CN103077393A - DSP (Digital Signal Processor)-based vehicle-mounted real-time moving target detection system and method thereof - Google Patents
DSP (Digital Signal Processor)-based vehicle-mounted real-time moving target detection system and method thereof Download PDFInfo
- Publication number
- CN103077393A CN103077393A CN2013100317587A CN201310031758A CN103077393A CN 103077393 A CN103077393 A CN 103077393A CN 2013100317587 A CN2013100317587 A CN 2013100317587A CN 201310031758 A CN201310031758 A CN 201310031758A CN 103077393 A CN103077393 A CN 103077393A
- Authority
- CN
- China
- Prior art keywords
- signal processor
- digital signal
- module
- processor dsp
- dsp
- 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.)
- Granted
Links
Images
Landscapes
- Image Analysis (AREA)
- Closed-Circuit Television Systems (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a DSP (Digital Signal Processor)-based vehicle-mounted real-time moving target detection system and a method thereof. According to the invention, the road condition on an open road can be detected in real time, and when an appointed target occurs, an alarm is sent. A charge coupling device (CCD) camera is used for collecting an analog signal, the analog signal is digitalized by using a decoding chip, digital image data is transmitted to a Ping-Pong Buffer of an extended memory by using a direct memory access (DMA) technology for temporary storage; and the DSP reads one frame of image from the extended memory according to an image count value, the images are classified by using a machine learning classifying algorithm, and when three continuous frames of images are classified as targets, a sound-light alarm is sent out for reminding a driver to attend. The DSP-based vehicle-mounted real-time moving target detection system has the advantages of low cost, small volume, light weight and strong instantaneity, and is especially suitable for real-time detection and alarm on a moving target on an open road by a vehicle-mounted system.
Description
Technical field
The invention belongs to technical field of image processing, further relate in the digital signal DSP processing technology field vehicle-mounted real time kinematics object detection system and method thereof based on DSP.The present invention utilizes digital signal dsp processor and moving object detection system and method thereof, can be implemented on the open highway moving target to vehicle front and detects in real time and report to the police.
Background technology
Mechanical transport is day by day flourishing at present, and open Highway Conditions is complicated, often has humans and animals to pass through, and therefore guarantees that vehicle security drive is significant.Generally at automobile imaging device is installed, vehicle front road conditions video is detected in real time.This method need to be utilized target detection technique.
The patented claim that Shanghai Communications University proposes " infrared target detects the recognition and tracking system " (number of patent application 200310109070.2, grant number CN1546993A) discloses a kind of target detection recognition and tracking system.This system comprises that infrared imaging, infrared image reception, infrared image processing and image and result show four parts, by the infrared thermoviewer imaging, deliver to double digital signal processor dsp board card system through receiving unit, infrared image is carried out target detection follow the tracks of and identify, result and raw image data are delivered on the display by the bus of main frame and are shown.But the deficiency that this patented claim still exists is: the first, and the infrared thermoviewer price is high, is unfavorable for practical application; The second, main frame and display volume weight are larger, are unfavorable for being applied in the onboard system; The 3rd, recognition time is longer, and real-time is poor.
Summary of the invention
The object of the invention is to overcome the deficiency of above-mentioned prior art, a kind of vehicle-mounted real time kinematics object detection system and method thereof based on DSP proposed, by using common charge coupled cell CCD camera, collection vehicle the place ahead road conditions video, use digital signal processor DSP that video is analyzed, real-time designated movement target to vehicle front detects and reports to the police, and simultaneity factor is light, cheap, can satisfy vehicle-mounted practical application.By using this system, can ensure the driving safety that open highway is got on the car.
System of the present invention comprises video acquisition decoder module, digital signal processor DSP module, extension storage module and acousto-optic alarm module; The video acquisition decoder module is connected with the parallel peripheral bus PPI interface of digital signal processor DSP module, the extension storage module is connected with the expansion bus interface unit EBIU of digital signal processor DSP module, and the sound and light alarm module is connected with the universal input output GPIO interface of digital signal processor DSP module; Wherein:
Described video acquisition decoder module comprises charge coupled cell CCD camera and decoding chip, is used for collection and the digitizing of real time video image;
Described digital signal processor DSP module has direct memory access DMA function, is used for carrying out detection algorithm image is detected, and whether analyze wherein has intended target;
Described extension storage module comprises synchronous DRAM SDRAM, be used for the temporary view data that collects, and storage will be identified the clarification of objective training set;
Described sound and light alarm module is used for sounding and light alarm.
The concrete steps of the inventive method are as follows:
(1) obtains video
1a) the analog video signal of charge coupled cell CCD camera collection vehicle front;
1b) the decoding chip in the video acquisition decoder module, analog video signal is converted into Digital Image Data after, again each frame of digital view data is sent to the digital signal processor DSP module;
(2) transmit image data
2a) digital signal processor DSP module, direct memory access DMA is set to PPI to EBIU passage, based on the two-dimensional model of descriptor;
2b) the digital signal processor DSP module gathers the Digital Image Data that decoder module sends by parallel peripheral bus PPI interface receiver, video;
2c) direct memory access DMA is transferred to expansion bus interface unit EBIU automatically with Digital Image Data from parallel peripheral bus PPI interface;
(3) image data temporary
3a) the extension storage module according to the shared storage size of every two field picture, marks off the storage area that can hold two two field pictures with synchronous DRAM SDRAM, as Ping-Pong Buffer zone;
3b) synchronous DRAM SDRAM from the expansion bus interface unit EBIU of digital signal processor DSP module, receives Digital Image Data;
3c) synchronous DRAM SDRAM, according to the parity frame order, circulation deposits in the Ping-Pong Buffer zone, whenever deposits a frame in the Digital Image Data that receives, sends a signal to digital signal processor DSP;
(4) count value initialization
Digital signal processor DSP respectively picture count value i and successive objective count value k is set to 0;
(5) read a two field picture
5a) when digital signal processor DSP received the signal that synchronous DRAM SDRAM sends, picture count value i added 1;
If 5b) picture count value i=3 changes execution in step 5a over to), otherwise, change execution in step 5c over to);
5c) digital signal processor DSP reads a frame image data that has just deposited in from the extension storage module;
(6) Images Classification
Digital signal processor DSP uses the machine learning classification algorithm, and a two field picture that reads is classified;
(7) whether image has target
If this two field picture is classified as the driftlessness class, change execution in step (4) over to; Otherwise, change execution in step (8) over to;
(8) whether report to the police
Successive objective count value k adds 1; If k=3 changes execution in step (9) over to, otherwise picture count value i=2 changes execution in step (5) over to;
(9) send warning
Digital signal processor DSP sends alerting signal to the sound and light alarm module, and the sound and light alarm module is sent ring sound and highlighted blinking light.
The present invention compared with prior art has the following advantages:
The first, owing to use charge coupled cell CCD camera collection video in the system of the present invention, overcome prior art and used the high shortcoming of infrared thermoviewer cost, so that system cost of the present invention is cheap, be conducive to practical application.
The second, owing to use the digital signal processor DSP performance objective to detect in the system of the present invention, use the sound and light alarm module to report to the police, overcome the large shortcoming of system bulk weight in the prior art, so that light and flexible of the present invention is conducive to vehicular applications.
The 3rd, because method of the present invention adopts machine learning algorithm, direct memory access DMA technology and Ping-Pong Buffer structure, overcome algorithm steps complexity in original technology, the shortcoming that detection time is long, so that detection time of the present invention is fast, real-time, has good actual application value.
Description of drawings
Fig. 1 is the block scheme of system of the present invention;
Fig. 2 is the process flow diagram of the inventive method.
Embodiment
Below in conjunction with Fig. 1 system of the present invention is further described.
System of the present invention comprises video acquisition decoder module, digital signal processor DSP module, extension storage module and acousto-optic alarm module.
The video acquisition decoder module, become with the decoding chipset by charge coupled cell CCD, the resolution of charge coupled cell CCD can be selected according to actual needs, be responsible for gathering vehicle front road conditions video analog signal, decoding chip sends to the digital signal processor DSP module with each frame of digital view data after being responsible for analog video signal is digitized as Digital Image Data again.
The digital signal processor DSP module comprises parallel peripheral bus PPI interface, expansion bus interface unit EBIU and universal input output GPIO interface.Wherein parallel peripheral bus PPI interface is connected with the video acquisition decoder module, and expansion bus interface unit EBIU is connected with the extension storage module, and universal input output GPIO interface is connected with the sound and light alarm module.The digital signal processor DSP module is responsible for transmission of digital view data, carries out image sorting algorithm and control sound and light alarm.
The extension storage module comprises synchronous DRAM SDRAM and expansion interface, be responsible for the Digital Image Data that temporary video acquisition decoder module obtains, and storage will be identified the clarification of objective training set.
The sound and light alarm module comprises audible alarm and light warning two parts, is responsible for sounding ringing and reports to the police and light flicker warning, reminds human pilot to note the target that vehicle front occurs.
Below in conjunction with Fig. 2 the inventive method is further described.
Step 1. is obtained video
After system starts, charge coupled cell CCD in the video acquisition decoder module begins to gather analog video signal, and signal sent into decoding chip, by decoding chip analog video signal is converted into and meets the Digital Image Data that international telecommunication is organized the ITU-656 form.
Step 2. transmit image data
The digital signal processor DSP module, at first direct memory access DMA is set to PPI to EBIU passage, based on the two-dimensional model of descriptor; Receive the Digital Image Data of video acquisition decoder module when the digital signal processor DSP module after, direct memory access DMA is transferred to expansion bus interface unit EBIU automatically with Digital Image Data from parallel peripheral bus PPI interface.
Step 3. image data temporary
The extension storage module according to the shared storage size of every two field picture, marks off the storage area that can hold two two field pictures with synchronous DRAM SDRAM, as Ping-Pong Buffer zone; Synchronous DRAM SDRAM from the expansion bus interface unit EBIU of digital signal processor DSP module, receives Digital Image Data; According to the parity frame order, circulation deposits in the Ping-Pong Buffer zone, whenever deposits a frame in the Digital Image Data that receives, sends a signal to digital signal processor DSP.
The initialization of step 4. count value
Digital signal processor DSP respectively picture count value i and successive objective count value k is set to 0; Picture count value i is used for the signal that synchronous DRAM SDRAM sends is counted, and judges whether needs reads image data from synchronous DRAM SDRAM according to this value; Successive objective count value k is used for the image that is classified as continuously target is counted, and judges whether that according to this value needs give the alarm.
Step 5. reads a two field picture
Digital signal processor DSP whenever receives synchronous DRAM SDRAM and sends signal one time, and picture count value i adds 1, until during picture count value i=3, digital signal processor DSP reads a frame image data that has just deposited in from the extension storage module.
Step 6. Images Classification
Digital signal processor DSP uses the machine learning classification algorithm, comprises two parts of off-line training and online classification, and a two field picture that reads is classified.
The first step, collection n frame training image, wherein, n＞100;
Second step, capable 8 row of n are set in the extension storage module matrix as the features training collection, the corresponding two field picture of every delegation of features training collection; Calculate 7 Hu moment characteristics values of every two field picture, concentrate front 7 elements of corresponding row as features training; Every two field picture is judged manually if target is arranged in the image, then the 8th element of the concentrated corresponding row of features training is 1, otherwise it is 0 that features training is concentrated the 8th element of corresponding row; The off-line that obtains of features training collection is finished, and is stored in the expansion interface part of extension storage module;
The 3rd step, for a two field picture that reads from the extension storage module, calculates 7 Hu moment characteristics values, form vector that 1 row 7 is listed as test sample book;
The 4th the step, with features training collection and test sample book, input together the k nearest neighbor sorter, if the k nearest neighbor sorter is categorized as 1 with test sample book, then there is intended target in view data corresponding to this test sample book, otherwise, if the k nearest neighbor sorter is categorized as 0 with test sample book, then there is not intended target in view data corresponding to this test sample book.
Whether step 7. image has target
According to the machine learning classification algorithm, if this two field picture is classified as the driftlessness class, change execution in step 4 over to, restart counting; Otherwise, change execution in step 8 over to.Picture count value i counts again since 0, otherwise, read continuously three two field pictures and classify; If three two field pictures all have been classified as target, then give the alarm, otherwise, change execution in step 5 over to.
Whether step 8. reports to the police
Successive objective count value k adds 1; If k=3 namely has continuous 3 two field pictures to be classified as target, change execution in step (9) over to, otherwise picture count value i=2 changes execution in step 5 over to;
Step 9. gives the alarm
Digital signal processor DSP sends alerting signal to the sound and light alarm module, and audible alarm partly sends the sound of ringing, and light is reported to the police and partly sent highlighted light flash, and reminding driver is noted.
Effect of the present invention by experiment result further embodies.
Experiment condition of the present invention is, builds the open highway scene of simulation, uses animal model as intended target, utilizes toy car to carry the present invention and carries out simulated experiment, and animal model different distance place before car is crossed along the direction of vertical car.After 300 experiments, analyze the results show, the present invention is 85% to the detection accuracy of moving target, and false-alarm probability is 13%, and false dismissal probability is 2%, and detection speed can reach more than frame p.s.s 30.Experimental result is analyzed, can be found out, the present invention has the real-time height, detects effective advantage, can satisfy actual vehicular applications.
Claims (3)
1. the vehicle-mounted real time kinematics object detection system based on DSP comprises video acquisition decoder module, digital signal processor DSP module, extension storage module and acousto-optic alarm module; The video acquisition decoder module is connected with the parallel peripheral bus PPI interface of digital signal processor DSP module, the extension storage module is connected with the expansion bus interface unit EBIU of digital signal processor DSP module, and the sound and light alarm module is connected with the universal input output GPIO interface of digital signal processor DSP module; Wherein:
Described video acquisition decoder module comprises charge coupled cell CCD camera and decoding chip, is used for collection and the digitizing of real time video image;
Described digital signal processor DSP module has direct memory access DMA function, is used for carrying out detection algorithm image is detected, and whether analyze wherein has intended target;
Described extension storage module comprises synchronous DRAM SDRAM, be used for the temporary view data that collects, and storage will be identified the clarification of objective training set;
Described sound and light alarm module is used for sounding and light alarm.
2. vehicle-mounted method for detecting real-time moving object based on DSP, its concrete steps are as follows:
(1) obtains video
1a) the analog video signal of charge coupled cell CCD camera collection vehicle front;
1b) the decoding chip in the video acquisition decoder module, analog video signal is converted into Digital Image Data after, again each frame of digital view data is sent to the digital signal processor DSP module;
(2) transmit image data
2a) digital signal processor DSP module, direct memory access DMA is set to PPI to EBIU passage, based on the two-dimensional model of descriptor;
2b) the digital signal processor DSP module gathers the Digital Image Data that decoder module sends by parallel peripheral bus PPI interface receiver, video;
2c) direct memory access DMA is transferred to expansion bus interface unit EBIU automatically with Digital Image Data from parallel peripheral bus PPI interface;
(3) image data temporary
3a) the extension storage module according to the shared storage size of every two field picture, marks off the storage area that can hold two two field pictures with synchronous DRAM SDRAM, as the Ping-PongBuffer zone;
3b) synchronous DRAM SDRAM from the expansion bus interface unit EBIU of digital signal processor DSP module, receives Digital Image Data;
3c) synchronous DRAM SDRAM, according to the parity frame order, circulation deposits in the Ping-Pong Buffer zone, whenever deposits a frame in the Digital Image Data that receives, sends a signal to digital signal processor DSP;
(4) count value initialization
Digital signal processor DSP respectively picture count value i and successive objective count value k is set to 0;
(5) read a two field picture
5a) when digital signal processor DSP received the signal that synchronous DRAM SDRAM sends, picture count value i added 1;
If 5b) picture count value i=3 changes execution in step 5a over to), otherwise, change execution in step 5c over to);
5c) digital signal processor DSP reads a frame image data that has just deposited in from the extension storage module;
(6) Images Classification
Digital signal processor DSP uses the machine learning classification algorithm, and a two field picture that reads is classified;
(7) whether image has target
If this two field picture is classified as the driftlessness class, change execution in step (4) over to; Otherwise, change execution in step (8) over to;
(8) whether report to the police
Successive objective count value k adds 1; If k=3 changes execution in step (9) over to, otherwise picture count value i=2 changes execution in step (5) over to;
(9) send warning
Digital signal processor DSP sends alerting signal to the sound and light alarm module, and the sound and light alarm module is sent ring sound and highlighted blinking light.
3. the vehicle-mounted method for detecting real-time moving object based on DSP according to claim 2 is characterized in that, the concrete steps of the machine learning classification algorithm described in the step (6) are as follows:
The first step, collection n frame training image, wherein, n＞100;
Second step, capable 8 row of n are set in the extension storage module matrix as the features training collection, the corresponding two field picture of every delegation of features training collection; Calculate 7 Hu moment characteristics values of every two field picture, concentrate front 7 elements of corresponding row as features training; Every two field picture is judged manually if target is arranged in the image, then the 8th element of the concentrated corresponding row of features training is 1, otherwise it is 0 that features training is concentrated the 8th element of corresponding row;
The 3rd step, for a two field picture that reads from the extension storage module, calculates 7 Hu moment characteristics values, form vector that 1 row 7 is listed as test sample book;
The 4th the step, with features training collection and test sample book, input together the k nearest neighbor sorter, if the k nearest neighbor sorter is categorized as 1 with test sample book, then there is intended target in view data corresponding to this test sample book, otherwise, if the k nearest neighbor sorter is categorized as 0 with test sample book, then there is not intended target in view data corresponding to this test sample book.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310031758.7A CN103077393B (en) | 2013-01-08 | 2013-01-08 | Based on vehicle-mounted real time kinematics object detection system and the method thereof of DSP |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310031758.7A CN103077393B (en) | 2013-01-08 | 2013-01-08 | Based on vehicle-mounted real time kinematics object detection system and the method thereof of DSP |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103077393A true CN103077393A (en) | 2013-05-01 |
| CN103077393B CN103077393B (en) | 2016-01-13 |
Family
ID=48153917
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310031758.7A Active CN103077393B (en) | 2013-01-08 | 2013-01-08 | Based on vehicle-mounted real time kinematics object detection system and the method thereof of DSP |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103077393B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103522970A (en) * | 2013-05-31 | 2014-01-22 | Tcl集团股份有限公司 | Vehicle driving safety detection method and system based on machine vision |
| CN104079881A (en) * | 2014-07-01 | 2014-10-01 | 中怡(苏州)科技有限公司 | Monitoring device and monitoring method related to monitoring device |
| DE102017201796A1 (en) | 2017-02-06 | 2018-08-09 | Audi Ag | Control device for determining a self-motion of a motor vehicle and motor vehicle and method for providing the control device |
| WO2021252873A1 (en) * | 2020-06-11 | 2021-12-16 | Tektronix, Inc. | System and method for separation and classification of signals using cyclic loop images |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101330606A (en) * | 2008-07-30 | 2008-12-24 | 浙江大学 | Built-in intelligent monitoring system for automatically detecting movement goal |
| CN101714213A (en) * | 2009-11-27 | 2010-05-26 | 中国科学技术大学 | Vehicle-mounted pedestrian detection method and system |
-
2013
- 2013-01-08 CN CN201310031758.7A patent/CN103077393B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101330606A (en) * | 2008-07-30 | 2008-12-24 | 浙江大学 | Built-in intelligent monitoring system for automatically detecting movement goal |
| CN101714213A (en) * | 2009-11-27 | 2010-05-26 | 中国科学技术大学 | Vehicle-mounted pedestrian detection method and system |
Non-Patent Citations (1)
| Title |
|---|
| 郑楚锋: "基于ADSP-BF533的运动目标检测及其在车载智能系统中的应用研究", 《中国优秀硕士学位论文全文数据库（电子期刊）》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103522970A (en) * | 2013-05-31 | 2014-01-22 | Tcl集团股份有限公司 | Vehicle driving safety detection method and system based on machine vision |
| CN103522970B (en) * | 2013-05-31 | 2016-04-27 | 深圳Tcl工业研究院有限公司 | Based on vehicle driving safety method of inspection and the system of machine vision |
| CN104079881A (en) * | 2014-07-01 | 2014-10-01 | 中怡(苏州)科技有限公司 | Monitoring device and monitoring method related to monitoring device |
| CN104079881B (en) * | 2014-07-01 | 2017-09-12 | 中磊电子（苏州）有限公司 | The relative monitoring method of supervising device |
| DE102017201796A1 (en) | 2017-02-06 | 2018-08-09 | Audi Ag | Control device for determining a self-motion of a motor vehicle and motor vehicle and method for providing the control device |
| WO2021252873A1 (en) * | 2020-06-11 | 2021-12-16 | Tektronix, Inc. | System and method for separation and classification of signals using cyclic loop images |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103077393B (en) | 2016-01-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102765365B (en) | Pedestrian detection method based on machine vision and pedestrian anti-collision warning system based on machine vision | |
| CN103366565B (en) | Method and system of detecting pedestrian running red light based on Kinect | |
| CN110738150B (en) | Camera linkage snapshot method and device and computer storage medium | |
| CN103832433A (en) | Lane departure and front collision warning system and achieving method thereof | |
| US20120212617A1 (en) | System and method for detecting traffic violations on restricted roadways | |
| CN103021175A (en) | Pedestrian red light running video detection method and device based on Davinci architecture | |
| US12045709B2 (en) | Vehicular driving assist system with enhanced data processing | |
| CN102303563B (en) | System and method for prewarning front vehicle collision | |
| CN101795397B (en) | Infrared imaging method for detecting passengers in vehicle | |
| US11620522B2 (en) | Vehicular system for testing performance of headlamp detection systems | |
| US11745749B2 (en) | Vehicular system for testing performance of object detection algorithms | |
| CN109977908B (en) | Vehicle driving lane detection method based on deep learning | |
| CN103077393B (en) | Based on vehicle-mounted real time kinematics object detection system and the method thereof of DSP | |
| CN109874099B (en) | Networking vehicle-mounted equipment flow control system | |
| RU2016121860A (en) | VEHICLE SYSTEM (OPTIONS) AND METHOD | |
| US11557133B1 (en) | Automatic license plate recognition | |
| CN104658261A (en) | Detection and monitoring system of cart lane occupation on expressway | |
| CN117372979A (en) | Road inspection method, device, electronic equipment and storage medium | |
| CN114119955A (en) | Method and device for detecting potential dangerous target | |
| Carrell et al. | Identification of driver phone usage violations via state-of-the-art object detection with tracking | |
| CN102840853A (en) | Obstacle detection and alarm method for vehicle-mounted night vision system | |
| CN112466121A (en) | Speed measuring method based on video | |
| Kulhandjian et al. | Pedestrian Detection and Avoidance at Night Using Multiple Sensors and Machine Learning | |
| US20160217581A1 (en) | Driving condition identification system | |
| CN202887456U (en) | Traffic flow video detection device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |