CN101790022A - Method for detecting moving vehicles based on JPEG (Joint Photographic Experts Group) image - Google Patents
Method for detecting moving vehicles based on JPEG (Joint Photographic Experts Group) image Download PDFInfo
- Publication number
- CN101790022A CN101790022A CN 200910219168 CN200910219168A CN101790022A CN 101790022 A CN101790022 A CN 101790022A CN 200910219168 CN200910219168 CN 200910219168 CN 200910219168 A CN200910219168 A CN 200910219168A CN 101790022 A CN101790022 A CN 101790022A
- Authority
- CN
- China
- Prior art keywords
- image
- jpeg
- pixel
- moving vehicles
- surveyed area
- 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
Images
Abstract
The invention discloses a method for detecting moving vehicles based on a JPEG (Joint Photographic Experts Group) image, belonging to the technical field of digital image processing for detecting moving vehicles. The method comprises the following steps of: classifying two adjacent frames of JPEG image in a sequence image according to the presence probability of source symbols based on respective Huffman coding length, sequencing maximal length codes by size, reversely coding by combination and Z-shaped decompression to obtain compressed domain images of the two frames of image, wherein the pixel of the compressed domain images corresponds to the position of a pixel of completely-decompressed images; respectively extracting the detection areas of the two frames of compressed-domain image according to the specifically-set detection area to form two new pixel matrixes, respectively storing the pixel matrixes in two specified storage areas, comparing the pixel value corresponding to the two image matrixes in the two storage areas, and detecting the moving vehicles by comparing the summation of the difference value and the set threshold. Under the same condition, the time for decompressing two 308x308 JPEG images is decreased from 1.985143 seconds of the prior art to 0.441066 seconds.
Description
Technical field
The present invention relates to a kind of moving vehicle detection method, particularly, be used for image encoding and pattern recognition based on the moving vehicle detection method of jpeg image.Belong to the digital image processing techniques field.
Background technology
In traffic video monitoring system, handle apace and transmit traffic image, extract information of vehicles traffic is monitored in real time, traffic guidance, traffic accident warnings etc. are significant.And vehicle detection is the speed of a motor vehicle, the basis that traffic parameters such as vehicle flowrate extract and break in traffic rules and regulations is supervised.
Existing video frequency vehicle detects and is primarily aimed at that two kinds of images inputs carry out, i.e. BMP image and jpeg image.And handled image is the BMP form in present most of traffic video monitoring system.The BMP image adopts position mapping storage format, except picture depth is optional, does not adopt other any compression, and institute takes up space very big.This is for the system of outdoor studies such as traffic video monitoring, and treatments B MP image is very high to the memory space and the processing speed requirement of system, and transmission has brought very big burden to image, and therefore, this picture format seldom is used to transmission.Compressed image as jpeg image, utilizes people's vision system characteristic, removes or reduce those to the insensitive data of eyes, and its compression ratio is very high, and is less to the picture quality influence simultaneously.Store and transmit undoubtedly and can reduce data volume significantly with jpeg image, increase work efficiency.Yet as can be known, the compression of jpeg image mainly comprises following step according to " JPEG stationary image compression coding standard ": color conversion, DCT (discrete cosine transform) quantizes, zigzag coding, huffman coding.Though behind the method coding, average code word is the shortest, most effective, code word is different in size.Even if just making, this changes two little frame adjacent images, behind coding, and the size of image no matter, the length of code word, still variation has all taken place in the value of every middle code word.This can not directly handle existing image processing algorithm.Patent 200610117273.X, patent 200410059169.0, gathered for the BMP image, and, obtain traffic parameter to the handling of this image, for satisfying the transmission needs, transmit after the BMP image is compressed simultaneously.The obvious like this storage pressure that has increased the weight of system.Development along with the CMOS technology, the traffic image of many traffic video monitoring system collections is a jpeg format, because this picture format is not easy to direct processing, usually need carry out whole or local decompression with canned software to jpeg image, and then carrying out the extraction of motion detection and traffic parameter, the traffic parameter with original jpeg image and extraction transmits together at last.The decompression process of jpeg image anti-cosine transform especially wherein is very consuming time.
Summary of the invention
In order to overcome long deficiency of prior art decompression time, the invention provides a kind of moving vehicle detection method based on jpeg image, adopt the compression of FPGA hardware local solution, the moving region in the abstraction sequence image can reduce processing time and hardware data storage pressure significantly.
The technical solution adopted for the present invention to solve the technical problems: a kind of moving vehicle detection method based on jpeg image, be characterized in: the selected surveyed area that meets the JPEG coding, sequence jpeg image surveyed area to adjacent two frames is analyzed, for first frame, FPGA at first carries out source symbol occurrence probability according to huffman coding length and classifies automatically, again the maximum length coding is sorted by size, decompress by combination and zigzag then and carry out Gray code, the respective value of the surveyed area behind the Gray code is placed on first storage zone of appointment; For second two field picture, to carry out Huffman Gray code and zigzag in the same way and decompress, the surveyed area respective value is placed on second zone storage of appointment; Ask first storage portions and the second storage portions difference, compare with preset threshold, carry out detection moving vehicle by the difference summation of two storage portions.
The invention has the beneficial effects as follows: at first, only jpeg image is carried out Huffman Gray code and zigzag decompression, obtain and the corresponding compression domain image of original image, this image can react original image information to a certain extent, make that conventional images processing method (as image difference computing, foreground detection, motion detection etc.) is used, simplify the decompression process of jpeg image complexity, greatly reduced the decompression time.Under identical experiment condition, the gray level image of two width of cloth 308*308 is handled in the example.Jpeg image is decompressed, and carrying out motion detection then needs 1.985143 seconds, and utilization the present invention only needs 0.441066 second, has saved for nearly 78% decompression time.Secondly, only in the surveyed area of compression domain, carry out the difference computing, detect moving object, removed non-surveyed area data, reduced deposit data amount and the operand of FPGA, improved detection time.The present invention can be widely used in the image processing real-time, image storage space, and in the system that transmission speed is had relatively high expectations, as the vehicle flowrate of intersection, the overspeed of vehicle monitoring in highway sections such as highway etc.
Below in conjunction with the drawings and specific embodiments the present invention is elaborated.
Description of drawings
Fig. 1 is the moving vehicle detection method flow chart that the present invention is based on jpeg image;
Fig. 2 is the jpeg image flow process figure that partly decompresses among Fig. 1;
Fig. 3 is the inventive method surveyed area schematic diagram (chain-dotted line zone).
Embodiment
In order to understand technical scheme of the present invention better, couple the present invention elaborates below in conjunction with Fig. 1～3:
At two frame jpeg images in the image sequence that obtains under the same traffic scene.Implementation step is as follows:
1.FPGA read in the two continuous frames image in the image sequence;
2. respectively two two field pictures are carried out following operation: FPGA and at first carry out source symbol occurrence probability and classify automatically, again the maximum length coding is sorted by size, then by combination and zigzag decompression carrying out Gray code according to huffman coding length; Obtain the compression domain image of two two field pictures, the pixel of compression domain image can react original image information to a certain extent, but its a lot of pixel values is 0, so data volume and shared memory space are less with the decompressing image location of pixels is corresponding fully;
3. because in the traffic video monitoring, video camera maintains static (ignoring the slight jitter that external factor causes), therefore can surveyed area be set according to concrete traffic application scenarios.Extract the surveyed area in the two frame compression domain images respectively, form new picture element matrix, and it is stored in first storage area of appointment, back one frame surveyed area picture element matrix is stored in second storage area;
4. ask the absolute value of two image array respective pixel value difference values in two storage areas.There is moving vehicle to exist if the difference sum greater than certain setting threshold, is then thought, extracts moving vehicle; Otherwise, think not have moving vehicle, continue to detect successive image.
Claims (1)
1. moving vehicle detection method based on jpeg image, it is characterized in that comprising the steps: the selected surveyed area that meets the JPEG coding, sequence jpeg image surveyed area to adjacent two frames is analyzed, for first frame, FPGA at first carries out source symbol occurrence probability according to huffman coding length and classifies automatically, again the maximum length coding is sorted by size, decompress by combination and zigzag then and carry out Gray code, the respective value of the surveyed area behind the Gray code is placed on first storage zone of appointment; For second two field picture, to carry out Huffman Gray code and zigzag in the same way and decompress, the surveyed area respective value is placed on second zone storage of appointment; Ask first storage portions and the second storage portions difference, compare with preset threshold, carry out detection moving vehicle by the difference summation of two storage portions.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910219168 CN101790022A (en) | 2009-11-26 | 2009-11-26 | Method for detecting moving vehicles based on JPEG (Joint Photographic Experts Group) image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200910219168 CN101790022A (en) | 2009-11-26 | 2009-11-26 | Method for detecting moving vehicles based on JPEG (Joint Photographic Experts Group) image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101790022A true CN101790022A (en) | 2010-07-28 |
Family
ID=42533063
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200910219168 Pending CN101790022A (en) | 2009-11-26 | 2009-11-26 | Method for detecting moving vehicles based on JPEG (Joint Photographic Experts Group) image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101790022A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111614514A (en) * | 2020-04-30 | 2020-09-01 | 北京邮电大学 | Network traffic identification method and device |
-
2009
- 2009-11-26 CN CN 200910219168 patent/CN101790022A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111614514A (en) * | 2020-04-30 | 2020-09-01 | 北京邮电大学 | Network traffic identification method and device |
| CN111614514B (en) * | 2020-04-30 | 2021-09-24 | 北京邮电大学 | Network traffic identification method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2016173277A1 (en) | Video coding and decoding methods and apparatus | |
| US20230336754A1 (en) | Video compression using deep generative models | |
| Wiseman | Real-time monitoring of traffic congestions | |
| KR102194499B1 (en) | Apparatus for CCTV Video Analytics Based on Object-Image Recognition DCNN and Driving Method Thereof | |
| Wiseman | Take a picture of your tire! | |
| Wiseman | Camera that takes pictures of aircraft and ground vehicle tires can save lives | |
| Wiseman | Tool for online observing of traffic congestions | |
| Wiseman | The effectiveness of JPEG images produced by a standard digital camera to detect damaged tyres | |
| CN111901604B (en) | Video compression method, video reconstruction method, corresponding devices, camera and video processing equipment | |
| Wiseman | Computerized traffic congestion detection system | |
| CN113158738B (en) | Port environment target detection method, system, terminal and readable storage medium based on attention mechanism | |
| WO2016205700A1 (en) | Steganographic depth images | |
| WO2022067656A1 (en) | Image processing method and apparatus | |
| US10198628B2 (en) | Method and apparatus for determining a document suitability for server-based optical character recognition (OCR) processing | |
| CN110826429A (en) | Scenic spot video-based method and system for automatically monitoring travel emergency | |
| CN110991310B (en) | Portrait detection method, device, electronic equipment and computer readable medium | |
| US20230127009A1 (en) | Joint objects image signal processing in temporal domain | |
| CN110674787A (en) | Video decompression method and system based on Hog feature and lgb classifier | |
| JP7255819B2 (en) | Systems and methods for use in object detection from video streams | |
| CN113810654A (en) | Image video uploading method and device, storage medium and electronic equipment | |
| Jianyong et al. | A novel vehicle's shadow detection and removal algorithm | |
| CN101790022A (en) | Method for detecting moving vehicles based on JPEG (Joint Photographic Experts Group) image | |
| Schreiber et al. | GPU-based non-parametric background subtraction for a practical surveillance system | |
| CN116052090A (en) | Image quality evaluation method, model training method, device, equipment and medium | |
| CN112714313A (en) | Image processing method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20100728 |