CN111192283A - Height limiting rod detection and height calculation method - Google Patents
Height limiting rod detection and height calculation method Download PDFInfo
- Publication number
- CN111192283A CN111192283A CN201911368175.7A CN201911368175A CN111192283A CN 111192283 A CN111192283 A CN 111192283A CN 201911368175 A CN201911368175 A CN 201911368175A CN 111192283 A CN111192283 A CN 111192283A
- Authority
- CN
- China
- Prior art keywords
- height
- edge
- point
- candidate
- limiting rod
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20228—Disparity calculation for image-based rendering
Abstract
The invention discloses a height limiting rod detection and height calculation method, which comprises the steps of acquiring image information of a limiting rod through binocular cameras with the heights of the two eyes staggered; extracting edge points from image information, and matching by using a high-robustness edge point matching algorithm to obtain an edge point disparity map, wherein the edge point disparity map is a map only with the edge points, and the value of each point is the left-right disparity of a binocular camera; and detecting a group of edges possibly representing height-limiting rods based on the edge disparity map, calculating the respective heights of the edges, wherein the edges with the heights falling into the height range of the height-limiting rods are the height-limiting rods, and outputting the heights of the edges. The advantages are that: the polar line is prevented from being parallel to the horizontal line, and the reliability of parallax calculation is ensured by adopting a high-robustness edge matching algorithm; the height-limiting rod detection algorithm based on the edge disparity map of the image has more accurate calculation result.
Description
Technical Field
The invention relates to a height limiting rod detection and height calculation method, and belongs to the technical field of automobile auxiliary driving.
Background
The height-limiting rod is a common safety device, and the purpose of the height-limiting rod is roughly divided into two types, namely, the height-limiting rod is used for warning a driver that an obstacle exists in front of the driver to limit the height (a parking lot entrance, a viaduct, a tunnel and the like), and the height-limiting rod is used for reminding a truck of not allowing to enter. The general height limiting rod is provided with a specific height limiting height, and a driver can judge whether the vehicle can safely pass through the height limiting rod. In recent years, in order to improve safety, some more accurate warning devices are provided, such as a laser height measuring device arranged on the roadside in front of a height limiting rod, and a warning signal is sent to a driver when the vehicle passes through an ultrahigh height.
Recently, it has been proposed to use binocular vision for height-limiting pole detection. The vehicle-mounted height limiting rod detection and height measurement has the advantages that when a driver makes a mistake or ignores (such as unseen, short napping and the like), the system can directly send out a warning signal to the driver, and even can be matched with an Automatic Emergency Braking (AEB) function of an automatic driving L3 level, so that the vehicle can be stopped in time, and accidents and casualties are avoided. At present, a vehicle-mounted binocular vision-based height limiting rod detection method comprises the steps of firstly calculating a disparity map, and then detecting the height limiting rod according to equal point disparity on the height limiting rod. This approach has two fundamental limitations: 1/because the height limiting rod is horizontal and is superposed with the polar line, and the appearance of the height limiting rod is a same-color or two-color paint strip, the calculated parallax reliability is low; 2/at a distance, the height-limiting rod has a small width in the image, has low matching reliability, and is removed in post-processing and is empty in the disparity map.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a height limiting rod detection and height calculation method.
In order to solve the technical problem, the invention provides a height limiting rod detection and height calculation method, which comprises the steps of acquiring image information of a limiting rod through binocular cameras with the heights of the two eyes staggered;
extracting edge points from the left image and the right image, and matching by using a high-robustness edge point matching algorithm to obtain an edge point disparity map, wherein the edge point disparity map is a map only with the edge points, and the value of each point is the left-right disparity of the binocular camera;
and detecting a group of edges possibly representing height-limiting rods based on the edge disparity map, calculating the respective heights of the edges, wherein the edges with the heights falling into the height range of the height-limiting rods are the height-limiting rods, and outputting the heights of the edges.
Further, the matching calculation process of the edge matching algorithm model is as follows:
converting image information acquired by a left camera and a right camera of a binocular camera into gray images to obtain a left image and a right image, and respectively acquiring a left edge point and a right edge point from the left image and the right image by using a Canny edge point detection algorithm;
respectively connecting the left edge point and the right edge point into a left edge and a right edge according to direct connection, wherein the edges reflect the outline or the picture characteristics of the corresponding objects;
calculating epipolar lines of each left edge point in the right image according to the calibration parameters, searching in a certain range along the epipolar lines, and searching candidate corresponding points (simple features such as brightness, gradient values, directions and the like) through simple features;
for each left edge point, on the left edge, each neighboring point in a neighborhood of fixed size centered on the edge point is examined;
determining whether candidate corresponding points exist in all adjacent points of the edge point, if at least one candidate corresponding point exists in the adjacent points, performing parallax gradient calculation on one of the candidate corresponding points of the current edge point and one of the candidate corresponding points of one adjacent point of the current edge point, and if the calculation result is smaller than a preset parallax gradient limit, throwing the adjacent point to one ticket of the current candidate corresponding point of the current edge point;
finding out the candidate corresponding point with the highest ticket number of each edge point, and if the ticket number is higher than a preset ticket number threshold, determining that the edge point has a matched candidate corresponding point.
Further, the certain range is a maximum parallax and a minimum parallax corresponding to a farthest and closest distance to be detected in front of the vehicle;
the determination criteria of the candidate corresponding points are as follows: the absolute value of the difference between the gradient and the direction of the left edge point and the right edge point is smaller than a preset threshold value.
Further, the formula of the parallax gradient calculation is as follows:
|disp1–disp2|<DG0*d12+1.5
wherein disp1And disp2For the disparity of the two candidate correspondences to be compared, d12Is the current point and its neighbors on the same edgeDistance of a neighbor within a domain, DG0Is the parallax gradient limit.
Further, the height calculation process of the height-limiting rod detection algorithm model is as follows:
gathering the edge points of each row in the left image, which have matching candidate corresponding points, into a set;
if the number of the aggregation points is higher than a preset point threshold value, calculating a parallax histogram of edge points in the aggregation;
determining each parallax value of a horizontal axis of the histogram, and if the sum of the current parallax value, the left parallax height and the right parallax height is larger than a preset height threshold value, determining the behavior candidate height limiting rod of the left image;
calculating the parallax of each candidate height-limiting rod and the height in the image of the candidate height-limiting rod to obtain the height of the candidate height-limiting rod;
and according to the setting specification of the height limit rods, deleting unreasonable candidate height limit rods in the candidate height limit rods, reserving the candidate height limit rods specified by the setting specification of the nearest height limit rod, and outputting the heights of the candidate height limit rods.
Further, if the number of candidate height-limiting rods is greater than one and the parallax values are close (when the height-limiting rods are close, the image is wide, the edge detection may give an upper edge contour and a lower edge contour, and the parallax values of the upper edge contour and the lower edge contour are close, the lower candidate height-limiting rod is taken at this time), and the candidate height-limiting rods with the lower candidate height-limiting rods in the image are merged. The invention achieves the following beneficial effects:
the binocular cameras are designed to be in high-dislocation arrangement, so that polar lines are prevented from being parallel to horizontal lines, and the reliability of parallax calculation is guaranteed; the edge points of the height limiting rod represent the height limiting rod, so that even if the image width is narrow when the distance is long, the edge of the height limiting rod can be detected, and then matching is carried out to calculate the parallax. The high reliability of parallax calculation is ensured by adopting the high-robustness edge matching algorithm, so that the reliability of the detection algorithm is ensured. The height-limiting rod detection algorithm based on the edge disparity map of the image has more accurate calculation result.
Drawings
FIG. 1 is a schematic flow diagram of the present invention.
Detailed Description
In order to make the objects, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the embodiments described below are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The technical scheme of the invention is further explained by the specific implementation mode in combination with the attached drawings.
As shown in fig. 1, in the height limiting rod detection and height calculation method, image information of a limiting rod is acquired through binocular cameras with the heights of the two eyes staggered;
matching by adopting a pre-constructed edge matching algorithm model with high robustness, and determining matching parameters;
and inputting the matching parameters into a pre-constructed height limit rod detection algorithm model based on the edge disparity map, and determining the height of the height limit rod.
The edge matching algorithm process of the edge matching algorithm model with high robustness comprises the following steps:
1. edge point detection (edge detection): the video images of the left camera and the right camera are converted into gray level images, and a Canny edge point detection algorithm is used.
2. Edge point connections (edge linking introcurves): directly connected edge points are connected as edges.
3. Selecting candidate corresponding points: for each point in the left image, the epipolar line of the point in the right image is calculated, and the search is carried out in a certain range along the epipolar line.
a) The search range is as follows: and calculating the maximum and minimum parallaxes corresponding to the farthest and closest distances of the three-dimensional space.
b) Candidate correspondent point acceptance criteria: the gradient and direction of the left and right edge points are less than a given threshold. Equality is not required here because the left and right cameras cannot do the same and the image is noisy.
c) Each edge point has 0 to N candidate correspondences.
4. The voting mechanism comprises the following steps:
a) for each edge point of the left image, each neighboring point, i.e. a neighborhood of fixed size belonging to the same edge and centered on the edge point, is examined.
b) And each candidate corresponding point of the current edge point is considered. If the neighboring point has at least one candidate corresponding point, the disparity gradient of the neighboring point, i.e. the disparity difference divided by the distance between the current edge point and the neighboring point, is less than a threshold value, which is called "disparity gradient limit", then the neighboring point is cast to the current candidate corresponding point of the current edge point by one vote. In the implementation, the above condition is rewritten to | disp in consideration of the edge position error1–disp2|<DG0*d12+1.5，disp1And disp2For the disparity of the two candidate correspondences to be compared, d12Is the distance between them, DG0For the parallax gradient limit, we choose 0.2.
c) The candidate corresponding point with the highest ticket number of each edge point is checked. If the total number of votes is above a given threshold, the candidate corresponding point is deemed to be correct.
The height-limiting rod detection algorithm process of the height-limiting rod detection algorithm model based on the edge disparity map comprises the following steps:
1. each line of the left image has edge points matching the corresponding points, grouped into a set.
2. If the number of points of the set is above a given threshold, a disparity histogram of the points within the set is computed.
3. Each disparity value is considered on the horizontal axis of the histogram. This image behavior is a candidate height-limiting bar if the sum of the current disparity and the left-right disparity height is greater than a given threshold.
4. If the number of the candidate height limiting rods is more than one and the candidate height limiting rods with similar parallax values exist, merging and taking the candidate height limiting rods with the lower parallax values.
5. And calculating the height of each candidate height-limiting rod according to the parallax and the height in the image of each candidate height-limiting rod.
6. And deleting unreasonable candidate height limiting rods according to the height limiting rod setting rules.
The invention is mainly characterized in that: the 1/binocular camera is designed to be in high-dislocation arrangement, so that polar lines are prevented from being parallel to a horizontal line, and the reliability of parallax calculation is ensured; 2/representing the height limiting rod by the edge point of the height limiting rod, so that even if the image width is narrow when the distance is long, the edge of the height limiting rod can be detected, and then matching is carried out to calculate the parallax. 3/the invention provides an edge matching algorithm with high robustness, which ensures high reliability of parallax calculation and further ensures the reliability of a detection algorithm. 4/the invention is a height-limiting rod detection algorithm based on an edge disparity map of an image, and the calculation is accurate.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (6)
1. A height limiting rod detection and height calculation method is characterized in that image information of a limiting rod is obtained through binocular cameras with the heights of the two eyes staggered;
extracting edge points from image information, and matching by using a high-robustness edge point matching algorithm to obtain an edge point disparity map, wherein the edge point disparity map is a map only with the edge points, and the value of each point is the left-right disparity of a binocular camera;
and detecting a group of edges possibly representing height-limiting rods based on the edge disparity map, calculating the respective heights of the edges, wherein the edges with the heights falling into the height range of the height-limiting rods are the height-limiting rods, and outputting the heights of the edges.
2. The height-limiting rod detecting and height calculating method according to claim 1, wherein the matching calculation process of the edge matching algorithm model is as follows:
converting image information acquired by a left camera and a right camera of a binocular camera into gray images to obtain a left image and a right image, and respectively acquiring a left edge point and a right edge point from the left image and the right image by using a Canny edge point detection algorithm;
respectively connecting the left edge point and the right edge point into a left edge and a right edge according to direct connection, wherein the edges reflect the outline or the picture characteristics of the corresponding objects;
calculating the epipolar line of each left edge point in the right image according to the calibration parameters, searching in a certain range along the epipolar line, and searching for candidate corresponding points through simple features;
for each left edge point, on the left edge, each neighboring point in a neighborhood of fixed size centered on the edge point is examined;
determining whether candidate corresponding points exist in all adjacent points of the edge point, if at least one candidate corresponding point exists in the adjacent points, performing parallax gradient calculation on one of the candidate corresponding points of the current edge point and one of the candidate corresponding points of one adjacent point of the current edge point, and if the calculation result is smaller than a preset parallax gradient limit, throwing the adjacent point to one ticket of the current candidate corresponding point of the current edge point;
finding out the candidate corresponding point with the highest ticket number of each edge point, and if the ticket number is higher than a preset ticket number threshold, determining that the edge point has a matched candidate corresponding point.
3. The height-limiting rod detecting and height calculating method according to claim 2, wherein the certain range is a maximum and a minimum parallax corresponding to a farthest and closest distance to be detected in front of the vehicle;
the determination criteria of the candidate corresponding points are as follows: the absolute value of the difference between the gradient and the direction of the left edge point and the right edge point is smaller than a preset threshold value.
4. The height-limiting rod detecting and height calculating method according to claim 2, wherein the formula of the parallax gradient calculation is:
|disp1–disp2|<DG0*d12+1.5
wherein disp1And disp2For the disparity of the two candidate correspondences to be compared, d12Is the distance, DG, between the current point on the same edge and a neighbouring point in its neighbourhood0Is the parallax gradient limit.
5. The height-limiting rod detecting and height calculating method according to claim 1, wherein the height calculating process of the height-limiting rod detecting algorithm model is as follows:
gathering the edge points of each row in the left image, which have matching candidate corresponding points, into a set;
if the number of the aggregation points is higher than a preset point threshold value, calculating a parallax histogram of edge points in the aggregation;
determining each parallax value of a horizontal axis of the histogram, and if the sum of the current parallax value, the left parallax height and the right parallax height is larger than a preset height threshold value, determining the behavior candidate height limiting rod of the left image;
calculating the parallax of each candidate height-limiting rod and the height in the image of the candidate height-limiting rod to obtain the height of the candidate height-limiting rod;
and according to the setting specification of the height limit rods, deleting unreasonable candidate height limit rods in the candidate height limit rods, reserving the candidate height limit rods specified by the setting specification of the nearest height limit rod, and outputting the heights of the candidate height limit rods.
6. The height-limiting rod detecting and height calculating method according to claim 5, wherein if the number of candidate height-limiting rods is greater than one and there are candidate height-limiting rods with similar disparity values, the candidate height-limiting rods with low positions in the image are merged and taken as the candidate height-limiting rods.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911368175.7A CN111192283A (en) | 2019-12-26 | 2019-12-26 | Height limiting rod detection and height calculation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911368175.7A CN111192283A (en) | 2019-12-26 | 2019-12-26 | Height limiting rod detection and height calculation method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111192283A true CN111192283A (en) | 2020-05-22 |
Family
ID=70708020
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911368175.7A Pending CN111192283A (en) | 2019-12-26 | 2019-12-26 | Height limiting rod detection and height calculation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111192283A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113378805A (en) * | 2021-08-13 | 2021-09-10 | 北京中科慧眼科技有限公司 | Height limiting device detection method and system based on deep learning and intelligent terminal |
| CN113658226A (en) * | 2021-08-26 | 2021-11-16 | 中国人民大学 | Height detection method and system for height limiting device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108027877A (en) * | 2015-09-18 | 2018-05-11 | 高通股份有限公司 | System and method for the detection of non-barrier |
| JP2018088092A (en) * | 2016-11-28 | 2018-06-07 | 京セラ株式会社 | Image processing apparatus, on-vehicle system, on-vehicle camera, moving body, and image processing method |
| US20180357495A1 (en) * | 2016-02-23 | 2018-12-13 | Genki WATANABE | Image processing apparatus, imaging apparatus, mobile device control system, image processing method, and recording medium |
| CN110069990A (en) * | 2019-03-18 | 2019-07-30 | 北京中科慧眼科技有限公司 | A kind of height-limiting bar detection method, device and automated driving system |
-
2019
- 2019-12-26 CN CN201911368175.7A patent/CN111192283A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108027877A (en) * | 2015-09-18 | 2018-05-11 | 高通股份有限公司 | System and method for the detection of non-barrier |
| US20180357495A1 (en) * | 2016-02-23 | 2018-12-13 | Genki WATANABE | Image processing apparatus, imaging apparatus, mobile device control system, image processing method, and recording medium |
| JP2018088092A (en) * | 2016-11-28 | 2018-06-07 | 京セラ株式会社 | Image processing apparatus, on-vehicle system, on-vehicle camera, moving body, and image processing method |
| CN110069990A (en) * | 2019-03-18 | 2019-07-30 | 北京中科慧眼科技有限公司 | A kind of height-limiting bar detection method, device and automated driving system |
Non-Patent Citations (1)
| Title |
|---|
| RUIHUA MA等: "A Robust and efficient stereo matching algorithm" * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113378805A (en) * | 2021-08-13 | 2021-09-10 | 北京中科慧眼科技有限公司 | Height limiting device detection method and system based on deep learning and intelligent terminal |
| CN113378805B (en) * | 2021-08-13 | 2021-12-17 | 北京中科慧眼科技有限公司 | Height limiting device detection method and system based on deep learning and intelligent terminal |
| CN113658226A (en) * | 2021-08-26 | 2021-11-16 | 中国人民大学 | Height detection method and system for height limiting device |
| CN113658226B (en) * | 2021-08-26 | 2023-09-05 | 中国人民大学 | Height detection method and system for height limiting device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107738612B (en) | Automatic parking space detection and identification system based on panoramic vision auxiliary system | |
| CN109670376B (en) | Lane line identification method and system | |
| JP6606610B2 (en) | Runway boundary estimation device and travel support system using the same | |
| CN106778593B (en) | Lane level positioning method based on multi-ground sign fusion | |
| CN101075376B (en) | Intelligent video traffic monitoring system based on multi-viewpoints and its method | |
| US11427193B2 (en) | Methods and systems for providing depth maps with confidence estimates | |
| CN103630122B (en) | Monocular vision lane line detection method and distance measurement method thereof | |
| Labayrade et al. | In-vehicle obstacles detection and characterization by stereovision | |
| CN102542843A (en) | Early warning method for preventing vehicle collision and device | |
| CN104021388A (en) | Reversing obstacle automatic detection and early warning method based on binocular vision | |
| CN110069990B (en) | Height limiting rod detection method and device and automatic driving system | |
| CN108764108A (en) | A kind of Foregut fermenters method based on Bayesian inference | |
| CN108319920A (en) | A kind of pavement strip detection and calculation method of parameters scanning three-dimensional point cloud based on line | |
| CN103770704A (en) | System and method for recognizing parking space line markings for vehicle | |
| CN109522847A (en) | A kind of track and road barricade object detecting method based on depth map | |
| CN200990147Y (en) | Intelligent video traffic monitoring system based on multi-view point | |
| JP4032843B2 (en) | Monitoring system and monitoring method, distance correction device and distance correction method in the monitoring system | |
| CN104951747A (en) | Method for determining state of obstruction of at least one camera installed in stereoscopic system | |
| JP2020013573A (en) | Three-dimensional image reconstruction method of vehicle | |
| CN111192283A (en) | Height limiting rod detection and height calculation method | |
| KR100508763B1 (en) | System for detecting and recognizing a car plate area by using stereo vision and method thereof | |
| JP2018055597A (en) | Vehicle type discrimination device and vehicle type discrimination method | |
| US20200193184A1 (en) | Image processing device and image processing method | |
| CN113432615B (en) | Detection method and system based on multi-sensor fusion drivable area and vehicle | |
| JP2016143364A (en) | Position identification equipment, position identification method, and 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 |