CN108108656B - Vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method - Google Patents
Vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method Download PDFInfo
- Publication number
- CN108108656B CN108108656B CN201711134164.3A CN201711134164A CN108108656B CN 108108656 B CN108108656 B CN 108108656B CN 201711134164 A CN201711134164 A CN 201711134164A CN 108108656 B CN108108656 B CN 108108656B
- Authority
- CN
- China
- Prior art keywords
- point
- width
- window
- theta
- platerect
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The invention provides a method for roughly positioning the position of a car window by detecting two angular points below the car window through an SVM (support vector machine), and then accurately positioning the edge of the car window by combining multidirectional projection; by using the method of the invention to perform a positioning operation on the window, the edge of the window can be positioned very accurately, and the positioning can be performed even if the upper edge and the lower edge of the window are inclined.
Description
Technical Field
The invention relates to the field of computer vision and intelligent traffic, in particular to a method for roughly positioning a car window by detecting two angular points below the car window through an SVM (support vector machine), and then accurately positioning the edge of the car window by combining multidirectional projection.
Background
With the continuous development of the field of intelligent transportation, the detection and analysis of the driver behavior in the road traffic safety monitoring system are more and more emphasized at present, including the detection of whether the driver is wearing a safety belt and whether illegal behaviors such as answering a call exist. The driver position must be located for the driver behavior detection, and the accurate location of the window can roughly determine the driver position, so the window location is a very critical step in the driver behavior detection.
Currently, many scholars propose different car window positioning methods, wherein the technical scheme closer to the invention is as follows: the method comprises the following steps that in a document (Panshiji intelligent traffic violation detection algorithm research and a software system to achieve [ D ]. Harbin industry university, 2016 ]), the upper and lower horizontal edges of a car window are positioned by a method of Canny edge detection and linear structural element opening operation, the method does not consider the situation that the car window is inclined when the linear structural element is used for opening operation, and once the upper and lower edges of the car window are slightly inclined, the upper and lower horizontal edges of the car window can be removed after the opening operation is carried out through the linear structural element; the document (study on a window detection technology [ D ]. beijing university of transportation 2012.) proposes that a vehicle is positioned first, then a window position is preliminarily obtained according to the ratio of a vehicle head and a window, Canny edge detection is performed in the approximate area of the window, then horizontal straight line filtering is performed by using template matching, and then detection of the upper and lower boundaries of the window is performed by using Hough transform, wherein the horizontal straight line filtering performed by using the template matching method does not consider the condition that the upper and lower edges of the window are inclined, and the inclined straight line is also filtered after the template matching; the document (yaotong, hannhua, et al, car body color identification method based on car window detection, study [ J ] information communication, 2017(2):87-88.) also uses a Canny edge detection method combined with horizontal line template filtering and Hough line detection car window horizontal band, and there are cases that the car window upper and lower edges cannot be detected when inclined, and the car window upper and lower horizontal bands start to be detected without roughly positioning the car window position, which may possibly detect the horizontal line at the top of the car, resulting in false detection.
In summary, the current car window positioning method has the following disadvantages: (1) no detection is made for the presence of a tilt in the upper and lower edges of the window; (2) horizontal straight lines at the top of the vehicle are easy to detect, and false detection of the edges of the vehicle windows is caused.
Disclosure of Invention
Aiming at the problems in the existing car window positioning method, the invention provides a method for roughly positioning the car window by detecting two angular points below the car window through an SVM (support vector machine), and then accurately positioning the edge of the car window by combining multidirectional projection.
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized by comprising the following steps of:
step 1: carrying out license plate positioning on a color image containing a vehicle, and recording the positioned license plate position as a rectangular area place Rect;
step 2: defining a detection region detectRIOL of a corner point at the lower left corner of a vehicle window and a detection region detectRIOR of a corner point at the lower right corner of the vehicle window according to the license plate position platRect;
and step 3: detecting candidate areas of corner points at the lower left corner of all windows by using an SVM in detectRoil, storing the candidate areas in a linked list as a linked list leftCornersList, detecting candidate areas of corner points at the lower right corner of all windows by using the SVM in detectRoil, storing the candidate areas in the linked list as a linked list rightCornersList;
and 4, step 4: and respectively selecting the final window left lower corner angular point region as winCornerRectL and the final window right lower corner angular point region as winCornerRectR from leftCornerList and rightCornerList, so that the window left lower corner angular point region and the window right lower corner angular point region meet the following conditions: the absolute value of (winCornerRectR.x + winCornerRectR.width/2) - (winCornerRectL.x + winCornerRectL.width/2) is the largest among all candidate regions, and the absolute value of (winCornerRectR.y + winCornerRectR.height/2) - (winCornerRectL.y + winCornerRectL.height/2) is less than max (winCornerRectL.height, winCornerRectR.height);
wherein winCornerRectL.x is the top left-hand abscissa of winCornerRectL, winCornerRectL.width is the width of winCornerRectL, winCornerRectL.y is the top left-hand ordinate of winCornerRectL, winCornerRectL.height is the height of winCornerRectL, winCornerRectR.x is the top left-hand abscissa of winCornerRectR, winCornerRectR.width is the width of winCornerRectR, winCornerRectR.y is the top left-hand ordinate of winCornerRectR, and winCornerRectR.height is the height of winCornerRectR, indicating the greater of the two numbers.
And 5: defining a window left edge detection area winLeftRect, a window right edge detection area winRightRect and a window lower edge detection area winDown Rect according to the positions of the winCornerRectL and the winCornerRectR and the width of the placeRect;
step 6: respectively intercepting a winLeftRect area part, a winLightRect area part and a winDownRect area part in an image to obtain a left window edge detection area image, a right window edge detection area image and a lower window edge detection area image, carrying out image graying, Sobel vertical edge detection and OTSU binaryzation on the intercepted left window edge detection area image and right window edge detection area image, respectively recording the finally obtained binaryzation images as winLeftImg and winLightImg, carrying out image graying, Sobel horizontal edge detection and OTSU binaryzation on the intercepted lower window edge detection area image, and recording the obtained binaryzation image as winDownImg;
and 7: taking the upper left corner of the winLeftImg as the origin of a coordinate system, horizontally rightwards as the positive direction of an x axis, vertically downwards as the positive direction of a y axis, recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is more than or equal to 90 degrees and less than or equal to 60 degrees, projecting the winLeftImg in each theta direction, and recording the angle theta corresponding to the maximum projection value as theta1And recording any point on the projection straight line corresponding to the maximum projection value as point1According to theta1And point1Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l1；
The original image coordinate system is a coordinate system which takes the upper left corner of the original image as the origin of the coordinate system, the horizontal right direction is the positive direction of the x axis, and the vertical downward direction is the positive direction of the y axis;
and 8: taking the upper left corner of the winRightImg as the origin of a coordinate system, horizontally rightwards as the positive direction of an x axis, vertically downwards as the positive direction of a y axis, recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is more than or equal to 60 degrees and less than or equal to 90 degrees, projecting the winRightImg in each theta direction, and recording the angle theta corresponding to the position with the maximum projection value as theta2And recording any point on the projection straight line corresponding to the maximum projection value as point2According to theta2And point2Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l2；
And step 9: taking the upper left corner of the winDownImg as the origin of a coordinate system, the horizontal right direction as the positive direction of an x axis, the vertical downward direction as the positive direction of a y axis, recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is more than or equal to-10 degrees and less than or equal to 10 degrees, projecting the winDownImg in each theta direction, and recording the angle theta corresponding to the position with the maximum projection value as theta3And recording any point on the projection straight line corresponding to the maximum projection value as point3According to theta3And point3Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l3；
Step 10: calculating a straight line l1And l3The intersection point of (c) is recorded as leftPoint, and a straight line l is calculated2And l3The intersection point of (a) is marked as rightPoint;
step 11: defining a vehicle window upper edge detection area winUpRect according to the positions of the winCornerRectL and the winCornerRectR and the positions of the leftPoint and the rightPoint;
step 12: intercepting a winUpRect region part in the image to obtain an image of an upper edge detection region of the vehicle window, performing image graying, Sobel horizontal edge detection and OTSU binaryzation on the intercepted image of the upper edge detection region of the vehicle window, and recording the finally obtained binaryzation image as winUpImg;
step 13: taking the upper left corner of the winUpImg as the origin of a coordinate system, the horizontal right direction as the positive direction of an x axis, and the vertical downward direction as the positive direction of a y axis, and recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is equal to theta3Projecting winUpImg in the direction, scanning from bottom to top in the obtained projection values, stopping scanning until the projection value scanned to a certain position is larger than threshProject, and marking any point on a corresponding projection straight line at the projection value as point4According to theta3And point4Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l4If no position is scanned where the projection value is greater than threshProject, directly order l4Comprises the following steps: y ═ 2- (rightPoint. x-leftPoint. x)/3, where threshProject is a preset projection threshold, leftPoint. x is the abscissa of leftPoint, leftPoint. y is the ordinate of leftPoint, rightPoint. x is the abscissa of rightPoint, and rightPoint. y is the ordinate of rightPoint;
step 14: calculating a straight line l1And l4Is recorded as leftPoint1Calculating a straight line l2And l4Is marked as rightPoint1；
Step 15: according to leftPoint, rightPoint, leftPoint1And rightPoint1The position winRect of the window in the image is calculated.
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in the step 2, a detection region detectroller of a car window lower left corner point and a detection region detectroller of a car window lower right corner point are defined as follows:
2.1)：detectRoiL.x＝plateRect.x-plateRect.width×3；
2.2)：detectRoiL.y＝plateRect.y-plateRect.width×3.5；
2.3)：detectRoiL.width＝plateRect.x+plateRect.width/2-detectRoiL.x；
2.4)：detectRoiL.height＝plateRect.y-plateRect.width-detectRoiL.y；
2.5)：detectRoiR.x＝plateRect.x+plateRect.width/2；
2.6)：detectRoiR.y＝plateRect.y-plateRect.width×3.5；
2.7)：detectRoiR.width＝plateRect.width×3.5；
2.8)：detectRoiR.height＝plateRect.y-plateRect.width-detectRoiR.y。
where platerect.x is the top left abscissa of plateRect, platerect.y is the top left ordinate of plateRect, detectrool.x is the top left abscissa of detectrool, detectrool.y is the top left ordinate of detectrool, detectrool.width is the width of detectrool, detectrool.height is the height of detectrool, detectrool.x is the top left abscissa of detectroor, detectroor.y is the top left ordinate of detectroor, detectroor.width is the width of detectroor, and detectroor.height is the height of detectroor.
The vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method is characterized in that in the step 5, a vehicle window left edge detection area winLeftRect, a vehicle window right edge detection area winRightRect and a vehicle window lower edge detection area winDown Rect are defined as follows:
5.1)：winLeftRect.x＝winCornerRectL.x；
5.2)：winLeftRect.y＝winCornerRectL.y+winCornerRectL.height-plateRect.width/2；
5.3)：winLeftRect.width＝plateRect.width；
5.4)：winLeftRect.height＝plateRect.width/2；
5.5)：winRightRect.x＝winCornerRectR.x+winCornerRectR.width-plateRect.width；
5.6)：winRightRect.y＝winCornerRectR.y+winCornerRectR.height-plateRect.width/2；
5.7)：winRightRect.width＝plateRect.width；
5.8)：winRightRect.height＝plateRect.width/2；
5.9)：winDownRect.x＝winCornerRectL.x；
5.10)：winDownRect.y＝Top；
5.11)：winDownRect.width＝winCornerRectR.x+winCornerRectR.width–winCornerRectL.x；
5.12)：winDownRect.height＝(Buttom–Top+plateRect.width)/2。
wherein, wire left-hand abscissa is the upper left-hand abscissa of wire left, wire left is the upper left-hand ordinate of wire left, wire left is the width of wire left, wire left is the height of wire left, wire right is the upper left-hand abscissa of wire right, wire right is the upper left-hand ordinate of wire right, wire down is the upper left-hand ordinate of wire down, wire down is the lower right, wire right + wire down + wire.
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 7, according to theta1And point1Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system1The steps are as follows:
7.1): order point1.x＝winLeftRect.x+point1.x，point1.y＝winLeftRect.y+point1.y；
7.2): if theta1When the angle is-90 deg., let the equation of straight line l1Comprises the following steps: x ═ point1.x；
7.3): if-90 °<θ1At most-60 deg., let the linear equation l1Comprises the following steps: y is tan (θ)1)×(x-point1.x)+point1.y。
Wherein point1X is point1Abscissa, point of1Y is point1The ordinate of (c).
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 8, the method is based on theta2And point2Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system2The steps are as follows:
8.1): order point2.x＝winRightRect.x+point2.x，point2.y＝winRightRect.y+point2.y；
8.2): if theta2Let the equation of a straight line l be 90 DEG2Comprises the following steps: x ═ point2.x；
8.3): if theta is less than or equal to 60 degrees1<90 DEG, then let the equation of the straight line l2Comprises the following steps: y is tan (θ)2)×(x–point2.x)+point2.y。
Wherein point2X is point2Abscissa, point of2Y is point2The ordinate of (c).
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 9, the method is based on theta3And point3Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system3The steps are as follows:
9.1): order point3.x＝winDownRect.x+point3.x，point3.y＝winDownRect.y+point3.y；
9.2): equation of the straight line l3Comprises the following steps: y is tan (θ)3)×(x–point3.x)+point3.y。
Wherein point3X is point3Abscissa, point of3Y is point3The ordinate of (c).
The vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method is characterized in that in the step 11, a vehicle window upper edge detection area winUpRect is defined as follows:
11.1)：winUpRect.x＝winCornerRectL.x；
11.2)：winUpRect.y＝(leftPoint.y+rightPoint.y)/2–(rightPoint.x-leftPoint.x)/3–
(rightPoint.x-leftPoint.x)/8×3；
11.3)：winUpRect.width＝winCornerRectR.x+winCornerRectR.width–winCornerRectL.x；
11.4)：winUpRect.height＝(rightPoint.x-leftPoint.x)/2。
wherein, winUpRect.x is the horizontal coordinate of the upper left corner of winUpRect, winUpRect.y is the vertical coordinate of the upper left corner of winUpRect, winUpRect.width is the width of winUpRect, and winUpRect.height is the height of winUpRect.
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 13, according to theta3And point4Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system4The steps are as follows:
13.1): order point4.x＝winUpRect.x+point4.x，point4.y＝winUpRect.y+point4.y；
13.2): equation of the straight line l4Comprises the following steps: y is tan (θ)3)×(x–point4.x)+point4.y。
Wherein point4X is point4Abscissa, point of4Y is point4The ordinate of (c).
The car window corner detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 15, the position winRect of a car window in an image is calculated as follows:
15.1)：winRect.x＝leftPoint.x；
15.2)：winRect.y＝(leftPoint1.y+rightPoint1.y)/2；
15.3)：winRect.width＝rightPoint.x-leftPoint.x；
15.4)：winRect.height＝(leftPoint.y+rightPoint.y)/2-(leftPoint1.y+rightPoint1.y)/2。
wherein, winRect.x is the horizontal coordinate of the upper left corner of winRect, winRect.y is the vertical coordinate of the upper left corner of winRect, winRect.width is the width of winRect, and winRect.height is the height of winRect, leftPoint1Y is leftPoint1Ordinate of (1), rightPoint1Y is rightPoint1The ordinate of (c).
By using the method of the invention to perform a positioning operation on the window, the edge of the window can be positioned very accurately, and the positioning can be performed even if the upper edge and the lower edge of the window are inclined.
Drawings
FIG. 1 is an image selected in an embodiment of the present invention;
FIG. 2 is a schematic diagram showing the positions of plateRect, detectRIOL, detectRIOIR, winCornerRectL and winCornerRectR in the example of the present invention;
FIG. 3 is a schematic diagram of the positions of winLeftRect and winLightRect in an embodiment of the present invention;
FIG. 4 is a schematic diagram of positions of winDownRect and winUpRect in the embodiment of the present invention;
FIG. 5 is an image winLeftImg in an embodiment of the present invention;
FIG. 6 is an embodiment of an image winLightImg;
fig. 7 is an image winDownImg in an embodiment of the present invention;
fig. 8 is an image winUpImg in an embodiment of the present invention;
fig. 9 is a diagram of a window positioning result in the embodiment of the present invention.
Detailed Description
The following describes in detail a specific embodiment of a car window corner detection and multidirectional projection-based car window accurate positioning method according to the present invention with reference to an embodiment. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The invention discloses a car window corner point detection and multidirectional projection-based car window accurate positioning method, which comprises the following steps of:
step 1: carrying out license plate positioning on a color image containing a vehicle, and recording the positioned license plate position as a rectangular area plateRect, wherein the selected image is shown in FIG. 1, and the position of the plateRect obtained by license plate positioning is shown in FIG. 2;
step 2: defining a detection region detectroller of a corner point at the lower left corner of a vehicle window and a detection region detectroller of a corner point at the lower right corner of the vehicle window according to a plateRect position, wherein detectroller and detectroller regions defined according to the plateRect position are shown in fig. 2;
and step 3: detecting candidate areas of corner points at the lower left corner of all windows by using an SVM in detectRoil, storing the candidate areas in a linked list as a linked list leftCornersList, detecting candidate areas of corner points at the lower right corner of all windows by using the SVM in detectRoil, storing the candidate areas in the linked list as a linked list rightCornersList;
and 4, step 4: and respectively selecting the final window left lower corner angular point region as winCornerRectL and the final window right lower corner angular point region as winCornerRectR from leftCornerList and rightCornerList, so that the window left lower corner angular point region and the window right lower corner angular point region meet the following conditions: among all candidate regions, the absolute value of (wincorerrectr.x + wincorectr.width/2) - (wincorectr.x + wincorectr.width/2) is the largest, and the absolute value of (wincorectr.y + wincorectr.height/2) - (wincorectr.y + wincorectr.height/2) is smaller than max (wincorectr.height, wincorectr.y + wincorectr.height), and the positions of wincorectr _ and wincorectr _ are selected in this example as shown in fig. 2;
wherein winCornerRectL.x is the top left-hand abscissa of winCornerRectL, winCornerRectL.width is the width of winCornerRectL, winCornerRectL.y is the top left-hand ordinate of winCornerRectL, winCornerRectL.height is the height of winCornerRectL, winCornerRectR.x is the top left-hand abscissa of winCornerRectR, winCornerRectR.width is the width of winCornerRectR, winCornerRectR.y is the top left-hand ordinate of winCornerRectR, winCornerRectR.height is the height of winCornerRectR, max being the greater of the two numbers;
and 5: defining a window left edge detection area winLeftRect, a window right edge detection area winRightRect and a window lower edge detection area winDownRect according to the positions of winCornerRectl and winCornerRectr and the width of the placeRect, wherein the positions of winLeftRect and winrightRect are shown in FIG. 3 and the position of windownRect is shown in FIG. 4;
step 6: respectively intercepting a winLeftRect area part, a winLightRect area part and a winDownRect area part in an image to obtain a left window edge detection area image, a right window edge detection area image and a lower window edge detection area image, performing image graying, Sobel vertical edge detection and OTSU binaryzation on the intercepted left window edge detection area image and right window edge detection area image, respectively recording the finally obtained binaryzation images as winLeftImg and winLightImg, performing image graying, Sobel horizontal edge detection and OTSU binaryzation on the intercepted lower window edge detection area image, and recording the obtained binaryzation images as winDownImg, wherein the obtained winLeftImg image in the embodiment is shown in FIG. 5, the obtained winLightRigh image is shown in FIG. 6, and the obtained winImg image is shown in FIG. 7;
and 7: taking the upper left corner of the winLeftImg as the origin of a coordinate system, horizontally rightwards as the positive direction of an x axis, vertically downwards as the positive direction of a y axis, recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is more than or equal to 90 degrees and less than or equal to 60 degrees, projecting the winLeftImg in each theta direction, and recording the angle theta corresponding to the maximum projection value as theta1And recording any point on the projection straight line corresponding to the maximum projection value as point1According to theta1And point1Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l1；
The original image coordinate system is a coordinate system which takes the upper left corner of the original image as the origin of the coordinate system, the horizontal right direction is the positive direction of the x axis, and the vertical downward direction is the positive direction of the y axis.
And 8: taking the upper left corner of the winRightImg as the origin of a coordinate system, horizontally rightwards as the positive direction of an x axis, vertically downwards as the positive direction of a y axis, recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is more than or equal to 60 degrees and less than or equal to 90 degrees, projecting the winRightImg in each theta direction, and recording the angle theta corresponding to the position with the maximum projection value as theta2And recording any point on the projection straight line corresponding to the maximum projection value as point2According to theta2And point2Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l2；
And step 9: taking the upper left corner of the winDownImg as the origin of the coordinate system, the horizontal right direction as the positive direction of the x axis, the vertical downward direction as the positive direction of the y axis, marking the included angle between the straight line in the coordinate system and the positive direction of the x axis as theta,theta is larger than or equal to-10 degrees and smaller than or equal to 10 degrees, the winDownImg is projected in each theta direction, and the angle theta corresponding to the position with the maximum projection value is recorded as theta3And recording any point on the projection straight line corresponding to the maximum projection value as point3According to theta3And point3Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l3；
Step 10: calculating a straight line l1And l3The intersection point of (c) is recorded as leftPoint, and a straight line l is calculated2And l3The intersection point of (a) is marked as rightPoint;
step 11: defining a vehicle window upper edge detection area winUpRect according to the positions of the winCornerRectL and the winCornerRectR and the positions of the leftPoint and the rightPoint, wherein the positions of the winUpRect are shown in fig. 4 in the embodiment;
step 12: intercepting a winUpRect region part in an image to obtain an image of an upper edge detection region of a vehicle window, performing image graying, Sobel horizontal edge detection and OTSU binaryzation on the intercepted image of the upper edge detection region of the vehicle window, and marking a finally obtained binaryzation image as winUpImg, wherein the obtained winUpImg image is shown in FIG. 8;
step 13: taking the upper left corner of the winUpImg as the origin of a coordinate system, the horizontal right direction as the positive direction of an x axis, and the vertical downward direction as the positive direction of a y axis, and recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is equal to theta3Projecting winUpImg in the direction, scanning from bottom to top in the obtained projection values, stopping scanning until the projection value scanned to a certain position is larger than threshProject, and marking any point on a corresponding projection straight line at the projection value as point4According to theta3And point4Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l4If no position is scanned where the projection value is greater than threshProject, directly order l4Comprises the following steps: y ═ 2- (rightpoint. x-leftPoint. x)/3, where threshProject is a preset projection threshold, in this embodiment, 84 is set, leftPoint. x is the abscissa of leftPoint, and leftPoint. y is lefthe ordinate of tPoint, rightpoint.x is the abscissa of rightPoint, and rightpoint.y is the ordinate of rightPoint;
step 14: calculating a straight line l1And l4Is recorded as leftPoint1Calculating a straight line l2And l4Is marked as rightPoint1；
Step 15: according to leftPoint, rightPoint, leftPoint1And rightPoint1The position of the window in the image winRect is calculated, and the calculated winRect area in the embodiment is shown in fig. 9.
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in the step 2, a detection region detectroller of a car window lower left corner point and a detection region detectroller of a car window lower right corner point are defined as follows:
2.1)：detectRoiL.x＝plateRect.x-plateRect.width×3；
2.2)：detectRoiL.y＝plateRect.y-plateRect.width×3.5；
2.3)：detectRoiL.width＝plateRect.x+plateRect.width/2-detectRoiL.x；
2.4)：detectRoiL.height＝plateRect.y-plateRect.width-detectRoiL.y；
2.5)：detectRoiR.x＝plateRect.x+plateRect.width/2；
2.6)：detectRoiR.y＝plateRect.y-plateRect.width×3.5；
2.7)：detectRoiR.width＝plateRect.width×3.5；
2.8)：detectRoiR.height＝plateRect.y-plateRect.width-detectRoiR.y。
where platerect.x is the top left abscissa of plateRect, platerect.y is the top left ordinate of plateRect, detectrool.x is the top left abscissa of detectrool, detectrool.y is the top left ordinate of detectrool, detectrool.width is the width of detectrool, detectrool.height is the height of detectrool, detectrool.x is the top left abscissa of detectroor, detectroor.y is the top left ordinate of detectroor, detectroor.width is the width of detectroor, and detectroor.height is the height of detectroor.
The vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method is characterized in that in the step 5, a vehicle window left edge detection area winLeftRect, a vehicle window right edge detection area winRightRect and a vehicle window lower edge detection area winDown Rect are defined as follows:
5.1)：winLeftRect.x＝winCornerRectL.x；
5.2)：winLeftRect.y＝winCornerRectL.y+winCornerRectL.height-plateRect.width/2；
5.3)：winLeftRect.width＝plateRect.width；
5.4)：winLeftRect.height＝plateRect.width/2；
5.5)：winRightRect.x＝winCornerRectR.x+winCornerRectR.width-plateRect.width；
5.6)：winRightRect.y＝winCornerRectR.y+winCornerRectR.height-plateRect.width/2；
5.7)：winRightRect.width＝plateRect.width；
5.8)：winRightRect.height＝plateRect.width/2；
5.9)：winDownRect.x＝winCornerRectL.x；
5.10)：winDownRect.y＝Top；
5.11)：winDownRect.width＝winCornerRectR.x+winCornerRectR.width–winCornerRectL.x；
5.12)：winDownRect.height＝(Buttom–Top+plateRect.width)/2。
wherein, wire left-hand abscissa is the upper left-hand abscissa of wire left, wire left is the upper left-hand ordinate of wire left, wire left is the width of wire left, wire left is the height of wire left, wire right is the upper left-hand abscissa of wire right, wire right is the upper left-hand ordinate of wire right, wire down is the upper left-hand ordinate of wire down, wire down is the lower right, wire right + wire down + wire.
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 7, according to theta1And point1Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system1The steps are as follows:
7.1): order point1.x＝winLeftRect.x+point1.x，point1.y＝winLeftRect.y+point1.y；
7.2): if theta1When the angle is-90 deg., let the equation of straight line l1Comprises the following steps: x ═ point1.x；
7.3): if-90 °<θ1At most-60 deg., let the linear equation l1Comprises the following steps: y is tan (θ)1)×(x-point1.x)+point1.y。
Wherein point1X is point1Abscissa, point of1Y is point1The ordinate of (c).
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 8, the method is based on theta2And point2Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system2The steps are as follows:
8.1): order point2.x＝winRightRect.x+point2.x，point2.y＝winRightRect.y+point2.y；
8.2): if theta2Let the equation of a straight line l be 90 DEG2Comprises the following steps: x ═ point2.x；
8.3): if theta is less than or equal to 60 degrees1<90 DEG, then let the equation of the straight line l2Comprises the following steps: y is tan (θ)2)×(x–point2.x)+point2.y。
Wherein point2X is point2Abscissa, point of2Y is point2Sit uprightAnd (4) marking.
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 9, the method is based on theta3And point3Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system3The steps are as follows:
9.1): order point3.x＝winDownRect.x+point3.x，point3.y＝winDownRect.y+point3.y；
9.2): equation of the straight line l3Comprises the following steps: y is tan (θ)3)×(x–point3.x)+point3.y。
Wherein point3X is point3Abscissa, point of3Y is point3The ordinate of (c).
The vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method is characterized in that in the step 11, a vehicle window upper edge detection area winUpRect is defined as follows:
11.1)：winUpRect.x＝winCornerRectL.x；
11.2)：winUpRect.y＝(leftPoint.y+rightPoint.y)/2–(rightPoint.x-leftPoint.x)/3–
(rightPoint.x-leftPoint.x)/8×3；
11.3)：winUpRect.width＝winCornerRectR.x+winCornerRectR.width–winCornerRectL.x；
11.4)：winUpRect.height＝(rightPoint.x-leftPoint.x)/2。
wherein, winUpRect.x is the horizontal coordinate of the upper left corner of winUpRect, winUpRect.y is the vertical coordinate of the upper left corner of winUpRect, winUpRect.width is the width of winUpRect, and winUpRect.height is the height of winUpRect.
The car window corner point detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 13, according to theta3And point4Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system4The steps are as follows:
13.1): order point4.x＝winUpRect.x+point4.x，point4.y＝winUpRect.y+point4.y；
13.2): equation of the straight line l4Comprises the following steps: y is tan (θ)3)×(x–point4.x)+point4.y。
Wherein point4X is point4Abscissa, point of4Y is point4The ordinate of (c).
The car window corner detection and multidirectional projection-based car window accurate positioning method is characterized in that in step 15, the position winRect of a car window in an image is calculated as follows:
15.1)：winRect.x＝leftPoint.x；
15.2)：winRect.y＝(leftPoint1.y+rightPoint1.y)/2；
15.3)：winRect.width＝rightPoint.x-leftPoint.x；
15.4)：winRect.height＝(leftPoint.y+rightPoint.y)/2-(leftPoint1.y+rightPoint1.y)/2。
wherein, winRect.x is the horizontal coordinate of the upper left corner of winRect, winRect.y is the vertical coordinate of the upper left corner of winRect, winRect.width is the width of winRect, and winRect.height is the height of winRect, leftPoint1Y is leftPoint1Ordinate of (1), rightPoint1Y is rightPoint1The ordinate of (c).
In the present embodiment, through the above processing, it can be seen that the window region in fig. 9 is accurately positioned.
Claims (8)
1. A car window corner detection and multidirectional projection-based car window accurate positioning method comprises the following steps:
step 1: carrying out license plate positioning on a color image containing a vehicle, and recording the positioned license plate position as a rectangular area place Rect;
step 2: defining a detection region detectRIOL of a corner point at the lower left corner of a vehicle window and a detection region detectRIOR of a corner point at the lower right corner of the vehicle window according to the license plate position platRect;
and step 3: detecting candidate areas of corner points at the lower left corner of all windows by using an SVM in detectRoil, storing the candidate areas in a linked list as a linked list leftCornersList, detecting candidate areas of corner points at the lower right corner of all windows by using the SVM in detectRoil, storing the candidate areas in the linked list as a linked list rightCornersList;
and 4, step 4: and respectively selecting the final window left lower corner angular point region as winCornerRectL and the final window right lower corner angular point region as winCornerRectR from leftCornerList and rightCornerList, so that the window left lower corner angular point region and the window right lower corner angular point region meet the following conditions: the absolute value of (winCornerRectR.x + winCornerRectR.width/2) - (winCornerRectL.x + winCornerRectL.width/2) is the largest among all candidate regions, and the absolute value of (winCornerRectR.y + winCornerRectR.height/2) - (winCornerRectL.y + winCornerRectL.height/2) is less than max (winCornerRectL.height, winCornerRectR.height);
wherein winCornerRectL.x is the top left-hand abscissa of winCornerRectL, winCornerRectL.width is the width of winCornerRectL, winCornerRectL.y is the top left-hand ordinate of winCornerRectL, winCornerRectL.height is the height of winCornerRectL, winCornerRectR.x is the top left-hand abscissa of winCornerRectR, winCornerRectR.width is the width of winCornerRectR, winCornerRectR.y is the top left-hand ordinate of winCornerRectR, winCornerRectR.height is the height of winCornerRectR, max being the greater of the two numbers;
and 5: defining a window left edge detection area winLeftRect, a window right edge detection area winRightRect and a window lower edge detection area winDown Rect according to the positions of the winCornerRectL and the winCornerRectR and the width of the placeRect;
step 6: respectively intercepting a winLeftRect area part, a winLightRect area part and a winDownRect area part in an image to obtain a left window edge detection area image, a right window edge detection area image and a lower window edge detection area image, carrying out image graying, Sobel vertical edge detection and OTSU binaryzation on the intercepted left window edge detection area image and right window edge detection area image, respectively recording the finally obtained binaryzation images as winLeftImg and winLightImg, carrying out image graying, Sobel horizontal edge detection and OTSU binaryzation on the intercepted lower window edge detection area image, and recording the obtained binaryzation image as winDownImg;
and 7: taking the upper left corner of the winLeftImg as the origin of a coordinate system, horizontally rightwards as the positive direction of an x axis, vertically downwards as the positive direction of a y axis, recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is more than or equal to 90 degrees and less than or equal to 60 degrees, projecting the winLeftImg in each theta direction, and recording the angle theta corresponding to the maximum projection value as theta1And recording any point on the projection straight line corresponding to the maximum projection value as point1According to theta1And point1Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l1；
The original image coordinate system is a coordinate system which takes the upper left corner of the original image as the origin of the coordinate system, the horizontal right direction is the positive direction of the x axis, and the vertical downward direction is the positive direction of the y axis;
and 8: taking the upper left corner of the winRightImg as the origin of a coordinate system, horizontally rightwards as the positive direction of an x axis, vertically downwards as the positive direction of a y axis, recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is more than or equal to 60 degrees and less than or equal to 90 degrees, projecting the winRightImg in each theta direction, and recording the angle theta corresponding to the position with the maximum projection value as theta2And recording any point on the projection straight line corresponding to the maximum projection value as point2According to theta2And point2Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l2；
And step 9: taking the upper left corner of the winDownImg as the origin of a coordinate system, the horizontal right direction as the positive direction of an x axis, the vertical downward direction as the positive direction of a y axis, recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is more than or equal to-10 degrees and less than or equal to 10 degrees, projecting the winDownImg in each theta direction, and recording the angle theta corresponding to the position with the maximum projection value as theta3And recording any point on the projection straight line corresponding to the maximum projection value as point3According to theta3And point3Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l3；
Step 10: calculating a straight line l1And l3The intersection point of (c) is recorded as leftPoint, and a straight line l is calculated2And l3The intersection point of (a) is marked as rightPoint;
step 11: defining a vehicle window upper edge detection area winUpRect according to the positions of the winCornerRectL and the winCornerRectR and the positions of the leftPoint and the rightPoint;
step 12: intercepting a winUpRect region part in the image to obtain an image of an upper edge detection region of the vehicle window, performing image graying, Sobel horizontal edge detection and OTSU binaryzation on the intercepted image of the upper edge detection region of the vehicle window, and recording the finally obtained binaryzation image as winUpImg;
step 13: taking the upper left corner of the winUpImg as the origin of a coordinate system, the horizontal right direction as the positive direction of an x axis, and the vertical downward direction as the positive direction of a y axis, and recording the included angle between a straight line in the coordinate system and the positive direction of the x axis as theta, wherein theta is equal to theta3Projecting winUpImg in the direction, scanning from bottom to top in the obtained projection values, stopping scanning until the projection value scanned to a certain position is larger than threshProject, and marking any point on a corresponding projection straight line at the projection value as point4According to theta3And point4Calculating the corresponding linear equation of the projection straight line in the original image coordinate system and recording as l4If no position is scanned where the projection value is greater than threshProject, directly order l4Comprises the following steps: y ═ 2- (rightPoint. x-leftPoint. x)/3, where threshProject is a preset projection threshold, which is set to 84, leftPoint. x is the abscissa of leftPoint, leftPoint. y is the ordinate of leftPoint, rightPoint. x is the abscissa of rightPoint, and rightPoint. y is the ordinate of rightPoint;
step 14: calculating a straight line l1And l4Is recorded as leftPoint1Calculating a straight line l2And l4Is marked as rightPoint1；
Step 15: according to leftPoint, rightPoint, leftPoint1And rightPoint1The position of the vehicle window in the image is calculatedPlacing winRect;
in step 5, the window left edge detection area winLeftRect, the window right edge detection area winRightRect and the window lower edge detection area winDownRect are defined as follows:
5.1)：winLeftRect.x＝winCornerRectL.x；
5.2)：winLeftRect.y＝winCornerRectL.y+winCornerRectL.height-plateRect.width/2；
5.3)：winLeftRect.width＝plateRect.width；
5.4)：winLeftRect.height＝plateRect.width/2；
5.5)：winRightRect.x＝winCornerRectR.x+winCornerRectR.width-plateRect.width；
5.6)：winRightRect.y＝winCornerRectR.y+winCornerRectR.height-plateRect.width/2；
5.7)：winRightRect.width＝plateRect.width；
5.8)：winRightRect.height＝plateRect.width/2；
5.9)：winDownRect.x＝winCornerRectL.x；
5.10)：winDownRect.y＝Top；
5.11)：winDownRect.width＝winCornerRectR.x+winCornerRectR.width–winCornerRectL.x；
5.12)：winDownRect.height＝(Buttom–Top+plateRect.width)/2；
wherein, wire left-hand abscissa is the upper left-hand abscissa of wire left, wire left is the upper left-hand ordinate of wire left, wire left is the width of wire left, wire left is the height of wire left, wire right is the upper left-hand abscissa of wire right, wire right is the upper left-hand ordinate of wire right, wire down is the upper left-hand ordinate of wire down, wire down is the lower right, wire right + wire down + wire.
2. The car window corner point detection and multidirectional projection-based car window accurate positioning method as claimed in claim 1, wherein: in step 2, a detection region detectroller of the corner point at the lower left corner of the vehicle window and a detection region detectroller of the corner point at the lower right corner of the vehicle window are defined as follows:
2.1)：detectRoiL.x＝plateRect.x-plateRect.width×3；
2.2)：detectRoiL.y＝plateRect.y-plateRect.width×3.5；
2.3)：detectRoiL.width＝plateRect.x+plateRect.width/2-detectRoiL.x；
2.4)：detectRoiL.height＝plateRect.y-plateRect.width-detectRoiL.y；
2.5)：detectRoiR.x＝plateRect.x+plateRect.width/2；
2.6)：detectRoiR.y＝plateRect.y-plateRect.width×3.5；
2.7)：detectRoiR.width＝plateRect.width×3.5；
2.8)：detectRoiR.height＝plateRect.y-plateRect.width-detectRoiR.y；
where platerect.x is the top left abscissa of plateRect, platerect.y is the top left ordinate of plateRect, detectrool.x is the top left abscissa of detectrool, detectrool.y is the top left ordinate of detectrool, detectrool.width is the width of detectrool, detectrool.height is the height of detectrool, detectrool.x is the top left abscissa of detectroor, detectroor.y is the top left ordinate of detectroor, detectroor.width is the width of detectroor, and detectroor.height is the height of detectroor.
3. The car window corner point detection and multidirectional projection-based car window accurate positioning method as claimed in claim 1, wherein: according to theta in step 71And point1Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system1The steps are as follows:
7.1): order point1.x＝winLeftRect.x+point1.x，point1.y＝winLeftRect.y+point1.y；
7.2): if theta1When the angle is-90 deg., let the equation of straight line l1Comprises the following steps: x ═ point1.x；
7.3): if-90 °<θ1At most-60 deg., let the linear equation l1Comprises the following steps: y is tan (θ)1)×(x-point1.x)+point1.y；
Wherein point1X is point1Abscissa, point of1Y is point1The ordinate of (c).
4. The car window corner point detection and multidirectional projection-based car window accurate positioning method as claimed in claim 1, wherein: according to theta in step 82And point2Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system2The steps are as follows:
8.1): order point2.x＝winRightRect.x+point2.x，point2.y＝winRightRect.y+point2.y；
8.2): if theta2Let the equation of a straight line l be 90 DEG2Comprises the following steps: x ═ point2.x；
8.3): if theta is less than or equal to 60 degrees1<90 DEG, then let the equation of the straight line l2Comprises the following steps: y is tan (θ)2)×(x–point2.x)+point2.y；
Wherein point2X is point2Abscissa, point of2Y is point2The ordinate of (c).
5. The car window corner point detection and multidirectional projection-based car window accurate positioning method as claimed in claim 1, wherein: according to theta in step 93And point3Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system3The steps are as follows:
9.1): order point3.x＝winDownRect.x+point3.x，point3.y＝winDownRect.y+point3.y；
9.2): equation of the straight line l3Comprises the following steps: y is tan (θ)3)×(x–point3.x)+point3.y；
Wherein point3X is point3Abscissa, point of3Y is point3The ordinate of (c).
6. The car window corner point detection and multidirectional projection-based car window accurate positioning method as claimed in claim 1, wherein: in step 11, the definition of the vehicle window upper edge detection area winUpRect is as follows:
11.1)：winUpRect.x＝winCornerRectL.x；
11.2)：winUpRect.y＝(leftPoint.y+rightPoint.y)/2–(rightPoint.x-leftPoint.x)/3–(rightPoint.x-leftPoint.x)/8×3；
11.3)：winUpRect.width＝winCornerRectR.x+winCornerRectR.width–winCornerRectL.x；
11.4)：winUpRect.height＝(rightPoint.x-leftPoint.x)/2；
wherein, winUpRect.x is the horizontal coordinate of the upper left corner of winUpRect, winUpRect.y is the vertical coordinate of the upper left corner of winUpRect, winUpRect.width is the width of winUpRect, and winUpRect.height is the height of winUpRect.
7. The car window corner point detection and multidirectional projection-based car window accurate positioning method as claimed in claim 1, wherein: according to theta in step 133And point4Calculating a corresponding linear equation l of the projection straight line in the original image coordinate system4The steps are as follows:
13.1): order point4.x＝winUpRect.x+point4.x，point4.y＝winUpRect.y+point4.y；
13.2): equation of the straight line l4Comprises the following steps: y is tan (θ)3)×(x–point4.x)+point4.y；
Wherein point4X is point4Abscissa, point of4Y is point4The ordinate of (c).
8. The car window corner point detection and multidirectional projection-based car window accurate positioning method as claimed in claim 1, wherein: in step 15, the position winRect of the window in the image is calculated as follows:
15.1)：winRect.x＝leftPoint.x；
15.2)：winRect.y＝(leftPoint1.y+rightPoint1.y)/2；
15.3)：winRect.width＝rightPoint.x-leftPoint.x；
15.4)：winRect.height＝(leftPoint.y+rightPoint.y)/2-(leftPoint1.y+rightPoint1.y)/2；
wherein, winRect.x is the horizontal coordinate of the upper left corner of winRect, winRect.y is the vertical coordinate of the upper left corner of winRect, winRect.width is the width of winRect, and winRect.height is the height of winRect, leftPoint1Y is leftPoint1Ordinate of (1), rightPoint1Y is rightPoint1The ordinate of (c).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711134164.3A CN108108656B (en) | 2017-11-15 | 2017-11-15 | Vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711134164.3A CN108108656B (en) | 2017-11-15 | 2017-11-15 | Vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108108656A CN108108656A (en) | 2018-06-01 |
| CN108108656B true CN108108656B (en) | 2020-07-07 |
Family
ID=62207226
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711134164.3A Active CN108108656B (en) | 2017-11-15 | 2017-11-15 | Vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108108656B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109165689B (en) * | 2018-08-29 | 2021-10-26 | 浙江工业大学 | Accurate positioning method for vehicle window |
| CN110059623B (en) * | 2019-04-18 | 2021-06-11 | 北京字节跳动网络技术有限公司 | Method and apparatus for generating information |
| CN110766756B (en) * | 2019-10-21 | 2022-09-30 | 大连理工大学 | Multi-direction projection-based drop point positioning method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103854029A (en) * | 2014-02-21 | 2014-06-11 | 杭州奥视图像技术有限公司 | Detection method for front automobile window top right corner point |
| CN105447490A (en) * | 2015-11-19 | 2016-03-30 | 浙江宇视科技有限公司 | Vehicle key point detection method based on gradient regression tree and apparatus thereof |
| CN106250824A (en) * | 2016-07-21 | 2016-12-21 | 乐视控股（北京）有限公司 | Vehicle window localization method and system |
| CN106919900A (en) * | 2017-01-19 | 2017-07-04 | 博康智能信息技术有限公司上海分公司 | One kind sets up vehicle window location model and vehicle window localization method and device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9171213B2 (en) * | 2013-03-15 | 2015-10-27 | Xerox Corporation | Two-dimensional and three-dimensional sliding window-based methods and systems for detecting vehicles |
-
2017
- 2017-11-15 CN CN201711134164.3A patent/CN108108656B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103854029A (en) * | 2014-02-21 | 2014-06-11 | 杭州奥视图像技术有限公司 | Detection method for front automobile window top right corner point |
| CN105447490A (en) * | 2015-11-19 | 2016-03-30 | 浙江宇视科技有限公司 | Vehicle key point detection method based on gradient regression tree and apparatus thereof |
| CN106250824A (en) * | 2016-07-21 | 2016-12-21 | 乐视控股（北京）有限公司 | Vehicle window localization method and system |
| CN106919900A (en) * | 2017-01-19 | 2017-07-04 | 博康智能信息技术有限公司上海分公司 | One kind sets up vehicle window location model and vehicle window localization method and device |
Non-Patent Citations (4)
| Title |
|---|
| Moving Vehicle Detection and Tracking in Traffic Images Based on Horizontal Edges;Hongjin Zhu et al;《TELKOMNIKA Indonesian Journal of Electrical Engineering》;20131101(第11期);第6477-6483页 * |
| 基于放射性投影直方图及角点探测的车辆识别与测距研究;凌军;《中国优秀硕士学位论文全文数据库 信息科技辑》;20110415(第04期);全文 * |
| 基于灰度跳变与字符间隔模式的车牌定位方法研究;高飞等;《计算机测量与控制》;20160425;第24卷(第4期);第219-221、225页 * |
| 智能交通违章监测算法研究及软件系统实现;潘世吉;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20170215(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108108656A (en) | 2018-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2031571B1 (en) | Vehicle type determination device, program, and method | |
| US8902053B2 (en) | Method and system for lane departure warning | |
| US10941019B2 (en) | User detection system and image processing device | |
| US8184859B2 (en) | Road marking recognition apparatus and method | |
| KR101517181B1 (en) | System and method for warning lane departure | |
| CN108108656B (en) | Vehicle window corner detection and multidirectional projection-based vehicle window accurate positioning method | |
| US8867792B2 (en) | Environment recognition device and environment recognition method | |
| KR101094752B1 (en) | Lane Classification Method Using Statistical Model of HSI Color Information | |
| US20090110286A1 (en) | Detection method | |
| US20100208071A1 (en) | Vehicle detection device, vehicle detection method, and vehicle detection program | |
| US9928429B2 (en) | Image processing apparatus and image processing method | |
| EP2492868A1 (en) | Far-infrared pedestrian detection device | |
| US20140119609A1 (en) | Image recognizing apparatus and method | |
| CN112052782A (en) | Around-looking-based parking space identification method, device, equipment and storage medium | |
| CN106022242B (en) | Driver's making and receiving calls recognition methods in intelligent transportation system | |
| KR101178508B1 (en) | Vehicle Collision Alarm System and Method | |
| CN108133170B (en) | Window positioning method for multi-directional vehicle | |
| CN112507767B (en) | Face recognition method and related computer system | |
| JP2011248525A (en) | Object detection device and detection method thereof | |
| CN108389177A (en) | A kind of vehicle bumper damage testing method and traffic security early warning method of traffic control | |
| CN108182376B (en) | Vehicle window corner detection-based vehicle window positioning method | |
| JPH07105487A (en) | Drive lane recognition device | |
| JP2893814B2 (en) | Plate cutout device for automatic car number reader | |
| KR102243528B1 (en) | System for monitoring truck and method thereof | |
| CN115565363A (en) | Signal recognition device |
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 |