CN104898657A - Robot visual sense path identification method based on DSP - Google Patents
Robot visual sense path identification method based on DSP Download PDFInfo
- Publication number
- CN104898657A CN104898657A CN201410085265.6A CN201410085265A CN104898657A CN 104898657 A CN104898657 A CN 104898657A CN 201410085265 A CN201410085265 A CN 201410085265A CN 104898657 A CN104898657 A CN 104898657A
- Authority
- CN
- China
- Prior art keywords
- dsp
- image
- road
- fork
- path identification
- 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
Abstract
The invention discloses a robot visual sense path identification method based on DSP, which is characterized in that an image photographed by a CCD camera is encoded as digital information through a DSP development board and is stored in a memory, and that a DSP chip obtains image information to process. The processing method adopts C++ language to program and comprises steps of image denoising, image thresholding, road fork detection, road edge detection and real time display of detection result. The robot visual sense path identification method based on DSP can quickly detect the road fork and the road edge and can display the fork direction prompt and the road edge prompt on real time. The robot visual sense path identification method based on DSP is simple, improves the real-time processing efficiency of the robot, can be implanted into any robot path identification system and can be widely generalized.
Description
Technical field
The invention belongs to robot vision field, relate to a kind of path identification method, particularly relate to a kind of robot vision path identification method based on DSP.
Background technology
Along with the fast development of science and technology, Robotics is maked rapid progress.In Robotics, robot vision occupies very important position, and it act as the effect of robot eyes.Airmanship is also an aspect very important in Robotics, utilizes robot vision to navigate, can reduce costs, and improves treatment effeciency.
The DSP process image chip (abbreviation of Digital Signal Processor, i.e. digital signal processor) inner structure of high-speed digital video camera is adapted to it, irreplaceable effect has been played in image processing field, especially processing in video flowing in real time, there is superior performance.Code Composer Studio
tM(CCS) be that (Integrated Development Environment (IDE) of TD flush bonding processor series, the i.e. development environment of the embedded series of DSP, author language is C++ for Texas Instrument.Be that the embedded system of process chip is not only convenient and swift but also time saving and energy saving with DSP with CCS software development.
At present in path identification method, mainly based on static images analysis.Path Recognition general flow is: image denoising, morphological analysis, image threshold (binaryzation), extracts character pixel value, and mathematical method calculates matching navigation straight line (or curve), carries out error analysis.Because picture self-information amount is comparatively large, needs to carry out a large amount of mathematical computations during process picture, and take larger memory headroom, so general disposal route will take a long time, be unfavorable for the process of real-time.So in the real-time process of dynamic video stream, simple method is also needed to complete the process of robot path identification fast.
Summary of the invention
For above-mentioned problem, the object of the present invention is to provide a kind of robot vision path identification method based on DSP, be a kind of detection method of video flowing, road edge and fork road situation can be detected fast, carry out the real-time results display of video flowing.
Principle of work of the present invention is as follows:
When detecting beginning, CCD camera extracts the first frame pavement image, and DSP development board is decoded, and store with hexadecimal format, then dsp chip can process image, and the image processed through the display of DSP development board coding on a display screen.The method utilizes C Plus Plus to programme, and idiographic flow is as follows:
(1) Y (gray-scale value) component of image YUV color space is extracted;
(2) 4 × 4 medium filtering image denoisings;
(3) be converted to bianry image, getting segmentation threshold is 161;
(4) fork surveyed area is determined;
(5) count threshold is the pixel of 255, and divided by this area pixel point sum, whether result is greater than 0.8, is greater than 0.8, is 40 (grey), is less than 0.8 to the pixel grey scale assignment of fork identified areas, then not assignment;
(6) road-edge detection region is determined;
(7) count threshold is the pixel of 0, and divided by this area pixel point sum, whether result is greater than 0.8, is greater than 0.8, and it is 40 (grey) that the grey scale pixel value of road edge identified areas is composed, and is less than 0.8, then not assignment;
Image information after process is shown in LCD screen through DSP development board coding, then carries out the detection of next frame.
Concrete processing procedure is as follows:
First denoising is carried out and thresholding to path image, become bianry image, be i.e. the image of black and white two kinds of colors.Through test of many times under general illumination, getting segmentation thresholds is 161, after segmentation, and the threshold value of pavement image is 0, and the image threshold beyond road edge and road surface is 255, and with a lot of noise spot.Then according to pavement structure characteristic, first fork road can appear at the top of image, so it is surveyed area that position suitable above image extracts two regions, left and right respectively, and marks off mark warning region, for display.When fork occurs, in image Threshold segmentation divisible go out road surface, fork, in monitored area, simultaneously statistical threshold is the number of pixels of 255 (being judged to be the pixel on road surface), then divided by the total pixel number of monitored area, if ratio is more than 0.8, display screen marks direction, fork immediately.
Then according to the position that camera is placed, two regions are got in portion respectively in the picture, as the surveyed area of left and right road edge, when robot is near road edge, the method similar to detecting fork, surveyed area starts the number of pixels that statistical threshold is 0 (being judged to be the pixel on non-road surface), then divided by the total pixel number of monitored area, if ratio is more than 0.8, display screen can mark symbol warning road edge.
Each two field picture carries out same detection method, and detection speed can reach about 25 frames/second.
Compared with the conventional method, the present invention has the following advantages:
1) method used in the present invention is simple, greatly improves chip travelling speed, meets real time handling requirement.
2) the method can detect fork road and road edge fast, then shows the prompting of direction, fork in real time and points out with road edge.
3) the method programming is comparatively simple, and relatively independent, can be transplanted in any robot path recognition system.
Accompanying drawing explanation
Fig. 1 is a kind of pixel coordinate system used based on the robot vision path identification method of DSP of the present invention;
Fig. 2 is the image processing flow figure of a kind of robot vision path identification method based on DSP of the present invention;
Fig. 3 is the fork road detection method process flow diagram of a kind of robot vision path identification method based on DSP of the present invention;
Fig. 4 is the road edge detection method process flow diagram of a kind of robot vision path identification method based on DSP of the present invention;
Fig. 5 a is the fork road example of a kind of robot vision path identification method based on DSP of the present invention;
Fig. 5 b is the fork Detection results figure of a kind of robot vision path identification method based on DSP of the present invention;
Fig. 6 a is the road edge example of a kind of robot vision path identification method based on DSP of the present invention;
Fig. 6 b is the road binary image of a kind of robot vision path identification method based on DSP of the present invention;
Fig. 6 c is the road-edge detection design sketch of a kind of robot vision path identification method based on DSP of the present invention.
Embodiment
Below in conjunction with accompanying drawing, the invention will be further described.
When detecting beginning, camera extracts pavement image, and DSP development board is decoded, and store with hexadecimal format, then dsp chip can process image, and the image processed through the display of DSP development board coding on a display screen.Concrete processing procedure is as follows:
What Fig. 1 represented is pixel space coordinate system, has following digitized image function according to this figure:
In this invention: M=576, N=720.
The operation workflow of this invention as shown in Figure 2, when detecting beginning, camera extracts a frame pavement image, acquisition image respectively rate is 720 × 576, store with hexadecimal format, then binaryzation is carried out to image, and then carry out fork detection and road-edge detection, by testing result display on a display screen, the process of next frame image is carried out after having processed again.
For the road of Fig. 6 a, first to path image denoising, then carry out thresholding, become bianry image, as Fig. 6 b, i.e. the image of black and white two kinds of colors.Under general illumination, get segmentation threshold after test of many times is 161, after segmentation, and the threshold value of pavement image is 255, and beyond road edge and road surface, the image threshold of part is 0.F
0(x, y) is the image function after binaryzation:
For the road of Fig. 5 a, according to pavement structure characteristic, first fork road can appear at the top of image, so it is surveyed area that position suitable above image extracts two regions, left and right respectively, selection pixel detection region, the left side is: x=0 to x=60, y=0 to y=100, the right is: x=0 to x=60, y=600 to y=720.The region selected can change according to road conditions difference.
Fork testing process as shown in Figure 3, when fork occurs, in image, threshold division goes out road surface, fork, statistical threshold f while of in surveyed area
0(x, y) is the number of pixels n of 255 (being judged to be the pixel on road surface), and after this region detection, with the total pixel number m of the n value of adding up divided by monitored area, now m=100 × 60=6000, if ratio μ
1more than μ
0=0.8, be 40 (grey) to the pixel grey scale assignment of fork identified areas, if be less than 0.8, then not assignment.Formula is as follows:
μ
1=n/m
Fig. 5 b is fork road Detection results figure, when camera photographs the fork on the right, can see and " → " that image upper right side occurs iris out with black line circle.
Road-edge detection is for Fig. 6 a, and according to the position that camera is placed, two regions are got in portion respectively in the picture, as the surveyed area of left and right road edge.Shown in Fig. 6 b, monitored area is arranged in figure black wire frame: left side surveyed area is: x=200 to x=280, y=100 to y=140, the right surveyed area: x=200 to x=280, y=580 to y=620.The region selected can change according to road conditions difference.
When camera is near road edge, as shown in Figure 4, the method similar to detecting fork, surveyed area starts the number of pixels n that statistical threshold is 0 (being judged to be the pixel on non-road surface), detect the total pixel number m of n divided by monitored area of rear statistics, now m=80 × 40=3200, if μ
2more than μ
0=0.8, display screen can mark symbol warning road edge.Formula is as follows:
μ
2=n/m
Fig. 6 c is road-edge detection design sketch, when camera photographs left side edge, can see there is inverted "L" shaped prompting mark on the left of black line frame, iris out with white line circle.
According to above-mentioned description, relevant staff in the scope not departing from this invention technological thought, can carry out various change and amendment completely.The technical scope of this invention is not limited to the content on instructions, and all any amendments done within the spirit and principles in the present invention, equivalent replacement and improvement etc., all should be included within protection scope of the present invention.
Claims (1)
1. the robot vision path identification method based on DSP, wherein: the image taken by CCD camera is decoded as numerical information through DSP development board and is stored in storer, then obtain image information by dsp chip to process, this disposal route C Plus Plus is programmed, and feature is as follows:
(1) Y (gray-scale value) component of image YUV color space is extracted;
(2) 4 × 4 medium filtering image denoisings;
(3) be converted to bianry image, getting segmentation threshold is 161;
(4) fork surveyed area is determined;
(5) count threshold is the pixel of 255, and divided by this area pixel point sum, whether result is greater than 0.8, is greater than 0.8, is 40 (grey), is less than 0.8 to the pixel grey scale of fork identified areas, then not assignment;
(6) road-edge detection region is determined;
(7) count threshold is the pixel of 0, and divided by this area pixel point sum, whether result is greater than 0.8, is greater than 0.8, and it is 40 (grey) that the grey scale pixel value of road edge identified areas is composed, and is less than 0.8, then not assignment;
Image information after process is shown in LCD screen through DSP development board coding, then carries out the detection of next frame.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410085265.6A CN104898657A (en) | 2014-03-06 | 2014-03-06 | Robot visual sense path identification method based on DSP |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410085265.6A CN104898657A (en) | 2014-03-06 | 2014-03-06 | Robot visual sense path identification method based on DSP |
Publications (1)
Publication Number | Publication Date |
---|---|
CN104898657A true CN104898657A (en) | 2015-09-09 |
Family
ID=54031369
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410085265.6A Pending CN104898657A (en) | 2014-03-06 | 2014-03-06 | Robot visual sense path identification method based on DSP |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104898657A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113204236A (en) * | 2021-04-14 | 2021-08-03 | 华中科技大学 | Intelligent agent path tracking control method |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20030026456A1 (en) * | 2001-08-06 | 2003-02-06 | Nissan Motor Co., Ltd. | White line detection apparatus and white line detection method |
CN101339616A (en) * | 2008-08-12 | 2009-01-07 | 北京中星微电子有限公司 | Roads recognition method and apparatus |
CN102541063A (en) * | 2012-03-26 | 2012-07-04 | 重庆邮电大学 | Line tracking control method and line tracking control device for micro intelligent automobiles |
CN202454087U (en) * | 2012-02-27 | 2012-09-26 | 上海理工大学 | Vehicle-mounted road edge detecting device |
US20130173232A1 (en) * | 2010-04-20 | 2013-07-04 | Conti Temic Microelectronic Gmbh | Method for determining the course of the road for a motor vehicle |
CN103389733A (en) * | 2013-08-02 | 2013-11-13 | 重庆市科学技术研究院 | Vehicle line walking method and system based on machine vision |
CN103472824A (en) * | 2013-08-29 | 2013-12-25 | 河海大学 | Camera-based navigation system and method for automatic navigation vehicle |
-
2014
- 2014-03-06 CN CN201410085265.6A patent/CN104898657A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20030026456A1 (en) * | 2001-08-06 | 2003-02-06 | Nissan Motor Co., Ltd. | White line detection apparatus and white line detection method |
CN101339616A (en) * | 2008-08-12 | 2009-01-07 | 北京中星微电子有限公司 | Roads recognition method and apparatus |
US20130173232A1 (en) * | 2010-04-20 | 2013-07-04 | Conti Temic Microelectronic Gmbh | Method for determining the course of the road for a motor vehicle |
CN202454087U (en) * | 2012-02-27 | 2012-09-26 | 上海理工大学 | Vehicle-mounted road edge detecting device |
CN102541063A (en) * | 2012-03-26 | 2012-07-04 | 重庆邮电大学 | Line tracking control method and line tracking control device for micro intelligent automobiles |
CN103389733A (en) * | 2013-08-02 | 2013-11-13 | 重庆市科学技术研究院 | Vehicle line walking method and system based on machine vision |
CN103472824A (en) * | 2013-08-29 | 2013-12-25 | 河海大学 | Camera-based navigation system and method for automatic navigation vehicle |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113204236A (en) * | 2021-04-14 | 2021-08-03 | 华中科技大学 | Intelligent agent path tracking control method |
CN113204236B (en) * | 2021-04-14 | 2022-05-20 | 华中科技大学 | Intelligent agent path tracking control method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021051604A1 (en) | Method for identifying text region of osd, and device and storage medium | |
CN103914843B (en) | The image partition method marked based on watershed algorithm and morphology | |
CN111191611B (en) | Traffic sign label identification method based on deep learning | |
CN104361353A (en) | Application of area-of-interest positioning method to instrument monitoring identification | |
CN105447489B (en) | A kind of character of picture OCR identifying system and background adhesion noise cancellation method | |
CN103093229B (en) | Car is demarcated method for position and device | |
CN104598907A (en) | Stroke width figure based method for extracting Chinese character data from image | |
CN102609702A (en) | Method and system for rapidly positioning road guide sign | |
CN106327464A (en) | Edge detection method | |
CN103914829B (en) | Method for detecting edge of noisy image | |
CN103793923A (en) | Method and device for acquiring moving object in image | |
CN106778736A (en) | The licence plate recognition method and its system of a kind of robust | |
CN114419006A (en) | Method and system for removing watermark of gray level video characters changing along with background | |
CN104732530A (en) | Image edge detection method | |
CN103971347A (en) | Method and device for treating shadow in video image | |
CN104898657A (en) | Robot visual sense path identification method based on DSP | |
CN115526881B (en) | Battery cell polarity detection method and device based on image modeling | |
ATE418117T1 (en) | SEGMENTATION OF AN IMAGE USING A GRAPH | |
CN107220943A (en) | The ship shadow removal method of integration region texture gradient | |
CN114529715B (en) | Image identification method and system based on edge extraction | |
CN104766331A (en) | Imaging processing method and electronic device | |
CN106951831B (en) | Pedestrian detection tracking method based on depth camera | |
CN101408977A (en) | Method and apparatus for dividing candidate barrier region | |
JP2009277001A (en) | Image processor | |
CN112330659B (en) | Geometric tolerance symbol segmentation method combining LSD (least squares) linear detection and connected domain marking method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150909 |