoramic video splicing system | |
| CN112085659B (en) | Panorama splicing and fusing method and system based on dome camera and storage medium | |
| CN100542303C (en) | A kind of method for correcting multi-viewpoint vedio color | |
| CN106791623A (en) | A kind of panoramic video joining method and device | |
| CN104408701A (en) | Large-scale scene video image stitching method | |
| CN107154022A (en) | A kind of dynamic panorama mosaic method suitable for trailer | |
| CN105488777A (en) | System and method for generating panoramic picture in real time based on moving foreground | |
| CN107085842A (en) | The real-time antidote and system of self study multiway images fusion | |
| CN113221665A (en) | Video fusion algorithm based on dynamic optimal suture line and improved gradual-in and gradual-out method | |
| CN105894443A (en) | Method for splicing videos in real time based on SURF (Speeded UP Robust Features) algorithm | |
| CN107992837A (en) | Road full-view modeling and vehicle detecting and tracking method based on single PTZ monitor cameras | |
| CN112784834A (en) | Automatic license plate identification method in natural scene | |
| CN112712487A (en) | Scene video fusion method and system, electronic equipment and storage medium | |
| CN114554158A (en) | Panoramic video stitching method and system based on road traffic scene | |
| CN108737743B (en) | Video splicing device and video splicing method based on image splicing | |
| CN114898353A (en) | License plate identification method based on video sequence image characteristics and information | |
| CN101945299B (en) | Camera-equipment-array based dynamic scene depth restoring method | |
| CN117114997B (en) | Image stitching method and device based on suture line search algorithm | |
| CN107330856B (en) | Panoramic imaging method based on projective transformation and thin plate spline | |
| Li et al. | Gyroflow+: Gyroscope-guided unsupervised deep homography and optical flow learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |