CN107255470A - Obstacle detector - Google Patents
Obstacle detector Download PDFInfo
- Publication number
- CN107255470A CN107255470A CN201710412721.7A CN201710412721A CN107255470A CN 107255470 A CN107255470 A CN 107255470A CN 201710412721 A CN201710412721 A CN 201710412721A CN 107255470 A CN107255470 A CN 107255470A
- Authority
- CN
- China
- Prior art keywords
- vertical edge
- detection
- class vertical
- module
- class
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C11/00—Photogrammetry or videogrammetry, e.g. stereogrammetry; Photographic surveying
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C11/00—Photogrammetry or videogrammetry, e.g. stereogrammetry; Photographic surveying
- G01C11/04—Interpretation of pictures
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C11/00—Photogrammetry or videogrammetry, e.g. stereogrammetry; Photographic surveying
- G01C11/36—Videogrammetry, i.e. electronic processing of video signals from a single source or from different sources to give parallax or range information
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention provides a kind of obstacle detector, including memory cell, class vertical edge detection module, intersection judge module, position judging module and detection of obstacles module.Class vertical edge detection module receives input picture and class vertical edge is carried out to this input picture and detects program and obtain multiple class vertical edges.Intersection judge module judges whether these class vertical edges intersect with default virtual ground horizontal line.If one of these class vertical edges intersect with virtual ground horizontal line, position judging module judges whether one end of one of the class vertical edge intersected with virtual ground horizontal line is located in detection zone.If one end of one of the class vertical edge intersected with virtual ground horizontal line is located in detection zone, detection of obstacles module judges to detect the presence of barrier and provides warning.
Description
The present invention is the Application No. 201410102894.5 proposed for 19th for 03 month in 2014, entitled《Obstacle
Analyte detection device》Application for a patent for invention divisional application.
Technical field
The invention relates to a kind of obstacle detector, and in particular to a kind of based on image processing techniques
Obstacle detector.
Background technology
As automobile quantity is growing, the probability for occurring road accident also increases year by year.It is apparent that vehicle science and technology except
Outside power section continuous advancement, the enhancement of security is also another problem for needing to pay close attention in driving conditions.If studies have shown that
If driver can obtain early warning before colliding, the probability of road accident generation can be greatly reduced.Therefore, it is good and accurate
Detection of obstacles and caution system an actually very important link in current automotive safety system.
In general, radar or ultrasonic system have been commonly used in automotive safety system now, but radar or Supersonic
Wave system system may reduce discrimination power because of that can not recognize environment, and the electromagnetic wave power of radar or ultrasonic system may also
Human body is impacted.On the other hand, with the progress of image procossing and photography detection technology, barrier is carried out using image
The mode of detection is also increasingly common.Many image-type safety warning systems remind driver by barrier in recognisable image
Safe distance is kept with it.However, for miscellaneous image processing and analysis mode, its each self-corresponding identification is accurate
System effectiveness performance all difference such as the time required to exactness or computing.Therefore, how one kind is provided based on image procossing mode
The obstacle detection system of high-accuracy actually one of those skilled in the art's subject under discussion of concern.
The content of the invention
Therefore, the present invention provides a kind of obstacle detector, can improve the standard by image detection and identification barrier
True rate, is warned to driver with time providing correct barrier.
The present invention proposes a kind of obstacle detector, and this obstacle detector includes memory cell, class vertical edge
Detection module, intersection judge module, position judging module and detection of obstacles module.Memory cell at least stores a figure
Picture.Class vertical edge detection module couples memory cell, receives input picture and carries out class vertical edge inspection to this input picture
Ranging sequence and obtain multiple class vertical edges.Intersection judge module judge these class vertical edges whether with default virtual ground
Horizontal line intersects.If one of these class vertical edges intersect with virtual ground horizontal line, position judging module judges and void
Whether the one end for intending one of the intersecting class vertical edge of surface water horizontal line is located in detection zone.If with virtual ground level
One end of one of the intersecting class vertical edge of line is located in detection zone, and detection of obstacles module judges to detect barrier
Presence and warning is provided.
In one embodiment of this invention, above-mentioned class vertical edge detection module system comprising first direction rim detection with
Second direction rim detection, to obtain multiple first direction edges and multiple second direction edges.Class vertical edge detection module
And obtain class vertical edge according to the angled relationships between these first direction edges and these second direction edges.
In one embodiment of this invention, above-mentioned vertical edge detection module polymerize multiple class vertical edges close to each other
Edge.
In one embodiment of this invention, the first direction rim detection of above-mentioned class vertical edge detection module is level
Rim detection, and second direction rim detection detects for vertical edge.
On the other hand, the present invention proposes a kind of obstacle detector, and this obstacle detector includes memory cell, class
Vertical edge detection module, class vertical edge tracing module, slope variation computing module and detection of obstacles module.Storage is single
Member at least storage includes the video streaming of multiple images.Class vertical edge detection module couples memory cell, receives video streaming
Input picture and class vertical edge is carried out to input picture detect program and obtain multiple class vertical edges.Class vertical edge is chased after
Track module couples class vertical edge detection module, continues to follow the trail of the one tracking time of these class vertical edges respectively.Slope variation
Computing module and judges whether these slope variation values are small in the slope variation value for comparing each class vertical edge in the tracking time
In threshold value.If one of slope variation value is less than threshold value, detection of obstacles module judges to detect the presence of barrier
And warning is provided.
In one embodiment of this invention, above-mentioned class vertical edge detection module includes first direction rim detection and the
Two direction rim detections, to obtain multiple first direction edges and multiple second direction edges.Class vertical edge detection module according to
First kind vertical edge is obtained according to the angled relationships between these first direction edges and these second direction edges.
In one embodiment of this invention, above-mentioned vertical edge detection module polymerize close to each other and similar slope many
Individual first kind vertical edge.
In one embodiment of this invention, the first direction rim detection of above-mentioned class vertical edge detection module is level
Rim detection, and second direction rim detection detects for vertical edge.
Based on above-mentioned, in one embodiment of this invention, obstacle detector carries out class vertical edge to input picture
Detection program and obtain class vertical edge.For further, the rim detection that obstacle detector passes through different directions
To detect close to vertical line or completely vertical edge.Thereby, by judging the class vertical edge produced by class vertical detection
Whether intersect to carry out detection of obstacles with horizon, can further improve the degree of accuracy of identification barrier and reduce erroneous judgement
Probability, to increase the travel safety of driver.
For the features described above and advantage of the present invention can be become apparent, special embodiment below, and it is detailed to coordinate accompanying drawing to make
Carefully it is described as follows.
Brief description of the drawings
Fig. 1 is the schematic diagram according to the obstacle detection system depicted in the present invention；
Fig. 2 is the block diagram according to the obstacle detector depicted in one embodiment of the invention；
Fig. 3 is the flow chart according to obstacle detection method depicted in one embodiment of the invention；
Fig. 4 is the situation schematic diagram according to the detection of obstacles depicted in one embodiment of the invention；
Fig. 5 is the block diagram according to the obstacle detector depicted in another embodiment of the present invention；
Fig. 6 is the flow chart according to obstacle detection method depicted in another embodiment of the present invention；
Fig. 7 A and Fig. 7 B are the situation schematic diagram according to the detection of obstacles depicted in one embodiment of the invention；
Fig. 8 is the block diagram according to the obstacle detector depicted in further embodiment of this invention；
Fig. 9 is the flow chart according to obstacle detection method depicted in further embodiment of this invention.
Description of reference numerals：
10：Obstacle detection system；
12：Image acquisition unit；
15、100、400、700：Obstacle detector；
110、410、710：Memory cell；
120、420、720：Class vertical edge detection module；
130、730：Intersect judge module；
140、740：Position judging module；
150、450、770：Detection of obstacles module；
430、750：Class vertical edge tracing module；
440、760：Slope variation computing module；
Img1、Img2、Img3：Image；
M1：Object；
L1：Virtual ground horizontal line；
E1、E2、E2’、E3、E3’：Class vertical edge；
Z1：Detection zone；
C1、C1’、C2、C2’：Unit area；
S201~S204：Each step of obstacle detection method described in one embodiment of the invention；
S501~S504：Each step of obstacle detection method described in another embodiment of the present invention；
S801~S806：Each step of obstacle detection method described in further embodiment of this invention.
Embodiment
Fig. 1 is the schematic diagram according to the obstacle detection system depicted in the present invention.It refer to Fig. 1, detection of obstacles system
System 10 includes image acquisition unit 12 and obstacle detector 15.Obstacle detection system 10 is applied on a vehicle, figure
As acquiring unit 12 is arranged on vehicle, the ambient image to obtain vehicle periphery.For example, image acquisition unit 12 can
The front of vehicle is arranged on, seems the top of vehicle windscreen, to obtain the carriageway image of vehicle front.Image acquisition unit
12 be, for example, to have charge coupled cell (Charge Coupled Device, abbreviation CCD) or CMOS
The imaging sensor of (Complementary Metal-Oxide Semiconductor, abbreviation CMOS) element, to obtain car
The ambient image of surrounding.More particularly, image acquisition unit 12 is, for example, the driving recording video camera with video equipment,
Can also be the digital camera with camera function, but the invention is not limited in these implementation aspects.
Obstacle detector 15 is the electronic installation with image processing function, can be implemented as computer, vehicle electric
Brain or other device for vehicular electronic.The direct or indirect connection image acquisition unit 12 of obstacle detector 15, to receive image
Image or video streaming accessed by acquiring unit 12.Base this, obstacle detector 15 can be according to image acquisition unit 12
Accessed image or video streaming carry out the detection and warning of barrier.In an embodiment of the present invention, obstacle quality testing
Marginal information of the device 15 in image is surveyed to carry out the detection of barrier.Specifically, obstacle detector
15 will carry out class vertical edge (Quai-vertical edge) detection, so as to detect the edge direction of its in image and Vertical Square
To close class vertical edge.
Therefore, the principle of the surface water horizontal line in driver's sight can be sheltered from based on barrier, one in the present invention is real
Apply in example, whether obstacle detector 15 is by judging the class vertical edge in image by the visual horizontal line on image
Learn whether vehicle's surroundings have the presence of barrier according to this.In addition, among another embodiment of the present invention, detection of obstacles dress
Put 15 can also further distinguish barrier profile and the car on road by the slope variation of class vertical edge in video streaming
Diatom.Detection of obstacles is carried out in order to further explain how to detect acquired class vertical edge according to class vertical edge,
Especially exemplified by multiple embodiments, the present invention will be described below
Fig. 2 is the block diagram according to the obstacle detector depicted in one embodiment of the invention.It refer to Fig. 2, obstacle
Analyte detection device 100 includes memory cell 110, class vertical edge detection module 120, intersection judge module 130, position judgment mould
Block 140 and detection of obstacles module 150.Memory cell 110 is, for example, random access memory (random access
Memory), flash memory (Flash) or other memories, to storage image data.Class vertical edge detection module
120 coupling memory cell 110, to read or receive at least one image that memory cell is stored.
On the other hand, class vertical edge detection module 120, intersection judge module 130, position judging module 140 and barrier
Hinder analyte detection module 150 to be obtained by software, hardware or its combination implementation, be not any limitation as herein.Software be, for example, source code,
Application software, driver are specifically intended for realizing software module or function of specific function etc..Hardware is, for example, central processing
Unit (Central Processing Unit, abbreviation CPU), programmable controller, digital signal processor (Digital
Signal Processor, abbreviation DSP), or other programmables general service or the microprocessor of specific use
(Microprocessor) etc..For example, class vertical edge detection module 120, intersection judge module 130, position judgment mould
Block 140 and e.g. computer program or the instruction of detection of obstacles module 150, it can be loaded into the place of obstacle detector 100
Device is managed, so as to perform the function of detection of obstacles.
Fig. 3 is the flow chart according to obstacle detection method depicted in one embodiment of the invention.Referring to Fig. 2 and figure
3, in the present embodiment, obstacle detection method can for example be performed by the obstacle detector 100 in Fig. 2.Take below
The step of with obstacle detection method of each item in obstacle detector 100 to illustrate the present embodiment.
First, in step S201, class vertical edge detection module 120 receives input picture and carries out class to this input picture
Vertical edge detects program and obtains multiple class vertical edges.Furthermore, it is understood that the detection of class vertical edge is to detect image
Middle edge direction is close with vertical direction or identical edge, wherein above-mentioned vertical direction is vertical with horizon direction.It is specific next
Say, in an embodiment of the present invention, the angle between the edge direction and vertical direction of class vertical edge can be less than an angle
Value, and depending on the visual practical application situation of this angle value, the present invention is not intended to limit to this.
Furthermore, it is understood that in one embodiment, class vertical edge detection module 120 is for example comprising first direction rim detection
With second direction rim detection, to obtain multiple first direction edges and multiple second direction edges.Class vertical edge detects mould
Block simultaneously obtains class vertical edge according to the angled relationships between these first direction edges and these second direction edges.Need
It is bright, due to first direction rim detection and each self-corresponding direction of second direction rim detection and differ but be known
Parameter, therefore class vertical edge detection module 120 can be based on the angled relationships between first direction edge and second direction edge
And the directionality at each edge in image is analyzed, and class vertical edge is further obtained by the edge direction in image.
In other words, when performing edge detection procedure, class vertical edge detection module 120 can carry out not Tongfang using different shades
The rim detection of tropism.So, to be not specially limited first direction rim detection each right with second direction rim detection by the present invention
Depending on the direction answered, visual practical application situation and demand.
For example, first direction rim detection can be horizontal edge detection, and second direction rim detection can be vertical
Rim detection.Therefore, in one embodiment, class vertical edge detection module 120 can utilize the Sobel of different directions
(Sobel) shade calculates the marginal value of horizontal direction and the marginal value of vertical direction.Then, class vertical edge detection module
120 not only can detect the edge in image by the marginal value of horizontal direction and the marginal value of vertical direction, also available
The proportion grading of the marginal value of horizontal direction and the marginal value of vertical direction goes out the directionality at edge.Class vertical edge detection module
120 can detect the class vertical edge in image by the directionality at edge.But the present invention is not limited to above-mentioned implementation aspect,
It can detect that the algorithm of edge direction is all applied to the present invention.For example, class vertical edge detection module 120 can also be used
Prewitt shades carry out the class vertical edge detection of the present embodiment.
Afterwards, in step S202, intersection judge module 130 judge these class vertical edges whether with default virtual ground
Horizontal line intersects.Virtual ground horizontal line is a default datum line, the position of virtual ground horizontal line in the picture in image
Put visual practical application situation and set it.In addition to directly judging class vertical edge whether by virtual ground horizontal line,
Intersect judge module 130 for example on the basis of virtual ground horizontal line line and define including the horizontal image of virtual ground
It is interval.For example, can be as detecting that class is vertical with the region within ten pixels of distance above and below virtual ground horizontal line
The image whether edge intersects with virtual ground horizontal line is interval.If it is interval interior that the part of class vertical edge is located at this image,
Intersection judge module 130 can judge that such vertical edge intersects with virtual ground horizontal line according to this.
It should be noted that, judge whether class vertical edge with virtual ground horizontal line intersects it in intersection judge module 130
Before, vertical edge detection module 120 may also aggregate (grouping) multiple class vertical edges close to each other.Specifically, hang down
Straight edge detection module 120 can be learnt according to the position of these class vertical edges a class vertical edge whether with it is another kind of vertical
Edge is close to each other, and polymerize multiple first kind vertical edges close to each other.In simple terms, the class vertical edges of same barrier
Edge can be integrated or merged by this step, thus reduction intersection judge module 130 judge class vertical edge whether with virtual ground
The intersecting amount of calculation of horizontal line.
If one of these first kind vertical edges intersect with virtual ground horizontal line, in step S203, position judgment
Module 140 judges whether one end of one of the class vertical edge intersected with virtual ground horizontal line is located in detection zone.
Furthermore, it is understood that position judging module 140 can judge these and virtual ground water according to the positional information of class vertical edge
Whether one end of one of the intersecting class vertical edge of horizontal line is located in detection zone, and this measure can further be told virtually
Whether the object associated by the intersecting class vertical edge of face horizontal line is the barrier for stopping driving.Specifically, position judgment
Module 140 can further be picked out and the enough close barriers of the distance between vehicle by step S203 judgement.Due to
Can't hinder the traveling of vehicle with vehicle distances too remote object, thus position judging module 140 can exclude with vehicle away from
From the possibility that too remote object is barrier.That is, the edge based on distant objects in the picture can't fall
In detection zone described in the present embodiment, obstacle detector 100 can be by judging whether one end of class vertical edge is located at
Barrier and the object in a distant place are told in detection zone.
Then, if one end of one of first kind vertical edge intersected with virtual ground horizontal line is located at detection zone
Interior, in step S204, detection of obstacles module 150 judges to detect the presence of barrier and provides warning.Detection of obstacles mould
Warning such as prompt text, the sound and the one of light that block 150 is provided or its combine, but be not restricted to this.Obstacle quality testing
The presentation mode of prompting warning can be changed with practical application request by surveying module 150.
Fig. 4 is the situation schematic diagram according to the detection of obstacles depicted in one embodiment of the invention.It refer to Fig. 4, it is assumed that
Image Img1 is the Chinese herbaceous peony image captured by the image acquisition unit on vehicle, and has object M1 in image Img1.First, hinder
Hinder analyte detection device to carry out class vertical edge to image Img1 and detect program.In the example shown in Fig. 4, obstacle detector
Class vertical edge E1 is at least can detect, and judges that class vertical edge E1 intersects with virtual ground horizontal line L 1.As shown in figure 4,
Class vertical edge E1 is the edge close with vertical direction angle.In addition, obstacle detector also judges class vertical edge E1
One end fall in detection zone Z1.Base this, because class vertical edge E1 intersects and its one end falls with virtual ground horizontal line L 1
In detection zone Z1, therefore obstacle detector will judge that du vehicule has barrier and provides warning to driver.
Fig. 5 is the block diagram according to the obstacle detector depicted in another embodiment of the present invention.Fig. 5 is refer to, is hindered
Hinder analyte detection device 400 include memory cell 410, class vertical edge detection module 420, class vertical edge tracing module 430, tiltedly
Rate changes computing module 440 and detection of obstacles module 450.The phase of memory cell 110 of memory cell 410 and previous embodiment
It is seemingly or identical, repeated no more with this.
On the other hand, in the present embodiment, the coupling of class vertical edge detection module 410 memory cell 410, to receive video
Multiple images in crossfire.The coupling class vertical edge of class vertical edge tracing module 430 detection module 410, with to multiple images
On edge be tracked.Class vertical edge detection module 420, class vertical edge tracing module 430, slope variation computing module
440 and detection of obstacles module 450 can by software, hardware or its combine implementation and obtain, be not any limitation as herein.
Software be, for example, source code, application software, driver or be specifically intended for realizing specific function software module or
Function etc..Hardware is, for example, CPU (Central Processing Unit, abbreviation CPU), programmable control
Device, digital signal processor (Digital Signal Processor, abbreviation DSP), or other programmables general use
Microprocessor (Microprocessor) of way or specific use etc..For example, class vertical edge detection module 420, class are hung down
Straight edge tracing module 430, slope variation computing module 440 and detection of obstacles module 450 are, for example, computer program or referred to
Order, it can be loaded into the processor of obstacle detector 400, so as to perform the function of detection of obstacles.
Fig. 6 is the flow chart according to obstacle detection method depicted in another embodiment of the present invention.Referring to Fig. 5 with
Fig. 6, in the present embodiment, obstacle detection method can for example be performed using the obstacle detector 400 in Fig. 5.Below
Arrange in pairs or groups obstacle detection method of each item to illustrate the present embodiment in obstacle detector 400 the step of.
First, in step S501, class vertical edge detection module 420 receives the input picture of video streaming and input is schemed
Multiple class vertical edges are obtained as carrying out class vertical edge detection program.In simple terms, class vertical edge detection module 420 from
Memory cell 410 obtains video streaming, and this video streaming is made up of multiple continuous images of shooting time.Class vertical edge
Detection module 420 carries out class vertical edge detection to the image included by these video streamings, so as to obtain many on each image
Individual class vertical edge.Wherein, class vertical edge detection module 420 perform the detailed implementation methods of class vertical edge detection with it is foregoing
The class vertical edge detection module 120 of embodiment is similar, is repeated no more in this.In one embodiment, class vertical edge detection mould
Block 420 can carry out the detection of class vertical edge for each image and obtain the class vertical edge image corresponding to each image, and all kinds of
Vertical image includes multiple class vertical edges.
Then, in step S502, class vertical edge tracing module 430, which continues to follow the trail of one section of these class vertical edges respectively, to be chased after
The track time.In step S503, slope variation computing module 440 compares the slope variation of each class vertical edge within the tracking time
Value, and judge whether these slope variation values are less than threshold value.Furthermore, it is understood that in order to mutual corresponding class on different images
Vertical edge is compared, and class vertical edge tracing module 430 continues to follow the trail of the one tracking time of these class vertical edges respectively.
On the other hand, slope variation computing module 440 will compare the mutual corresponding respective slope of class vertical edge on different images, and
Judge whether slope variation value is less than threshold value.
Among an embodiment, class vertical edge tracing module 430 and slope variation computing module 440 can utilize gradient
Angular histogram (Histogram of Oriented Gradient, abbreviation HOG) follows the trail of these class vertical edges and ratio
The slope of relatively mutual corresponding class vertical edge.Specifically, class vertical edge image can be distinguished into multiple unit areas
(cell) the gradient angular histogram in constituent parts region can, be set up by counting the characteristic value of each pixel.Base this, by comparing
The gradient angular histogram in constituent parts region, class vertical edge tracing module 430, which can be obtained, represents identical scenery on different images
Unit area, to obtain mutual corresponding unit area on different images, thus reach follow the trail of class vertical edge purpose.
Slope variation computing module 440 can for example judge that slope becomes by comparing the histogrammic statistical result of gradient angle
Whether change value is less than threshold value.Furthermore, it is understood that by the histogrammic statistics of gradient angle, slope variation computing module 440 can
Learn the angle value corresponding to highest statistics groove (most significant bin) in each gradient angular histogram.Base this,
By comparing the angle value corresponding to the highest statistics groove of mutually corresponding unit area, slope variation computing module 440 can evidence
To learn the slope variation of all kinds of vertical edges.
It should be noted that, for the lane line on road, the edge of lane line can change with the steering of vehicle
Direction.On the contrary, the edge of barrier can't be with changing direction because of Vehicular turn, therefore obstacle detector can lead to
Cross the slope variation value for analyzing and comparing class vertical edge within the tracking time and determine whether to detect barrier.Change speech
It, by the slope variation value of class vertical edge, obstacle detector can tell the stereo object in image according to this and put down
Still it is not enough to hinder vehicle to advance on face object, wherein plane object.Thereby, if one of slope variation value is less than threshold
Value, in step S504, detection of obstacles module 450 judges to detect the presence of barrier and provides warning.
Fig. 7 A and Fig. 7 B are the situation schematic diagram according to the detection of obstacles depicted in one embodiment of the invention.Please join simultaneously
According to Fig. 7 A and Fig. 7 B, it is assumed that the video streaming acquired in image acquisition unit on vehicle includes image Img2 and image Img3,
And this vehicle obtains image Img2 and image Img3 in the state of steering.In other words, image Img2 and image Img3 is vehicle
On image acquisition unit in the Chinese herbaceous peony image acquired in different time, and image Img2 the acquisition time earlier than image Img3's
The acquisition time.Furthermore, it was found from the scenery in image Img2 and image Img3, it is assumed that vehicle is in turn right in the state of.
First, obstacle detector carries out class vertical edge to image Img2 and image Img3 respectively and detects program.
In example shown in Fig. 7 A and Fig. 7 B, obstacle detector at least can detect image Img2 class vertical edge E2 and class is hung down
Straight edge E3, and obstacle detector at least can detect image Img3 class vertical edge E2 ' and class vertical edge E3 '.
Furthermore, the calculating based on HOG, obstacle detector can learn image Img2 unit area C1 and image Img3 unit area
Domain C1 ' is mutual corresponding relation, and can learn that image Img2 unit area C2 and image Img3 unit area C2 ' are phase
Mutual corresponding relation.
Then, by comparing unit area C1 gradient orientation histogram and unit area C1 ' gradient orientation histogram,
Obstacle detector can learn whether the slope variation value between class vertical edge E3 and class vertical edge E3 ' is more than threshold
Value.Likewise, by comparing unit area C2 gradient orientation histogram and unit area C2 ' gradient orientation histogram, barrier
Analyte detection device is hindered to learn whether the slope variation value between class vertical edge E2 and class vertical edge E2 ' is more than threshold value.
In this example, class vertical edge E3 and class vertical edge E3 ' is the edge for being associated with front obstacle, and class is vertical
Edge E2 and class vertical edge E2 ' is the edge for being associated with lane line.As shown in figures 7 a and 7b, when Vehicular turn, class is vertical
Slope variation between edge E3 and class vertical edge E3 ' is simultaneously little.On the contrary, when Vehicular turn, class vertical edge E2 with
Slope variation between class vertical edge E2 ' is clearly.Base this, obstacle detector only hangs down class vertical edge E3 with class
Straight edge E3 ' is considered as correspondence to the edge of barrier.That is, obstacle detector can be by comparing within the tracking time
The slope variation value of each class vertical edge, to judge whether to detect barrier.
Fig. 8 is the block diagram according to the obstacle detector depicted in further embodiment of this invention.Fig. 8 is refer to, is hindered
Analyte detection device 700 is hindered to include memory cell 710, class vertical edge detection module 720, intersection judge module 730, position judgment
Module 740, class vertical edge tracing module 750, slope variation computing module 760 and detection of obstacles module 770.It is above-mentioned each
Item is same or similar with embodiment shown in Fig. 2 and Fig. 5, and those skilled in the art can refer to Fig. 2 and Fig. 5 phase
Speak on somebody's behalf bright and analogize it, repeated no more in this.
Fig. 9 is the flow chart according to obstacle detection method depicted in further embodiment of this invention.Referring to Fig. 8 with
Fig. 9, in the present embodiment, obstacle detection method can for example be performed using the obstacle detector 700 in Fig. 8.In addition,
In the present embodiment, obstacle detector can carry out obstacle by multiple images in single image and video streaming simultaneously
Analyte detection, thus improve the degree of accuracy of detection of obstacles.Each item in collocation obstacle detector 700 illustrates below
The step of obstacle detection method of the present embodiment.
In step S801, class vertical edge detection module 720 receives the input picture of video streaming and input picture is entered
Row class vertical edge detects program and obtains multiple class vertical edges.In step S802, intersection judge module 730 judges these classes
Whether vertical edge intersects with default virtual ground horizontal line.If step S802 is judged as NO, in step S803, class vertical edges
Edge tracing module 750 continues to follow the trail of the one tracking time of these class vertical edges respectively.
Afterwards, in step S805, slope variation computing module 760 compares the oblique of each class vertical edge within the tracking time
Rate changing value, and judge whether these slope variation values are less than threshold value.If step S805 is judged as YES or step S802 judges
It is yes, in step S804, position judging module 740 judges the class vertical edge therein one intersected with virtual ground horizontal line
Whether one end is located in detection zone, and judges that correspondence is less than one of class vertical edge of threshold value to slope variation value
One end whether be located at detection zone in.If step S804 is judged as YES, in step S806, detection of obstacles module 770 judges
Detect the presence of barrier and warning is provided.Above-mentioned steps S801~step S806 detailed content is referred to Fig. 1 to Fig. 7
Related description and analogize it, will not be repeated here.
In summary, the obstacle detector in one embodiment of the invention carries out obstacle by way of image procossing
Analyte detection.Obstacle detector is detected close to vertical line but non-fully vertical by the rim detection of different directions
Edge, and determine whether the presence of barrier using the information produced by these class vertical edges.Base this, it is vertical by class
The detection at edge, can avoid the erroneous judgement caused by excessively complicated marginal information or specific edge feature, be distinguished so as to improve
Know the degree of accuracy of barrier.In addition, during using class vertical edge detection barrier, the present invention can also pass through list
Video streaming produced by camera picks out stereo object and plane object in image, vertical compared to many cameras
Body vision technology, the present invention can save substantial amounts of cost and reduce operand.
Finally it should be noted that：Various embodiments above is merely illustrative of the technical solution of the present invention, rather than its limitations；To the greatest extent
The present invention is described in detail with reference to foregoing embodiments for pipe, it will be understood by those within the art that：Its according to
The technical scheme described in foregoing embodiments can so be modified, or which part or all technical characteristic are entered
Row equivalent substitution；And these modifications or replacement, the essence of appropriate technical solution is departed from various embodiments of the present invention technology
The scope of scheme.
Claims (4)
1. a kind of obstacle detector, it is characterised in that including：
Memory cell, at least storage include the video streaming of multiple images；
Class vertical edge detection module, couples the memory cell, the input picture of the video streaming is received, to the input
Image carries out class vertical edge detection program and obtains multiple class vertical edges；
Class vertical edge tracing module, couples the class vertical edge detection module, continues to follow the trail of the class vertical edge respectively
One tracking time；
Slope variation computing module, compares the slope variation value of each class vertical edge within the tracking time, and sentences
Whether the slope variation value of breaking is less than threshold value；And
Detection of obstacles module, if one of described slope variation value is less than the threshold value, the detection of obstacles module
Judge to detect the presence of barrier and warning is provided.
2. obstacle detector according to claim 1, it is characterised in that the class vertical edge detection module is included
First direction rim detection and second direction rim detection, to obtain multiple first direction edges and multiple second direction edges,
And obtain the class vertical edge according to the angled relationships between the first direction edge and the second direction edge.
3. obstacle detector according to claim 2, it is characterised in that the class vertical edge detection module polymerization
The similar class vertical edge of close to each other and slope.
4. obstacle detector according to claim 3, it is characterised in that the institute of the class vertical edge detection module
State first direction rim detection to detect for horizontal edge, the second direction rim detection detects for vertical edge.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710412721.7A CN107255470B (en) | 2014-03-19 | 2014-03-19 | Obstacle detection device |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710412721.7A CN107255470B (en) | 2014-03-19 | 2014-03-19 | Obstacle detection device |
| CN201410102894.5A CN104931024B (en) | 2014-03-19 | 2014-03-19 | Obstacle detector |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410102894.5A Division CN104931024B (en) | 2014-03-19 | 2014-03-19 | Obstacle detector |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107255470A true CN107255470A (en) | 2017-10-17 |
| CN107255470B CN107255470B (en) | 2020-01-10 |
Family
ID=54118302
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710412721.7A Active CN107255470B (en) | 2014-03-19 | 2014-03-19 | Obstacle detection device |
| CN201410102894.5A Active CN104931024B (en) | 2014-03-19 | 2014-03-19 | Obstacle detector |
Family Applications After (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410102894.5A Active CN104931024B (en) | 2014-03-19 | 2014-03-19 | Obstacle detector |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN107255470B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111198385A (en) * | 2019-12-26 | 2020-05-26 | 北京旷视机器人技术有限公司 | Obstacle detection method, obstacle detection device, computer device, and storage medium |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3309707A1 (en) * | 2016-10-13 | 2018-04-18 | Conti Temic microelectronic GmbH | Device and method for gradient-based object recognition of an input image |
| CN112965474A (en) * | 2019-11-27 | 2021-06-15 | 深圳市大富科技股份有限公司 | AGV navigation control method and system and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0961157A (en) * | 1995-08-25 | 1997-03-07 | Nippon Seiki Co Ltd | Distance-measuring device |
| JPH1097699A (en) * | 1996-09-19 | 1998-04-14 | Toyota Motor Corp | Obstacle detecting device for vehicle |
| CN1576123A (en) * | 2003-07-03 | 2005-02-09 | 黄保家 | Anticollision system for motor vehicle |
| CN1782668A (en) * | 2004-12-03 | 2006-06-07 | 曾俊元 | Method and device for preventing collison by video obstacle sensing |
| EP1722332A1 (en) * | 2005-05-13 | 2006-11-15 | Nissan Motor Company Limited | Apparatus and method for image processing |
| CN101436300A (en) * | 2008-12-10 | 2009-05-20 | 东软集团股份有限公司 | Method and apparatus for dividing barrier |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3267105B2 (en) * | 1995-06-28 | 2002-03-18 | トヨタ自動車株式会社 | Object recognition device |
| CN1987357B (en) * | 2006-12-26 | 2010-05-19 | 浙江工业大学 | Intelligent parking auxiliary device based on omnibearing computer sight |
| JP4410292B1 (en) * | 2008-10-20 | 2010-02-03 | 本田技研工業株式会社 | Vehicle periphery monitoring device |
| JP2011033594A (en) * | 2009-08-06 | 2011-02-17 | Panasonic Corp | Distance calculation device for vehicle |
-
2014
- 2014-03-19 CN CN201710412721.7A patent/CN107255470B/en active Active
- 2014-03-19 CN CN201410102894.5A patent/CN104931024B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0961157A (en) * | 1995-08-25 | 1997-03-07 | Nippon Seiki Co Ltd | Distance-measuring device |
| JPH1097699A (en) * | 1996-09-19 | 1998-04-14 | Toyota Motor Corp | Obstacle detecting device for vehicle |
| CN1576123A (en) * | 2003-07-03 | 2005-02-09 | 黄保家 | Anticollision system for motor vehicle |
| CN1782668A (en) * | 2004-12-03 | 2006-06-07 | 曾俊元 | Method and device for preventing collison by video obstacle sensing |
| EP1722332A1 (en) * | 2005-05-13 | 2006-11-15 | Nissan Motor Company Limited | Apparatus and method for image processing |
| CN101436300A (en) * | 2008-12-10 | 2009-05-20 | 东软集团股份有限公司 | Method and apparatus for dividing barrier |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111198385A (en) * | 2019-12-26 | 2020-05-26 | 北京旷视机器人技术有限公司 | Obstacle detection method, obstacle detection device, computer device, and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104931024A (en) | 2015-09-23 |
| CN104931024B (en) | 2017-07-14 |
| CN107255470B (en) | 2020-01-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112292711B (en) | Associating LIDAR data and image data | |
| US10685246B2 (en) | Systems and methods for curb detection and pedestrian hazard assessment | |
| US11305691B2 (en) | Vehicular vision system | |
| CN105835880B (en) | Lane following system | |
| CN106169244B (en) | Device and method is provided using the guidance information of crossing recognition result | |
| US9047518B2 (en) | Method for the detection and tracking of lane markings | |
| US9547795B2 (en) | Image processing method for detecting objects using relative motion | |
| EP2463843A2 (en) | Method and system for forward collision warning | |
| US10699567B2 (en) | Method of controlling a traffic surveillance system | |
| US10719699B2 (en) | Pedestrian detection method and system in vehicle | |
| CN102792314A (en) | Cross traffic collision alert system | |
| Lin et al. | Lane departure and front collision warning using a single camera | |
| JP2015153295A (en) | Environment recognition system, vehicle, and camera dirt detection method | |
| CN111932901A (en) | Road vehicle tracking detection apparatus, method and storage medium | |
| US10832428B2 (en) | Method and apparatus for estimating a range of a moving object | |
| WO2013116598A1 (en) | Low-cost lane marker detection | |
| Petrovai et al. | A stereovision based approach for detecting and tracking lane and forward obstacles on mobile devices | |
| CN104931024B (en) | Obstacle detector | |
| US9245342B2 (en) | Obstacle detection device | |
| Leu et al. | High speed stereo vision based automotive collision warning system | |
| JP4864953B2 (en) | Vehicle periphery monitoring device | |
| KR20160125803A (en) | Apparatus for defining an area in interest, apparatus for detecting object in an area in interest and method for defining an area in interest | |
| KR102003387B1 (en) | Method for detecting and locating traffic participants using bird's-eye view image, computer-readerble recording medium storing traffic participants detecting and locating program | |
| JP4765113B2 (en) | Vehicle periphery monitoring device, vehicle, vehicle periphery monitoring program, and vehicle periphery monitoring method | |
| JP2020201746A (en) | Distance estimation device, distance estimation method, and distance estimation computer program |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20211115 Address after: Taiwan Hsinchu Science Park, 12 Li Hang Road, China Patentee after: Huajing Technology Co., Ltd Address before: 3 / F, No. 12, Lixing Road, Hsinchu Patentee before: Nengjing Technology Co., Ltd |
|
| TR01 | Transfer of patent right |