CN107610152A - A kind of path detection method for being used to hide the water surface and barrier - Google Patents
A kind of path detection method for being used to hide the water surface and barrier Download PDFInfo
- Publication number
- CN107610152A CN107610152A CN201710771679.8A CN201710771679A CN107610152A CN 107610152 A CN107610152 A CN 107610152A CN 201710771679 A CN201710771679 A CN 201710771679A CN 107610152 A CN107610152 A CN 107610152A
- Authority
- CN
- China
- Prior art keywords
- depth
- height
- infrared
- value
- map
- 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
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a kind of path detection method for being used to hide the water surface and barrier, for guiding blind person to go to the passage region of no barrier and the water surface.This method is realized by the hardware such as the red speckle projector, two infrared cameras, attitude angle transducers.The present invention has been significantly increased blind person's avoiding obstacles with the water surface and has selected the security and convenience in current direction, the water surface and barrier zone are avoided in trip beneficial to blind person, correct path is found, there is good environmental adaptability can be used under motion conditions, the advantages of learning cost is low.
Description
Technical field
It is used to hide the path detection method of the water surface and barrier the present invention relates to a kind of,.
Background technology
According to the data of the World Health Organization, the whole world there are about 2.85 hundred million visually impaired peoples, wherein blind person up to 39,000,000.It is blind
People can not effectively distinguish the landform in front due to lacking normal vision, and predicting barrier and the water surface and selection can pass through road
Line.Depth transducer based on laser speckle coding or binocular vision technology, the color data stream and depth of high-speed can be obtained
Degrees of data stream, therefore sprawled using applying of being detected of these sensors in blind visual field of auxiliary.
However, the depth transducer based on laser speckle coding, depth letter is calculated according to the change in location for receiving speckle
Breath, because speckle is flooded in outdoor by sunlight, depth information can not be calculated according to the change of speckle, therefore in outdoor failure.And
Depth transducer based on binocular vision technology, carry out images match according to two images and calculate depth information, but indoors
White wall that many textures lack can not calculate depth information at floor, therefore generally in the outside work of texture-rich.
The content of the invention
In view of the above-mentioned deficiencies in the prior art, it is an object of the present invention to provide a kind of path for being used to hide the water surface and barrier
Detection method.
The purpose of the present invention is achieved through the following technical solutions:A kind of path inspection for being used to hide the water surface and barrier
Survey method, comprises the following steps:
(1) the infrared speckle of scene simulation forwards;Scene under infrared projection is infrared by two after linear polarizer filters
Camera gathers, and obtains two infrared images respectively;The polarization state of the image of two camera collections is vertical, the polarization state of one of them
For horizontal direction, another is vertical direction, and the angle of pitch of two infrared cameras is a, roll angle b;
(2) binocular solid matching is carried out to two width infrared images, obtains a width disparity map Disparity;Utilize disparity map
Disparity, the corresponding points (u', v) corresponding to any point (u, v) in left infrared figure are found in right view, meet u'-u
=disparity, disparity are the parallax value of pixel (u, v) in disparity map Disparity;Then by left infrared figure
Any point (u, v) infrared gray value IRL(u,v)With the infrared gray value IR of corresponding points (u', v) in right infrared figureL(u',v)Do
Difference, polarization differential figure Polarization is obtained, in wherein polarization differential figure Polarization, the value of pixel (u, v) is (partially
Shake difference value polarization) be | IRL(u,v)-IRL(u',v)|;
(3) disparity map Disparity is utilized, obtains pixel (u, v) in width a depth map Depth, depth map Depth
Depth value depth is depth=f × Baseline/disparity, wherein, f is the focal length of two infrared cameras, Baseline
For the parallax range between two infrared cameras;
(4) every bit (u, v) is calculated in depth map Depth relative to left infrared camera photocentre (u0,v0) three-dimensional coordinate
(x, y, z), so as to obtain a width height map Height;In height map Height, the height value height of pixel (u, v) is:
Height=x × sin (b)+y × cos (a) × cos (b)-z × sin (a) × cos (b);
Wherein x=(u-u0) × depth/f, y=(v-v0) × depth/f, z=depth;Depth is depth map Depth
The depth value of middle pixel (u, v);
(5) depth map Depth each row are traveled through, finds out and meets corresponding points in height map Height in each row
Height value CorrHeight is less than threshold value HeightThreshold minimum depth value MinDepthObs, while finds out each
Row meet that height value CorrHeight of the corresponding points in height map Height is more than threshold value HeightThreshold and polarized
Polarization differential value CorrPolarization in difference diagram Polarization is more than threshold value PolarizationThreshold
Minimum depth value MinDepthWat.Less value MinDepth in MinDepthObs and MinDepthWat is taken in each row,
Form the distance Curve MinTraversableLine that can pass through recently.
(6) the peak Peak that can pass through recently on distance Curve MinTraversableLine is searched, while from highest
Point Peak to two-sided search, search can pass through distance for peak half two nearest half height point HalfLeft with
HalfRight, it is to pass through relative to the direction of left infrared camera photocentre to take the midpoint Mid of two half height points, midpoint Mid
Direction.
Further, the threshold value HeightThreshold is equal to the height h of infrared camera;Threshold value
PolarizationThreshold is adjusted according to the maximum brightness of infrared image, generally the 0.75~0.8 of maximum brightness
Times.
Further, if the peak Peak that can be passed through recently on distance Curve MinTraversableLine in step 6
Have multiple, then search for half height point HalfLeft to the left from leftmost peak, searched for the right from the peak of rightmost
Half height point HalfRight.
The beneficial effects of the present invention are:
1. good environmental adaptability.The present invention obtains the three-dimensional information of environment by projecting infrared speckle, indoors or night
The region of texture missing, infrared speckle can strengthen texture, near in sunlight due to there is an irradiation of sunlight on outdoor or daytime
Infrared composition can also allow the infrared image collected to have abundant texture.Outdoor indoors, day and night, infrared camera all may be used
To obtain the infrared image of texture-rich, this is advantageous to processor and obtains dense three-dimensional information, so as to ensure blind person's reliever
Mirror has good environmental suitability.In addition, this method can detect the water surface and barrier simultaneously, therefore can also be adapted to fine
My god, the trip of the different weather such as rainy days, can help blind person's to avoid the danger zones such as the water surface, barrier.
2. it can be used under motion conditions.The attitude angle information of the infrared camera of collection in real time of the invention, so as to real-time
The actual height of every bit in scene is obtained, the ground in case of motion and barrier can be helped distinguish between, be adapted to blind person
Use.
3. learning cost is low.Patent of the present invention due to transmit every time to be that the most concerned current scene of blind person is most adapted to logical
Capable direction, blind person's avoiding obstacles are helped, guides them to go on correct path, the information of redundancy will not be transmitted, therefore
Learning cost is low.
Brief description of the drawings
Fig. 1 is depth map;
The face clear region and water-surface areas that Fig. 2 is represented;
Fig. 3 can pass through distance Curve figure recently, and the height of dotted line represents the distance of passing through of each row.
Embodiment
The invention will be further described below in conjunction with the accompanying drawings.
The present invention relates to a kind of path detection method for being used to hide the water surface and barrier, this method is thrown based on red speckle
The hardware such as emitter, two infrared cameras, polarizer, attitude angle transducers is realized.The wherein infrared speckle projector is used for forwards
The infrared speckle of scene simulation;Scene under infrared projection is gathered by two infrared cameras, obtained respectively after linear polarizer filters
Two infrared images;The angle of pitch that attitude angle transducer is used to obtain two infrared cameras in real time is a, roll angle b;Based on upper
Hardware is stated, detection method of the invention is as follows:
(1) the infrared speckle projector infrared speckle of scene simulation forwards;Scene under infrared projection is filtered through linear polarizer
After ripple, gathered by two infrared cameras, obtain two infrared images respectively;The polarization state of the image of two camera collections is vertical,
The polarization state of one of them is horizontal direction, and another is vertical direction, and the angle of pitch of two infrared cameras is a, and roll angle is
B, collected by attitude angle transducer;
(2) binocular solid matching is carried out to two width infrared images, obtains a width disparity map Disparity;(2) parallax is utilized
Scheme Disparity, the corresponding points (u', v) corresponding to any point (u, v) in left infrared figure are found in right view, meet u'-
U=disparity, disparity are the parallax value of pixel (u, v) in disparity map Disparity;Then by left infrared figure
Any point (u, v) infrared gray value IRL(u,v)With the infrared gray value IR of corresponding points (u', v) in right infrared figureL(u',v)Do
Difference, polarization differential figure Polarization is obtained, in wherein polarization differential figure Polarization, the value of pixel (u, v) is (partially
Shake difference value polarization) be | IRL(u,v)-IRL(u',v)|;
(3) disparity map Disparity is utilized, a width depth map Depth is obtained, as shown in figure 1, picture in depth map Depth
The depth value depth of vegetarian refreshments (u, v) is depth=f × Baseline/disparity, wherein, f is Jiao of two infrared cameras
Away from Baseline is the parallax range between two infrared cameras;
(4) every bit (u, v) is calculated in depth map Depth relative to left infrared camera photocentre (u0,v0) three-dimensional coordinate
(x, y, z), so as to obtain a width height map Height;In height map Height, the height value height of pixel (u, v) is:
Height=x × sin (b)+y × cos (a) × cos (b)-z × sin (a) × cos (b);
Wherein x=(u-u0) × depth/f, y=(v-v0) × depth/f, z=depth;Depth is depth map Depth
The depth value of middle pixel (u, v);
(5) depth map Depth each row are traveled through, finds out and meets corresponding points in height map Height in each row
Height value CorrHeight is less than threshold value HeightThreshold minimum depth value MinDepthObs, while finds out each
Row meet that height value CorrHeight of the corresponding points in height map Height is more than threshold value HeightThreshold and polarized
Polarization differential value CorrPolarization in difference diagram Polarization is more than threshold value PolarizationThreshold
Minimum depth value MinDepthWat.Less value MinDepth in MinDepthObs and MinDepthWat is taken in each row,
The distance Curve MinTraversableLine that can pass through recently is formed, as shown in Figure 3.As can be seen that passing through of detecting away from
From curve with being actually consistent.
(6) the peak Peak that can pass through recently on distance Curve MinTraversableLine is searched, while from highest
Point Peak to two-sided search, search can pass through distance for peak half two nearest half height point HalfLeft with
HalfRight, it is to pass through relative to the direction of left infrared camera photocentre to take the midpoint Mid of two half height points, midpoint Mid
Direction.
Above-mentioned threshold value HeightThreshold is equal to the height h of infrared camera;Threshold value PolarizationThreshold
It is adjusted according to the maximum brightness of infrared image, generally the 0.75~0.8 of maximum brightness times.
If the peak Peak that can be passed through recently on distance Curve MinTraversableLine in step 6 have it is multiple,
Half height point HalfLeft is searched for the left from leftmost peak, and half height point is searched for the right from the peak of rightmost
HalfRight。
Claims (3)
1. a kind of path detection method for being used to hide the water surface and barrier, comprises the following steps:
(1) the infrared speckle of scene simulation forwards;Scene under infrared projection is after linear polarizer filters, by two infrared cameras
Collection, two infrared images are obtained respectively;The polarization state of the image of two camera collections is vertical, and the polarization state of one of them is water
Square to another is vertical direction, and the angle of pitch of two infrared cameras is a, roll angle b;
(2) binocular solid matching is carried out to two width infrared images, obtains a width disparity map Disparity;Utilize disparity map
Disparity, the corresponding points (u', v) corresponding to any point (u, v) in left infrared figure are found in right view, meet u'-u
=disparity, disparity are the parallax value of pixel (u, v) in disparity map Disparity;Then by left infrared figure
Any point (u, v) infrared gray value IRL(u,v)With the infrared gray value IR of corresponding points (u', v) in right infrared figureL(u',v)Do
Difference, polarization differential figure Polarization is obtained, in wherein polarization differential figure Polarization, the value of pixel (u, v) is (partially
Shake difference value polarization) be | IRL(u,v)-IRL(u',v)|;
(3) disparity map Disparity is utilized, obtains the depth of pixel (u, v) in width a depth map Depth, depth map Depth
Value depth is depth=f × Baseline/disparity, wherein, f is the focal length of two infrared cameras, Baseline two
Parallax range between individual infrared camera;
(4) every bit (u, v) is calculated in depth map Depth relative to left infrared camera photocentre (u0,v0) three-dimensional coordinate (x, y,
Z), so as to obtaining a width height map Height;In height map Height, the height value height of pixel (u, v) is:
Height=x × sin (b)+y × cos (a) × cos (b)-z × sin (a) × cos (b);
Wherein x=(u-u0) × depth/f, y=(v-v0) × depth/f, z=depth;Depth is picture in depth map Depth
The depth value of vegetarian refreshments (u, v);
(5) depth map Depth each row are traveled through, finds out and meets height of the corresponding points in height map Height in each row
Value CorrHeight is less than threshold value HeightThreshold minimum depth value MinDepthObs, while it is full to find out each row
Height value CorrHeight of the sufficient corresponding points in height map Height is more than threshold value HeightThreshold and in polarization differential
The polarization differential value CorrPolarization schemed in Polarization is more than threshold value PolarizationThreshold most
Small depth value MinDepthWat.Less value MinDepth in MinDepthObs and MinDepthWat is taken in each row, is formed
Can pass through distance Curve MinTraversableLine recently.
(6) the peak Peak that can pass through recently on distance Curve MinTraversableLine is searched, while from peak
Peak to two-sided search, search can pass through distance for peak half two nearest half height point HalfLeft with
HalfRight, it is to pass through relative to the direction of left infrared camera photocentre to take the midpoint Mid of two half height points, midpoint Mid
Direction.
2. according to the method for claim 1, it is characterised in that the threshold value HeightThreshold is equal to infrared camera
Height h;Threshold value PolarizationThreshold is adjusted according to the maximum brightness of infrared image, generally most highlighted
0.75~0.8 times of degree.
3. according to the method for claim 1, it is characterised in that if distance Curve can be passed through in step 6 recently
Peak Peak on MinTraversableLine has multiple, then searches for half height point to the left from leftmost peak
HalfLeft, half height point HalfRight is searched for the right from the peak of rightmost.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710771679.8A CN107610152B (en) | 2017-08-31 | 2017-08-31 | Passage detection method for avoiding water surface and obstacles |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710771679.8A CN107610152B (en) | 2017-08-31 | 2017-08-31 | Passage detection method for avoiding water surface and obstacles |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107610152A true CN107610152A (en) | 2018-01-19 |
CN107610152B CN107610152B (en) | 2020-02-28 |
Family
ID=61057052
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710771679.8A Active CN107610152B (en) | 2017-08-31 | 2017-08-31 | Passage detection method for avoiding water surface and obstacles |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107610152B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109931946A (en) * | 2019-04-10 | 2019-06-25 | 福州大学 | Blind visual range-finding navigation method based on Android intelligent |
CN109998873A (en) * | 2019-04-12 | 2019-07-12 | 赖振龙 | A kind of wearable blindmen intelligent positioning and blind guiding system |
CN111310651A (en) * | 2020-02-13 | 2020-06-19 | 杭州凌像科技有限公司 | Water surface detection method based on polarization camera and RGB-D sensor |
CN111354027A (en) * | 2018-12-21 | 2020-06-30 | 沈阳新松机器人自动化股份有限公司 | Visual obstacle avoidance method for mobile robot |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101701828A (en) * | 2009-11-23 | 2010-05-05 | 常州达奇信息科技有限公司 | Blind autonomous navigation method based on stereoscopic vision and information fusion |
CN102313536A (en) * | 2011-07-21 | 2012-01-11 | 清华大学 | Method for barrier perception based on airborne binocular vision |
US20140160244A1 (en) * | 2010-09-21 | 2014-06-12 | Mobileye Technologies Limited | Monocular cued detection of three-dimensional structures from depth images |
CN106203390A (en) * | 2016-07-22 | 2016-12-07 | 杭州视氪科技有限公司 | A kind of intelligent blind aid system |
CN106228110A (en) * | 2016-07-07 | 2016-12-14 | 浙江零跑科技有限公司 | A kind of barrier based on vehicle-mounted binocular camera and drivable region detection method |
-
2017
- 2017-08-31 CN CN201710771679.8A patent/CN107610152B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101701828A (en) * | 2009-11-23 | 2010-05-05 | 常州达奇信息科技有限公司 | Blind autonomous navigation method based on stereoscopic vision and information fusion |
US20140160244A1 (en) * | 2010-09-21 | 2014-06-12 | Mobileye Technologies Limited | Monocular cued detection of three-dimensional structures from depth images |
CN102313536A (en) * | 2011-07-21 | 2012-01-11 | 清华大学 | Method for barrier perception based on airborne binocular vision |
CN106228110A (en) * | 2016-07-07 | 2016-12-14 | 浙江零跑科技有限公司 | A kind of barrier based on vehicle-mounted binocular camera and drivable region detection method |
CN106203390A (en) * | 2016-07-22 | 2016-12-07 | 杭州视氪科技有限公司 | A kind of intelligent blind aid system |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111354027A (en) * | 2018-12-21 | 2020-06-30 | 沈阳新松机器人自动化股份有限公司 | Visual obstacle avoidance method for mobile robot |
CN109931946A (en) * | 2019-04-10 | 2019-06-25 | 福州大学 | Blind visual range-finding navigation method based on Android intelligent |
CN109998873A (en) * | 2019-04-12 | 2019-07-12 | 赖振龙 | A kind of wearable blindmen intelligent positioning and blind guiding system |
CN111310651A (en) * | 2020-02-13 | 2020-06-19 | 杭州凌像科技有限公司 | Water surface detection method based on polarization camera and RGB-D sensor |
CN111310651B (en) * | 2020-02-13 | 2023-05-02 | 杭州凌像科技有限公司 | Water surface detection method based on polarization camera and RGB-D sensor |
Also Published As
Publication number | Publication date |
---|---|
CN107610152B (en) | 2020-02-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106681353B (en) | The unmanned plane barrier-avoiding method and system merged based on binocular vision with light stream | |
CN105225482B (en) | Vehicle detecting system and method based on binocular stereo vision | |
CN205905026U (en) | Robot system based on two mesh stereovisions | |
CN107610152A (en) | A kind of path detection method for being used to hide the water surface and barrier | |
Bertozzi et al. | Low-level pedestrian detection by means of visible and far infra-red tetra-vision | |
JP4956452B2 (en) | Vehicle environment recognition device | |
US10129521B2 (en) | Depth sensing method and system for autonomous vehicles | |
CN102750711B (en) | A kind of binocular video depth map calculating method based on Iamge Segmentation and estimation | |
CN106204443A (en) | A kind of panorama UAS based on the multiplexing of many mesh | |
CN101986717B (en) | Image data generating system for stereo display | |
CN105411490A (en) | Real-time positioning method of mobile robot and mobile robot | |
CN104331901A (en) | TLD-based multi-view target tracking device and method | |
CN101408978B (en) | Method and apparatus for detecting barrier based on monocular vision | |
CN101887589A (en) | Stereoscopic vision-based real low-texture image reconstruction method | |
CN108398139A (en) | A kind of dynamic environment visual odometry method of fusion fish eye images and depth image | |
WO2004008744A1 (en) | Road and other flat object video plan-view developing image processing method, reverse developing image conversion processing method, plan-view developing image processing device, and reverse developing image conversion processing device | |
CN106183995B (en) | A kind of visual parking device method based on stereoscopic vision | |
Rodríguez et al. | Obstacle avoidance system for assisting visually impaired people | |
CN108776989A (en) | Low texture plane scene reconstruction method based on sparse SLAM frames | |
CN106846375A (en) | A kind of flame detecting method for being applied to autonomous firefighting robot | |
CN108205315A (en) | A kind of robot automatic navigation method based on binocular vision | |
CN112115889B (en) | Intelligent vehicle moving target detection method based on vision | |
CN107341789A (en) | One kind is based on RGB D cameras and stereosonic visually impaired people's path precognition system and method | |
CN106846350A (en) | One kind is based on RGB D cameras and stereosonic visually impaired people's barrier early warning system and method | |
Yiruo et al. | Complex ground plane detection based on v-disparity map in off-road environment |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |