CN106228531A - Automatic vanishing point scaling method based on horizon search and system - Google Patents
Automatic vanishing point scaling method based on horizon search and system Download PDFInfo
- Publication number
- CN106228531A CN106228531A CN201610492617.9A CN201610492617A CN106228531A CN 106228531 A CN106228531 A CN 106228531A CN 201610492617 A CN201610492617 A CN 201610492617A CN 106228531 A CN106228531 A CN 106228531A
- Authority
- CN
- China
- Prior art keywords
- horizon
- vanishing point
- top view
- straight
- line
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- 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
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30248—Vehicle exterior or interior
- G06T2207/30252—Vehicle exterior; Vicinity of vehicle
- G06T2207/30256—Lane; Road marking
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses automatic vanishing point scaling method based on horizon search and system, method includes: set up horizon situation template;According to horizontal line position, current position, generate the inverse perspective mapping matrix corresponding to described horizon;The top view that described horizon is corresponding is generated by described inverse perspective mapping matrix;Carry out straight-line detection in a top view, it is determined that this horizon is the most correct;If correct, then, by the straight line that detects in top view on inverse perspective mapping to perspective view, take these straight lines intersection point at perspective view as vanishing point;If incorrect, then base area horizontal line situation template, using next horizon as current position horizontal line, continues checking.The present invention utilizes the method for cyclic search, and search pressure is shared time dimension, and calculation resources consumes low.It addition, the present invention is without any priori conditions, can realize full automatic calibration vanishing point, the automatic vanishing point of realization truly is demarcated.
Description
Technical field
The present invention relates to image processing field, particularly to the automatic vanishing point scaling method searched for based on horizon and be
System.
Background technology
The senior drive assist system (ADAS) of view-based access control model algorithm, vanishing point position is the key input of the most every technology
Information, including leading vehicle distance calculating, pedestrian's range estimation, lane detection.The diversity installed due to photographic head, causes vanishing point
Position in acquired image varies, and existing ADAS system is typically determined by artificial demarcation or automatic Calibration
The position of vanishing point.Manually demarcate it is generally required to professional and technical personnel accurately installs, high cost, be not discussed.At automatic Calibration
In the method for vanishing point, common method includes:
1. by full figure detection of straight lines or the method for lane line
This method is many by filtering image by the method for edge filter in artwork, then by Hough transformation detection of straight lines.
In the straight line detected, find intersection two-by-two, in the intersection point of gained, filter out the point in heart region the most in the picture, and with surplus
The mean location of remaining point is as vanishing point.
2. by G-sensor sensor and lane detection
This method is by G-sensor sensor, it is thus achieved that the photographic head angle of pitch, then estimates image according to image-forming principle
Position, horizon.Then utilize lane detection module, take the intersection point of a plurality of lane line as vanishing point.
3. by Foregut fermenters and lane detection
There is a following relation proportionate relationship of front height and photographic head height and position, horizon:
Wherein h0Represent front truck true altitude, hcRepresent photographic head true altitude, vtFor front truck top vertical seat in the picture
Mark, vbFor front truck bottom vertical coordinate in the picture, voHorizon vertical coordinate.
Counter pushing away can obtain:
Owing to car height domain of walker is the least, h0Being approximately 1.5m, remaining variables is also known.Thus can be evaluated whether
Go out horizontal position.Then on the premise of position, known horizon, utilize lane detection module, take a plurality of lane line
Intersection point is as vanishing point.
There is following defect in said method:
In method 1, in the case of horizontal line indefinitely, carrying out straight-line detection in the range of full figure needs to consume in a large number
Calculating resource, and have many interference straight line (such as road surface word, roadside railing etc.) not converge at vanishing point, can be to result
Impact.
In method 2, depending on hardware, the hardware technique of G-sensor can directly affect horizontal estimation result, carefully
Micro-deviation may result in vanishing point and is out of one's reckoning, and process is uncontrollable.
In method 3, depend on Foregut fermenters module, in the case of front is inaccurate without vehicle or vehicle detection,
Horizon can be caused to be out of one's reckoning, thus affect the calibration result of vanishing point.
Summary of the invention
The technical problem to be solved in the present invention is, does not relies on the hardware of Foregut fermenters and G-sensor, automatically goes out
Point is demarcated.
Solve above-mentioned technical problem, the invention provides a kind of automatic vanishing point scaling method based on horizon search, bag
Include:
Set up horizon situation template;
According to horizontal line position, current position, generate the inverse perspective mapping matrix corresponding to described horizon;
The top view that described horizon is corresponding is generated by described inverse perspective mapping matrix;
Carry out straight-line detection in a top view, it is determined that this horizon is the most correct;
If correct, then, by the straight line that detects in top view on inverse perspective mapping to perspective view, take these straight lines and exist
The intersection point of perspective view is as vanishing point;
If incorrect, then base area horizontal line situation template, using next horizon as current position horizontal line, continues checking.
Further, choose 11 alternatively horizontal line positions, set up horizontal line situation template as follows:
{y0、y1、y2、y3、y4、y5、y6、y7、y8、y9、y10, 11 that choose image central region are separated by uniform horizontal line.
Further, the method generating inverse perspective mapping matrix corresponding to described horizon is as follows:
Calculate inverse perspective mapping matrix T-1With horizon center point P0Function;
Tn -1=f-1(yn)
Wherein, P0Vertical coordinate be horizon coordinate yn, abscissa be fixed value be the half of fluoroscopy images width.
Further, carry out straight-line detection result in a top view to be represented by the two of straight line end points:
Straight line the L=((x detectedP,yp),(xQ,yQ)),
Wherein, (xP,yp), (xQ,yQ) it is respectively upper extreme point P and the coordinate of lower extreme point Q, the seat of upper and lower two end points of straight line
Mark can represent line correspondence.
Further, straight-line detection is carried out in a top view, it is determined that when this horizon is the most correct,
If any two straight lines meet during following condition parallel with the angle of x-axis, i.e. on top view, the straight line detected
Angle with x-axis
|α1-α2|≤ε
ε is the threshold value judging the most parallel minimum angles difference of two straight lines.
Further, wherein ε=5 °.
Further, wherein angle with x-axis can be tried to achieve by the extreme coordinates of straight line:
Wherein, xP,ypAnd xQ,yQIt is respectively upper extreme point P and the coordinate of lower extreme point Q.
Further, it is determined that this horizon the most correctly includes following condition:
If having at least 3 frame horizon to meet parallel condition in continuous 10 frames, then it is assumed that horizon is correct.
Present invention also offers a kind of automatic vanishing point calibration system based on horizon search, including:
Modular unit, sets up horizon situation template;
Top view signal generating unit, according to horizontal line position, current position, generates the inverse perspective mapping matrix corresponding to described horizon;
The top view that described horizon is corresponding is generated by described inverse perspective mapping matrix;
Straight-line detection unit, in order to carry out straight-line detection in a top view;
Checking parallel lines unit, the most correct in order to judge this horizon;If correct, then straight by what top view detected
Line, on inverse perspective mapping to perspective view, takes these straight lines intersection point at perspective view as vanishing point;If incorrect, then base area
Horizontal line situation template, using next horizon as current position horizontal line, continues checking.
Further, also include search unit, in order to verify parallel lines unit judge that this horizon is incorrect time, according to
Horizon situation template, using next horizon as current position horizontal line, continues checking.Until being verified.If institute
There is horizon the most not verified, the most again start repeated authentication from first horizon.
Beneficial effects of the present invention:
1) the automatic vanishing point scaling method that the present invention searches for based on horizon, carries out straight-line detection in a top view, it is determined that
This horizon is the most correct;If correct, then, by the horizon in top view in perspective transform to perspective view, take those Horizons
Line intersection point in the perspective is as vanishing point;If incorrect, then continue search for the Horizon of next frame according to horizon situation template
Line position.The present invention utilizes the method for cyclic search, and search pressure is shared time dimension, and calculation resources consumes low.
2) present invention is without any priori conditions, can realize full automatic calibration vanishing point, by setting up position, horizon mould
Plate;According to the current position horizontal line coordinate that detects in the situation template of described horizon, generate corresponding to described horizon is inverse
Perspective transformation matrix;The top view that described horizon is corresponding is generated by described inverse perspective mapping matrix;Carry out in a top view
Straight-line detection, it is determined that this horizon is the most correct.
3) vanishing point in the present invention refers to, dead ahead pole, visual field point at a distance, i.e. horizon and the friendship of sight line dead ahead
Point, vanishing point be in the visual field before and after the convergent point of straight line (lane line) that extends.Senior driving auxiliary at view-based access control model algorithm is
In system (ADAS), vanishing point position is the key input information of the most every technology, includes but not limited to leading vehicle distance calculating, pedestrian
Range estimation, lane detection.So having the biggest value and effect for ADAS.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of the automatic vanishing point scaling method that the present invention searches for based on horizon.
Fig. 2 is to set up horizon situation template simulation schematic diagram in Fig. 1.
Fig. 3 (a) Fig. 3 (c) is the top view schematic diagram generated in Fig. 1.
Fig. 4 is to verify in Fig. 1 that whether horizon is the schematic diagram of parallel lines.
Fig. 5 is the vanishing point method for searching schematic diagram in Fig. 1.
Fig. 6 is the structural representation of the automatic vanishing point calibration system that the present invention searches for based on horizon.
Detailed description of the invention
For making the object, technical solutions and advantages of the present invention clearer, below in conjunction with specific embodiment, and reference
Accompanying drawing, the present invention is described in more detail.
Fig. 1 is the schematic flow sheet of the automatic vanishing point scaling method that the present invention searches for based on horizon.
Step S100 sets up horizon situation template;
Step S101, according to horizontal line position, current position, generates the inverse perspective mapping matrix corresponding to described horizon;
Step S102 generates, by described inverse perspective mapping matrix, the top view that described horizon is corresponding;
Step S103 carries out straight-line detection in a top view,
Step S104 judges that this horizon is the most correct;
If step S105 is correct, then, by the straight line that detects in top view on inverse perspective mapping to perspective view, take this
A little straight lines at the intersection point of perspective view as vanishing point;
If step S106 is incorrect, then base area horizontal line situation template, using next horizon as current position horizontal line, continues
Continuous checking.
Automatic vanishing point scaling method in the present embodiment, it is not necessary to any priori conditions, can realize full automatic calibration vanishing point,
By setting up horizon situation template;According to the current position horizontal line coordinate detected in the situation template of described horizon, generate
Inverse perspective mapping matrix corresponding to described horizon;Corresponding the bowing in described horizon is generated by described inverse perspective mapping matrix
View;Carry out straight-line detection in a top view, it is determined that this horizon is the most correct.
Preferred as in the present embodiment, chooses 11 alternatively horizontal line positions in described step S100, sets up as follows
Horizon situation template:
{y0、y1、y2、y3、y4、y5、y6、y7、y8、y9、y10}。
Preferred as in the present embodiment, in described step S100 step S101 generate corresponding to described horizon inverse
The method of perspective transformation matrix is as follows:
Calculate inverse perspective mapping matrix T-1With horizon center point P0Function;
Tn -1=f-1(yn)
Wherein, P0Vertical coordinate be horizon coordinate yn, abscissa be fixed value be the half of fluoroscopy images width.
Preferred as in the present embodiment, carries out straight-line detection result the most in a top view by straight line
Two end points represent:
Straight line the L=((x detectedP,yp),(xQ,yQ)),
Wherein, (xP,yp), (xQ,yQ) it is respectively upper extreme point P and the coordinate of lower extreme point Q.
Preferred as in the present embodiment, carries out straight-line detection, it is determined that this horizon in step S104 in a top view
Time the most correct,
If any two straight lines meet during following condition parallel with the angle of x-axis,
|α1-α2|≤ε
ε is the threshold value judging the most parallel minimum angles difference of two straight lines, wherein ε=5 °.
Preferred as in the present embodiment, wherein the angle with x-axis can be tried to achieve by the extreme coordinates of straight line:
Wherein, xP,ypAnd xQ,yQIt is respectively upper extreme point P and the coordinate of lower extreme point Q.
Preferred as in the present embodiment, judges in step S105 that this horizon the most correctly includes following condition:
If having at least 3 frame horizon to meet parallel condition in continuous 10 frames, then it is assumed that horizon is correct.
Inventive principle:
First, utilize the functional relationship of top view inverse perspective mapping and horizon coordinate, generate corresponding to the horizontal line of current position
Inverse perspective mapping matrix T-1.In continuous 10 two field pictures afterwards, use this T-1Generate top view, and carry out straight-line detection.If 10
Have more than the straight line that 3 frames detect in frame and present parallel relation, then exit search.If exit criteria cannot be met, then change ground
Horizontal line, repeat the above steps continues checking, until exiting search.After searching for successfully, the parallel lines in top view is entered
Row perspective transform, takes its intersection point in the perspective as vanishing point.
Specifically, the concept of search is incorporated in vanishing point demarcation by the present invention, position, cyclic search horizon.
As in figure 2 it is shown, be that Fig. 1 sets up horizon situation template simulation schematic diagram.In figure, the line of labelling covers the biggest
Position in the picture, horizon under part photographic head installation situation.Vertical coordinate corresponding to every horizon is designated as yn.From y0Open
Begin, each horizon is carried out the checking of continuous 10 frames, if not met exit criteria, then jump to next horizon and continue
Checking.If searching y10The most do not exit search, then continue from y0Restart.
Owing to, in actual application scenarios, the image that photographic head catches is in real-time change, the information that each two field picture comprises
Amount is different from.Further, the demarcation of vanishing point is not required at once completing, it is only necessary to complete the most as precisely as possible
Demarcate.So, the present invention does not travel through all of search target at same two field picture, but have employed the side of cyclic search
Method, has shared search pressure time dimension, every 10 one horizon of frame checking, will not produce a large amount of computing in a frame.
1). top view generates
The first step of horizon proof procedure is to generate top view, and top view is by the lane line region of Visual Angle in Perspective image
Be converted to overlook visual angle by inverse perspective mapping.
Calculate inverse perspective mapping matrix T-1Method the most highly developed, just have in OPENCV correspondence function, at this
No longer introduce ins and outs.The method that base area horizontal line central point calculates inverse perspective mapping matrix is the most highly developed, and the present invention is not
Reinflated introduction.
The most only need to know T-1Being the function of horizon center point P 0, the vertical coordinate of P0 is horizon coordinate yn, abscissa
For the half of fluoroscopy images width, it it is fixed value.So, Tn -1Only relevant to yn.
Tn -1=f-1(yn)
Fig. 3 (a) Fig. 3 (c) is the top view schematic diagram generated in Fig. 1, and wherein Fig. 3 (a) Fig. 3 (c) is y4 respectively,
The top view that y5, y6 horizon is corresponding.
2). straight-line detection
Carry out straight-line detection in a top view, owing to the size of top view is more much smaller than artwork, so the speed of straight-line detection
Degree is quickly.The method of straight-line detection is very universal, such as Hough transformation and the method for enhancing projection, technology is the most no longer described in detail in detail thin
Joint.
The straight line L detected, with upper extreme point P and the coordinate representation of lower extreme point Q:
L=((xP,yP),(xQ,yQ))
3). checking parallel lines
From top view above, if current position horizontal line is correct, then converge at vanishing point (example in Visual Angle in Perspective
Such as lane line) straight line overlook in visual angle in parallel.Utilize this priori, it is possible to determine that horizon is the most just
Really.
As shown in Figure 4, it is that Fig. 1 verifying, whether horizon is the schematic diagram of parallel lines, if any two straight lines and x-axis
Angle meets following condition, the most parallel.
|α1-α2|≤ε
Through great many of experiments, ε=5 °.
The angle of straight line and x-axis can be tried to achieve by the extreme coordinates of straight line:
If having more than 3 frames or 3 frames to meet parallel condition in continuous 10 frames, then it is assumed that horizon is correct, exits search
Flow process.
Finally, by the parallel lines in top view in perspective transform to perspective view.Perspective transform is inverse perspective mapping
Inverse transformation, in OPENCV, have respective function, be not described in detail in this.
Being the vanishing point method for searching schematic diagram in Fig. 1 as shown in Figure 5, the inverse transformation making inverse perspective mapping obtains perspective change
Change:
(P'x,y,Q'x,y)=f (Px,y,Qx,y)
Taking those horizon intersection point in the perspective as vanishing point, the computational methods of vanishing point coordinate can be according to following public
Formula:
It is the structural representation of the automatic vanishing point calibration system that the present invention searches for based on horizon as shown in Figure 6.
Automatic vanishing point calibration system 10 based on horizon search in the present embodiment, including:
Modular unit 1, sets up horizon situation template;
Top view signal generating unit 2, according to horizontal line position, current position, generates the inverse perspective mapping square corresponding to described horizon
Battle array;The top view that described horizon is corresponding is generated by described inverse perspective mapping matrix;
Straight-line detection unit 3, in order to carry out straight-line detection in a top view;
Checking parallel lines unit 4, the most correct in order to judge this horizon;If correct, then will top view detect
Straight line, on inverse perspective mapping to perspective view, takes these straight lines intersection point at perspective view as vanishing point;If incorrect, then basis
Horizon situation template, using next horizon as current position horizontal line, continues checking.
Preferred as the present embodiment, automatic vanishing point calibration system 10 also includes search unit 5, in order at checking parallel lines
Unit judges when this horizon is incorrect, base area horizontal line situation template, using next horizon as current position horizontal line, continues
Checking.Until being verified.If all horizon are the most not verified, the most again start from first horizon to repeat
Checking.
Automatic vanishing point calibration system in the present embodiment, the method utilizing cyclic search in search unit 5, search is pressed
Power shares time dimension, and calculation resources consumes low.
Those of ordinary skill in the field it is understood that more than, described be only the present invention specific embodiment, and
Be not used in the restriction present invention, all within the spirit and principles in the present invention, any modification, equivalent substitution and improvement etc. done, all
Within protection scope of the present invention should being included in.
Claims (10)
1. an automatic vanishing point scaling method based on horizon search, it is characterised in that including:
Set up horizon situation template;
According to horizontal line position, current position, generate the inverse perspective mapping matrix corresponding to described horizon;
The top view that described horizon is corresponding is generated by described inverse perspective mapping matrix;
Carry out straight-line detection in a top view, it is determined that this horizon is the most correct;
If correct, then, by the straight line that detects in top view on inverse perspective mapping to perspective view, take these straight lines in perspective
The intersection point of figure is as vanishing point;
If incorrect, then base area horizontal line situation template, using next horizon as current position horizontal line, continues checking.
Automatic vanishing point scaling method the most according to claim 1, it is characterised in that choose 11 alternatively horizontal line positions,
Set up horizontal line situation template as follows:
{y0、y1、y2、y3、y4、y5、y6、y7、y8、y9、y10}。
Automatic vanishing point scaling method the most according to claim 1, it is characterised in that generate corresponding to described horizon is inverse
The method of perspective transformation matrix is as follows:
Calculate inverse perspective mapping matrix T-1With horizon center point P0Function;
Tn -1=f-1(yn)
Wherein, P0Vertical coordinate be horizon coordinate yn, abscissa be fixed value be the half of fluoroscopy images width.
Automatic vanishing point scaling method the most according to claim 1, it is characterised in that carry out straight-line detection knot in a top view
Fruit is represented by the two of straight line end points:
Straight line the L=((x detectedP,yp),(xQ,yQ)),
Wherein, (xP,yp), (xQ,yQ) it is respectively upper extreme point P and the coordinate of lower extreme point Q.
Automatic vanishing point scaling method the most according to claim 1, it is characterised in that carry out straight-line detection in a top view,
Judge when this horizon is the most correct,
If any two straight lines meet during following condition parallel with the angle of x-axis,
|α1-α2|≤ε
ε is the threshold value judging the most parallel minimum angles difference of two straight lines.
Automatic vanishing point scaling method the most according to claim 5, it is characterised in that wherein ε=5 °.
7. the automatic vanishing point scaling method stated according to claim 5, it is characterised in that wherein the angle with x-axis can pass through straight line
Extreme coordinates try to achieve:
Wherein, xP, ypAnd xQ, yQIt is respectively upper extreme point P and the coordinate of lower extreme point Q.
Automatic vanishing point scaling method the most according to claim 1, it is characterised in that judge that this horizon the most correctly includes
Following condition:
If having at least 3 frame horizon to meet parallel condition in continuous 10 frames, then it is assumed that horizon is correct.
9. an automatic vanishing point calibration system based on horizon search, it is characterised in that including:
Modular unit, sets up horizon situation template;
Top view signal generating unit, according to horizontal line position, current position, generates the inverse perspective mapping matrix corresponding to described horizon;Pass through
Described inverse perspective mapping matrix generates the top view that described horizon is corresponding;
Straight-line detection unit, in order to carry out straight-line detection in a top view;
Checking parallel lines unit, the most correct in order to judge this horizon;If correct, then the straight line warp that will detect in top view
Cross inverse perspective mapping to perspective view, take these straight lines intersection point at perspective view as vanishing point;If incorrect, then base area horizontal line
Situation template, using next horizon as current position horizontal line, continues checking.
10. according to the calibration system described in 9 claim 1, it is characterised in that also include search unit, in order to parallel in checking
Line unit judges when this horizon is incorrect, base area horizontal line situation template, using next horizon as current position horizontal line, continues
Continuous checking.Until being verified.If all horizon are the most not verified, the most again start weight from first horizon
Review card.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610492617.9A CN106228531B (en) | 2016-06-27 | 2016-06-27 | Automatic vanishing point calibration method and system based on horizon line search |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610492617.9A CN106228531B (en) | 2016-06-27 | 2016-06-27 | Automatic vanishing point calibration method and system based on horizon line search |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106228531A true CN106228531A (en) | 2016-12-14 |
CN106228531B CN106228531B (en) | 2020-03-20 |
Family
ID=57519965
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610492617.9A Active CN106228531B (en) | 2016-06-27 | 2016-06-27 | Automatic vanishing point calibration method and system based on horizon line search |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106228531B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107316331A (en) * | 2017-08-02 | 2017-11-03 | 浙江工商大学 | For the vanishing point automatic calibration method of road image |
CN111077893A (en) * | 2019-12-31 | 2020-04-28 | 驭势科技(南京)有限公司 | Navigation method based on multiple vanishing points, electronic equipment and storage medium |
CN111174796A (en) * | 2019-12-31 | 2020-05-19 | 驭势科技(南京)有限公司 | Navigation method based on single vanishing point, electronic equipment and storage medium |
CN111210411A (en) * | 2019-12-31 | 2020-05-29 | 驭势科技(南京)有限公司 | Detection method of vanishing points in image, detection model training method and electronic equipment |
CN113409235A (en) * | 2020-03-17 | 2021-09-17 | 杭州海康威视数字技术股份有限公司 | Vanishing point estimation method and device |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102103747A (en) * | 2009-12-16 | 2011-06-22 | 中国科学院电子学研究所 | Method for calibrating external parameters of monitoring camera by adopting reference height |
CN102324017A (en) * | 2011-06-09 | 2012-01-18 | 中国人民解放军国防科学技术大学 | FPGA (Field Programmable Gate Array)-based lane line detection method |
CN105644785A (en) * | 2015-12-31 | 2016-06-08 | 哈尔滨工业大学 | Unmanned aerial vehicle landing method based on optical flow method and horizon line detection |
-
2016
- 2016-06-27 CN CN201610492617.9A patent/CN106228531B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102103747A (en) * | 2009-12-16 | 2011-06-22 | 中国科学院电子学研究所 | Method for calibrating external parameters of monitoring camera by adopting reference height |
CN102324017A (en) * | 2011-06-09 | 2012-01-18 | 中国人民解放军国防科学技术大学 | FPGA (Field Programmable Gate Array)-based lane line detection method |
CN105644785A (en) * | 2015-12-31 | 2016-06-08 | 哈尔滨工业大学 | Unmanned aerial vehicle landing method based on optical flow method and horizon line detection |
Non-Patent Citations (4)
Title |
---|
RADU DANESCU 等: "Detection and Classification of Painted Road Objects for Intersection Assistance Applications", 《2010 13TH INTERNATIONAL IEEE ANNUAL CONFERENCE ON INTELLIGENT TRANSPORTATION SYSTEMS》 * |
李勃 等: "路况PTZ 摄像机自动标定方法", 《北京邮电大学学报》 * |
焦欣欣 等: "基于单目视觉的结构化道路检测算法研究", 《现代计算机》 * |
王宝锋 等: "基于线性逼近的车道线弯道识别方法", 《北京理工大学学报》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107316331A (en) * | 2017-08-02 | 2017-11-03 | 浙江工商大学 | For the vanishing point automatic calibration method of road image |
CN107316331B (en) * | 2017-08-02 | 2020-04-14 | 浙江工商大学 | Vanishing point automatic calibration method for road image |
CN111077893A (en) * | 2019-12-31 | 2020-04-28 | 驭势科技(南京)有限公司 | Navigation method based on multiple vanishing points, electronic equipment and storage medium |
CN111174796A (en) * | 2019-12-31 | 2020-05-19 | 驭势科技(南京)有限公司 | Navigation method based on single vanishing point, electronic equipment and storage medium |
CN111210411A (en) * | 2019-12-31 | 2020-05-29 | 驭势科技(南京)有限公司 | Detection method of vanishing points in image, detection model training method and electronic equipment |
CN111174796B (en) * | 2019-12-31 | 2022-04-29 | 驭势科技(浙江)有限公司 | Navigation method based on single vanishing point, electronic equipment and storage medium |
CN111210411B (en) * | 2019-12-31 | 2024-04-05 | 驭势科技(浙江)有限公司 | Method for detecting vanishing points in image, method for training detection model and electronic equipment |
CN113409235A (en) * | 2020-03-17 | 2021-09-17 | 杭州海康威视数字技术股份有限公司 | Vanishing point estimation method and device |
CN113409235B (en) * | 2020-03-17 | 2023-08-22 | 杭州海康威视数字技术股份有限公司 | Vanishing point estimation method and apparatus |
Also Published As
Publication number | Publication date |
---|---|
CN106228531B (en) | 2020-03-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106228531A (en) | Automatic vanishing point scaling method based on horizon search and system | |
CN110501018B (en) | Traffic sign information acquisition method for high-precision map production | |
Oniga et al. | Processing dense stereo data using elevation maps: Road surface, traffic isle, and obstacle detection | |
WO2020007189A1 (en) | Obstacle avoidance notification method and apparatus, electronic device, and readable storage medium | |
CN106156723A (en) | A kind of crossing fine positioning method of view-based access control model | |
CN102208019B (en) | Method for detecting lane change of vehicle based on vehicle-mounted camera | |
CN103218605B (en) | A kind of fast human-eye positioning method based on integral projection and rim detection | |
CN106056100A (en) | Vehicle auxiliary positioning method based on lane detection and object tracking | |
WO2021051344A1 (en) | Method and apparatus for determining lane lines in high-precision map | |
CN101833791A (en) | Scene modeling method under single camera and system | |
CN112329776B (en) | License plate detection method and device based on improved CenterNet network | |
CN110031829A (en) | A kind of targeting accuracy distance measuring method based on monocular vision | |
CN101211411A (en) | Human body detection process and device | |
WO2023240805A1 (en) | Connected vehicle overspeed early warning method and system based on filtering correction | |
CN106384355A (en) | Automatic calibration method applied to projection interactive system | |
CN112798811A (en) | Speed measurement method, device and equipment | |
CN103065163B (en) | A kind of fast target based on static images detects recognition system and method | |
CN106157219A (en) | Road axis extracting method based on vehicle-mounted scanning system and device | |
CN104574993A (en) | Road monitoring method and device | |
CN110147748A (en) | A kind of mobile robot obstacle recognition method based on road-edge detection | |
CN112446915B (en) | Picture construction method and device based on image group | |
CN111160132B (en) | Method and device for determining lane where obstacle is located, electronic equipment and storage medium | |
CN108573280A (en) | A kind of unmanned boat independently passes through the method for bridge | |
CN115273005A (en) | Visual navigation vehicle environment perception method based on improved YOLO algorithm | |
CN103134476B (en) | Sea and land boundary detection method based on level set algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | 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 |