CN106856004A - A kind of camera marking method - Google Patents

Abstract

The invention provides a kind of camera marking method, by the use of road surface upper graticule as object of reference, geometrical model between pixel column in traffic video and road surface actual range is taken out according to video camera imaging principle, geometrical model is parsed, obtain pixel column and actual road surface on video image apart from mapping relations.Without considering intrinsic parameters of the camera, geometry derivation goes out external parameter to the method, simple and easy to apply, and amount of calculation is small.

Description

A kind of camera marking method

Technical field

The present invention relates to a kind of camera marking method, a kind of more particularly to a kind of video camera suitable for traffic video Scaling method.

Background technology

Video detection technology has important application in transport information detection, main to utilize image processing techniques to obtaining To traffic video image carry out a series for the treatment of, extract useful information, complete traffic parameter, traffic and traffic thing The detection of part, is a hot research direction in ITS (Intelligent Transportation System).Video is examined In the application in the fields such as photogrammetric, computer vision, a fundamental problem is the problem of calibrating of video camera to survey technology.In reality In the image/video treatment on border for the corresponding points on the three-dimensional coordinate and the image that determine real space object mapping relations, it is necessary to By setting up corresponding geometrical model, the parameter of geometrical model is solved, the parameter of geometrical model is exactly camera parameters, this ginseng Several solution procedurees turns into the demarcation of video camera.It is most basic and most important during the demarcation of video camera in video detection technology A step, its precision and the degree of accuracy directly affect the result of follow-up work.Therefore the degree of accuracy and calculation of camera calibration are improved The succinct degree of method is the emphasis place of research work.

The demarcation of video camera can be taken under specific experimental condition, based on object of reference known to shape, size, utilized Image converts the method with mathematical computations to obtain the parameter of video camera, as traditional camera marking method.Such demarcation Method substantially has following several：Using the scaling method of optimization algorithm, Typical Representative has the conventional method in photogrammetry And linear transformation method；Using the method for the demarcation of video camera transformation matrix；Consider the two-step method of distortion compensation；Biplane side Method；Improved Zhang Zhengyou standardizations etc..It is corresponding with image that these methods obtain real space object using specific object of reference The mapping relations of point, algorithm is simple, but has certain experimental condition requirement.

Further, it is also possible to take that object of reference is independent of, using the method for the corresponding relation between object on different two field pictures Demarcated, i.e. self-calibrating method.The main shooting having using absolute conic and polar curve Transformation Properties solution Kruppa equations Machine self-calibrating method；Based on quadric self-calibrating method；Camera self-calibration technology based on active vision etc..Such side Method is flexibly practical, but precision needs further raising.

The content of the invention

The technical problem to be solved in the present invention is to provide a kind of simple and easy to apply, small camera marking method of amount of calculation.

The technical solution adopted by the present invention is as follows：A kind of camera marking method, specific method is：Using road surface upper graticule As object of reference, the geometry between the pixel column in traffic video and road surface actual range is taken out according to video camera imaging principle Model, parses to geometrical model, obtain pixel column and actual road surface on video image apart from mapping relations.

The specific method for setting up geometry module is：

Geometrical model benchmark is set up, on the basis of the video original image that the camera acquisition for being set up on road surface by road is arrived, is drawn The distance of the distance between fixed N number of point, the fixed point of picture on actual road surface is equal, forms N-1 pixel fragment；Take each pixel fragment Lower extreme point separately down with it is upper extension representative image on pixel fragment change；The N is more than or equal to 3；

Set up geometrical model, with the lower extreme point of each pixel fragment separately down with it is upper extension representative image on pixel fragment shape Into the individual different domain of 2 (N-1).

The N=3,3 points form two pixel fragments m and n；Their distances on actual road surface are equal.Take n sections Lower breakpoint separately down with upper extension representative image on pixel fragment change, take the upper extreme point of m separately down with upper extension The change of the pixel fragment in representative image, forms 4 different domain：Region is 1., 2., 3. and 4.；If L is extension pixel Section, K is the variable quantity of the corresponding road surface actual range when L is moved.

M, n represent two pixel fragments on video image, and the corresponding actual range of the two pixel fragments is equal, it is assumed that be all A；Region 1. in, L represents the pixel fragment for being extended downwardly on the basis of the lower extreme point of n on the video images；Region 2. in, L The pixel fragment that expression is upwardly extended on the basis of the lower extreme point of n on the video images；Region 3. in, L is represented in video image On the pixel fragment that is extended downwardly on the basis of the upper extreme point of m；Region 4. in L represent and be with the upper extreme point of m on the video images The pixel fragment that benchmark is upwardly extended.K is the variable quantity of the corresponding road surface actual range when L is moved, the road surface representated by m and n Actual range is equal, and K also follows suit and change accordingly while L changes.

It is to the specific method that geometrical model is analyzed：Looked on video on the road surface corresponding to point M, M a point Point is N, and the actual range NA in video image corresponding to pixel distance MD, O point are the focus of video camera, and NC is in actual road surface Lane line, MF be video camera shoot traffic video image in corresponding lane line, ON, OA, OB and OC are video camera pipelines, D, E and F are three points of selection, and n=DE, m=EF, corresponding actual road surface are respectively AB and BC, wherein AB=BC=h；It is right 1. region is asked for, and is obtained

I.e.：

Similarly, the result for asking for other several regions is：

Region is 2.：

Region is 3.：

Region is 4.：

Compared with prior art, the beneficial effects of the invention are as follows：By the use of road surface upper graticule as object of reference, there is provided a kind of line Video camera geometric calibration method under property model, the method takes out geometrical model, extrapolates figure according to video camera imaging principle The functional relation of corresponding road surface actual range change when changing as pixel fragment, without considering intrinsic parameters of the camera, geometry External parameter is derived, simple and easy to apply, amount of calculation is small.

Brief description of the drawings

Fig. 1 is the geometrical model schematic diagram of a wherein embodiment of the invention.

Fig. 2 is the geometry derivation schematic diagram in embodiment illustrated in fig. 1.

Specific embodiment

In order to make the purpose , technical scheme and advantage of the present invention be clearer, it is right below in conjunction with drawings and Examples The present invention is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the present invention, not For limiting the present invention.

Any feature disclosed in this specification (including summary and accompanying drawing), unless specifically stated otherwise, can be equivalent by other Or the alternative features with similar purpose are replaced.I.e., unless specifically stated otherwise, each feature is a series of equivalent or class An example in like feature.

As shown in figure 1, geometrical model benchmark is set up, the video original image that the camera acquisition for being set up on road surface by road is arrived On the basis of, to draw and determine 3 points, distance of the distance between the fixed point of picture on actual road surface is equal, forms 2 pixel fragments m and n； Take n sections of lower breakpoint separately down with upper extension representative image on pixel fragment change, take the upper extreme point of m separately down with it is upper Extend the change of the pixel fragment in representative image, form 4 different domain：Region is 1., 2., 3. and 4.；If L is extension Pixel fragment, K is the variable quantity of the corresponding road surface actual range when L is moved.

M, n represent two pixel fragments on video image, and the corresponding actual range of the two pixel fragments is equal, it is assumed that be all A；Region 1. in, L represents the pixel fragment for being extended downwardly on the basis of the lower extreme point of n on the video images；Region 2. in, L The pixel fragment that expression is upwardly extended on the basis of the lower extreme point of n on the video images；Region 3. in, L is represented in video image On the pixel fragment that is extended downwardly on the basis of the upper extreme point of m；Region 4. in L represent and be with the upper extreme point of m on the video images The pixel fragment that benchmark is upwardly extended.K is the variable quantity of the corresponding road surface actual range when L is moved, the road surface representated by m and n Actual range is equal, and K also follows suit and change accordingly while L changes.

As Fig. 2 at any time, the point looked on video on the road surface corresponding to point M, M a point be N, in video image pixel away from It is the focus of video camera from actual range NA, the O point corresponding to MD, NC is the lane line in actual road surface, and MF is clapped for video camera Corresponding lane line in the traffic video image taken the photograph, ON, OA, OB and OC are video camera pipelines, and D, E and F are three points of selection, N=DE, m=EF, corresponding actual road surface are respectively AB and BC, wherein AB=BC=h.Do two boost lines：AG and NH, and AG MF is respectively parallel to NH.Known conditions is：Known to DE, EF and MD pixel distance；AB=BC, and it is all known；MF parallel to NH is parallel to AG.Seek the actual range of NA.Solution procedure is as follows：

1. region is asked for,

By MF parallel to AG, draw

Can draw

By MF parallel to NH, draw

Can draw

Bring AK, NP and QH into following two formula：

By Δ ABK~Δ NBQ, drawI.e.

By Δ ACG~Δ NCH, drawI.e.

Obtain

Now, in two relational expressions only have two variables, NA andThis linear equation in two unknowns is solved to obtain

I.e.：

Similarly, the result for asking for other several regions is：

Region is 2.：

Region is 3.：

Region is 4.：

Claims (4)

1. a kind of camera marking method, specific method is：It is former according to video camera imaging by the use of road surface upper graticule as object of reference Reason takes out the geometrical model between the pixel column in traffic video and road surface actual range, and geometrical model is parsed, and obtains Pixel column and actual road surface are apart from mapping relations on to video image.

2. camera marking method according to claim 1, the specific method for setting up geometry module is：

Geometrical model benchmark is set up, on the basis of the video original image that the camera acquisition for being set up on road surface by road is arrived, draws fixed N number of The distance of the distance between point, the fixed point of picture on actual road surface is equal, forms N-1 pixel fragment；Take under each pixel fragment End points changes with the pixel fragment in upper extension representative image separately down；The N is more than or equal to 3；

Geometrical model is set up, 2 are formed with the pixel fragment in upper extension representative image separately down with the lower extreme point of each pixel fragment (N-1) individual different domain.

3. camera marking method according to claim 2, the N=3,3 points form two pixel fragments m and n；Take n The lower breakpoint of section separately down with upper extension representative image on pixel fragment change, the upper extreme point for taking m prolongs separately down and above The change of the pixel fragment in representative image is stretched, 4 different domain are formed：Region is 1., 2., 3. and 4.；If L is extension picture Plain section, K is the variable quantity of the corresponding road surface actual range when L is moved.

4. camera marking method according to claim 3, be to the specific method that geometrical model is analyzed：In video On the point looked on the road surface corresponding to point M, M a point be N, actual range NA, O in video image corresponding to pixel distance MD Point is the focus of video camera, and NC is the lane line in actual road surface, corresponding in the traffic video image that MF shoots for video camera Lane line, ON, OA, OB and OC are video camera pipelines, and D, E and F are three points of selection, n=DE, m=EF, corresponding actual road Face is respectively AB and BC, wherein AB=BC=h；1. region is asked for, is obtained

$NA=\frac{2\·AB\·EF\·MD}{EF\·DE+DE\·DE+MD\·DE-MD\·EF}$

I.e.： $K=\frac{2hmL}{mn+n^2+nL-mL}$

Similarly, the result for asking for other several regions is：

Region is 2.： $K=\frac{2hmL}{mn+n^2+mL-nL}$

Region is 3.： $K=\frac{2hnL}{mn+m^2+nL-mL}$

Region is 4.： $K=\frac{2hnL}{mn+m^2+mL-nL}.$