CN103413135A - Method, device and system for detecting brightness and darkness border line of vehicle headlight - Google Patents

Abstract

The invention relates to the technical field of image processing, in particular to a method for detecting a brightness and darkness border line of a vehicle headlight. The method for detecting the brightness and darkness border line of the vehicle headlight comprises the following steps that a lighting image of the vehicle headlight is obtained; edge detection is conducted on the lighting image, so that an edge image is obtained; a first line segment set located in a first direction is obtained through detection from the edge image, a second line segment set located in a second direction is obtained through detection from the edge image, and a candidate brightness and darkness border line set meeting the preset conditions is obtained based on the first line segment set and the second line segment set, wherein the first direction corresponds to the horizontal brightness and darkness border line direction, the second direction corresponds to the inclining brightness and darkness border line direction, and one candidate brightness and darkness border line is composed of one line segment of the first line segment set and one line segment of the second line segment set; based on the characteristics of light beams of the vehicle headlight, the brightness and darkness border line of the vehicle headlight is determined among the obtained candidate brightness and darkness border line set. The method for detecting the brightness and darkness border line of the vehicle headlight can be applicable to complicated outdoor scenes and is high in detection sensitivity, good in accuracy and strong in robustness.

Description

The bright dark border line detecting method of a kind of vehicle head lamp, device and system

Technical field

The present invention relates to technical field of image processing, particularly relate to the bright dark border line detecting method of a kind of vehicle head lamp based on image, device and system.

Background technology

The effect of vehicle head lamp is in night or other blind situation, for the vehicle driver provides illumination, and prevents that headlamp from causing and dazzling the eyes driver and passerby.The headlight beam irradiating angle is too high or too low all can cause adverse effect to driving, brings potential safety hazard.General vehicle head lamp has all been installed the two-filament bulb with instant and near light filament, and before car, existing bright and illumination uniformly can prevent dazzlingly when guaranteeing vehicle going at night, and distance-light luminous intensity distribution performance and Lighting direction are all had to certain requirement.In daily driving process, owing to being subject to effect of vibration, make headlamp depart from original installation site, thereby change Lighting direction.The vehicle head lamp Lighting direction is improper, has become one of main hidden danger that affects the Vehicular night safety traffic.Therefore, guarantee correct headlamp direction of illumination, significant to traffic safety.

In prior art, there is the method for utilizing screen or head light checking equipment to detect the headlight beam direction of illumination.When carrying out the detection of headlamp direction of illumination, usually can detect according to the light beam irradiates characteristics of distance light and dipped beam, wherein, the graphic feature that the distance light luminous intensity distribution forms is symmetric form, light shape center is the brightest; The graphic feature that the dipped beam luminous intensity distribution forms is asymmetric, has obvious bright dark side boundary line.Therefore, when detecting the lower beam headlamp direction of illumination, the position in bright dark side boundary line is important detected parameters.

The inventor finds in realizing process of the present invention, the detection method that prior art exists often need to be under the specific environment such as measuring station and is carried out during stationary vehicle, for example need to be by being projected in the position in the bright dark side of the image detection headlamp boundary line on specific screens when stationary vehicle.This method can't be applicable to exist the outdoor environment of strong interference, can't in outdoor environment, detect exactly the position in the bright dark side of headlamp boundary line.

Summary of the invention

The embodiment of the present invention provides the bright dark border line detecting method of a kind of vehicle head lamp based on image, device and system, goes for existing the bright dark side of the vehicle head lamp boundary line of the outdoor environment of strong interference to detect, and accuracy is high, strong robustness.

Technical scheme is as follows:

According to the first aspect of the embodiment of the present invention, the bright dark border line detecting method of a kind of vehicle head lamp based on image is disclosed, described method comprises:

Obtain the irradiation image of vehicle head lamp;

Described irradiation image is carried out to rim detection, obtain edge image;

In described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition; Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of candidate boundary line in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form;

According to the vehicle head lamp beam characteristics, in the bright dark side of the candidate boundary line set obtained, determine the bright dark side boundary line of vehicle head lamp.

Further, described the second line segment aggregate that detect to obtain respectively the first line segment aggregate that meets first direction and meet second direction comprises:

Detection obtains the first line segment aggregate that meets first direction and is specially: for each marginal point in edge image, obtain the gradient direction angle of marginal point; Whether the difference that judges the angle that the gradient direction angle of marginal point is corresponding with first direction is less than the first predetermined threshold value, obtains the first judged result; When described the first judged result shows that the difference of the angle that the gradient direction angle of marginal point is corresponding with first direction is less than the first predetermined threshold value, preserve described marginal point to the first edge point set and close; The marginal point obtained is carried out to clustering processing, obtain corresponding line segment;

The second line segment aggregate that detection meets second direction is specially: for each marginal point in edge image, obtain the gradient direction angle of marginal point; Whether the difference that judges the angle that the gradient direction angle of marginal point is corresponding with second direction is less than the second predetermined threshold value, obtains the second judged result; When described the second judged result shows that the difference of the angle that the gradient direction angle of marginal point is corresponding with second direction is less than the first predetermined threshold value, preserve described marginal point to the second edge point set and close; The marginal point obtained is carried out to clustering processing, obtain corresponding line segment.

Further, described the marginal point obtained is carried out to clustering processing, obtains corresponding line segment and comprise:

Obtain the corresponding polar coordinates straight-line equation of each marginal point parameter ρ in described marginal point set, by size parameter ρ is sorted, statistics obtains the number of the marginal point with identical parameters ρ, obtains and has the corresponding parameter ρ of local maximum marginal point number and marginal point set corresponding to this parameter ρ;

Marginal point in marginal point set corresponding to each parameter ρ is sorted;

Marginal point to after sequence, calculate the distance between adjacent two edges point, determines according to distance whether adjacent two edges point belongs to same line segment.

Further, based on described the first line segment aggregate and described the second line segment aggregate, obtain and to meet the pre-conditioned bright dark side of a candidate boundary line and comprise:

From described the first line segment aggregate, choosing a line segment as the first line segment, from described the second line segment aggregate, choosing a line segment as the second line segment;

Judge whether the horizontal ordinate of the terminal of described the first line segment is less than the horizontal ordinate of the terminal of the second line segment, obtain the 4th judged result;

Judge whether the ordinate of the terminal of described the first line segment is greater than the ordinate of the terminal of the second line segment, obtain the 5th judged result;

Whether the number that judges the marginal point of the number of marginal point of the first line segment and described the second line segment all is greater than the 4th predetermined threshold value, obtains the 6th judged result;

When described the 4th judged result, described the 5th judged result and described the 6th judged result are while being, using the combination of described the first line segment, described the second line segment as the bright dark side of a candidate boundary line.

Further, described according to the vehicle head lamp beam characteristics, in the bright dark side of the candidate boundary line set obtained, determine that bright dark side boundary line comprises:

Obtain the response in the bright dark side of each candidate boundary line, described response is sorted, the bright dark side boundary line using the bright dark side of the candidate of response maximum boundary line as vehicle head lamp.

Wherein, the response in the bright dark side of each candidate of described acquisition boundary line comprises:

The mean value of the gradient-norm value of the marginal point of the line segment of the mean value of the gradient-norm value of the marginal point of the line segment of first direction and second direction in the acquisition bright dark side of candidate boundary line;

Obtain the mean value of gradient-norm value of marginal point of line segment of described first direction and described second direction line segment marginal point the gradient-norm value mean value with the response of value as the bright dark side of this candidate boundary line.

Further, before the response in each bright dark side boundary line, described method also comprises in obtaining the combination of the bright dark side of candidate boundary line:

Utilize texture region feature and/or regional luminance feature, further screen the bright dark side of candidate boundary line.

Further, described method also comprises:

Export the positional information of turning point corresponding to bright dark side boundary line and/or form the level bright dark side boundary line in this bright dark side boundary line and the positional information in oblique bright dark side boundary line.

According to the second aspect of the embodiment of the present invention, the bright dark border line detector of a kind of vehicle head lamp based on image is disclosed, described device comprises:

Image obtains unit, be used to obtaining the irradiation image of vehicle head lamp;

Edge detection unit, for described irradiation image is carried out to rim detection, obtain edge image;

Candidate combinations is chosen unit, be used at described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition; Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of candidate boundary line in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form;

Determining unit, for according to the vehicle head lamp beam characteristics, determine the bright dark side boundary line of vehicle head lamp in the bright dark side of the candidate boundary line set obtained.

Further, described candidate combinations is chosen unit and is comprised that line segment detecting unit and combination choose unit, and described line segment detecting unit comprises the first line segment detecting unit and the second line segment detecting unit, and wherein, described the first line segment detecting unit comprises:

The first gradient direction angle acquiring unit, for each marginal point for edge image, obtain the gradient direction angle of marginal point;

Whether the first judging unit, be less than the first predetermined threshold value be used to the difference of the gradient direction angle that the judges marginal point angle corresponding with first direction, obtains the first judged result;

The first storage unit, while for the difference of the gradient direction angle that shows marginal point when described the first judged result angle corresponding with first direction, being less than the first predetermined threshold value, preserving described marginal point to the first edge point set and close;

The clustering processing unit, carry out clustering processing for the marginal point to obtaining, and obtains corresponding line segment;

Wherein, described the second line segment detecting unit comprises:

The second gradient direction angle acquiring unit, for each marginal point for edge image, obtain the gradient direction angle of marginal point;

Whether the second judging unit, be less than the second predetermined threshold value be used to the difference of the gradient direction angle that the judges marginal point angle corresponding with second direction, obtains the second judged result;

The second storage unit, while for the difference of the gradient direction angle that shows marginal point when described the second judged result angle corresponding with second direction, being less than the first predetermined threshold value, preserving described marginal point to the second edge point set and close;

The clustering processing unit, carry out clustering processing for the marginal point to obtaining, and obtains corresponding line segment.

Further, described clustering processing unit comprises:

Marginal point set acquiring unit, for obtaining the corresponding polar coordinates straight-line equation of each marginal point parameter ρ of described marginal point set, by size parameter ρ is sorted, statistics obtains the number of the marginal point with identical parameters ρ, obtains and has the corresponding parameter ρ of local maximum marginal point number and marginal point set corresponding to this parameter ρ;

Sequencing unit, sort for the marginal point to marginal point set corresponding to each parameter ρ;

The line segment processing unit, the marginal point for to after sequence, calculate the distance between adjacent two edges point, determines according to distance whether adjacent two edges point belongs to same line segment.

Further, described determining unit specifically for:

Obtain the response in the bright dark side of each candidate boundary line, described response is sorted, the bright dark side boundary line using the bright dark side of the candidate of response maximum boundary line as vehicle head lamp.

According to the third aspect of the embodiment of the present invention, the bright dark side of a kind of vehicle head lamp based on image boundary line detection system is disclosed, described system comprises in-vehicle camera and the bright dark border line detector of vehicle head lamp, wherein:

Described in-vehicle camera is used for irradiating the irradiation image of vehicle head lamp, and to the bright dark border line detector of described vehicle head lamp, sends the irradiation image of described vehicle head lamp;

The bright dark border line detector of described vehicle head lamp is for the irradiation image of the vehicle head lamp that receives described in-vehicle camera and send; Described irradiation image is carried out to rim detection, obtain edge image; In described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition; Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of described candidate boundary line in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form; According to the vehicle head lamp beam characteristics, in the bright dark side of the candidate boundary line set obtained, determine the bright dark side boundary line of vehicle head lamp.

The beneficial effect that the embodiment of the present invention can reach is: in method provided by the invention, obtain the irradiation image of vehicle head lamp, and described irradiation image is carried out to rim detection, obtain edge image; In described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition; Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of described candidate boundary line in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form; According to the vehicle head lamp beam characteristics, in the bright dark side of the candidate boundary line set obtained, determine bright dark side boundary line.In method provided by the invention, due to level bright dark side boundary line and oblique bright dark side boundary line are detected respectively and combine, obtained all possible bright dark side boundary line, and utilize the headlight beam feature from the boundary line set of the bright dark side of candidate, filtering out the most qualified bright dark side boundary line as final testing result, effectively got rid of the impact in the interference edge boundary line caused due to jamming light source, interference texture in the outdoor environment, go for complicated Outdoor Scene, detection sensitivity is high, accuracy good, strong robustness.

The accompanying drawing explanation

In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, below will the accompanying drawing of required use in embodiment or description of the Prior Art be briefly described, apparently, the accompanying drawing the following describes is only some embodiment that put down in writing in the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain according to these accompanying drawings other accompanying drawing.

Fig. 1 is bright dark border line detecting method the first embodiment process flow diagram of the vehicle head lamp based on image provided by the invention;

Fig. 2 is the filter template schematic diagram;

Fig. 3 is the bright dark side of headlamp boundary line schematic diagram;

Fig. 4 is that line segment is cut apart schematic diagram;

Fig. 5 is bright dark border line detecting method the second embodiment process flow diagram of the vehicle head lamp based on image provided by the invention;

Fig. 6 is the bright dark border line detector of the vehicle head lamp based on image provided by the invention schematic diagram;

Fig. 7 is the bright dark side of the vehicle head lamp based on image provided by the invention boundary line detection system schematic diagram.

Embodiment

The embodiment of the present invention provides the bright dark border line detecting method of a kind of vehicle head lamp based on image, device and system, goes for existing the bright dark side of the vehicle head lamp boundary line of the outdoor environment of strong interference to detect, and accuracy is high, strong robustness.

In order to make those skilled in the art person understand better the technical scheme in the present invention, below in conjunction with the accompanying drawing in the embodiment of the present invention, technical scheme in the embodiment of the present invention is clearly and completely described, obviously, described embodiment is only the present invention's part embodiment, rather than whole embodiment.Based on the embodiment in the present invention, those of ordinary skills, not making under the creative work prerequisite the every other embodiment obtained, should belong to the scope of protection of the invention.

At first thought provided by the invention is set forth.

When detecting the headlight beam direction of illumination, while particularly detecting the dim head lights direction of illumination, usually need to detect the position in bright dark side boundary line.This be because, when the light beam irradiates of headlamp is on road surface or metope, can produce the hot spot of certain form, this hot spot can imaging in in-vehicle camera, forms vehicle front lighting light irradiation image.The hot spot of headlamp is usually expressed as bright dark side boundary line feature in irradiating image, the coordinate of the position in therefore definite bright dark side boundary line and the turning point in bright dark side boundary line is the important measurement parameter that carries out the vehicle head lamp detection.

In the method that prior art provides, often need to be under the specific environments such as measuring station and carry out the detection of headlight beam direction of illumination during stationary vehicle, with by being projected in the position in the bright dark side of the image detection headlamp boundary line on specific screens.This method can't be applicable to complicated outdoor environment.In outdoor environment, usually there are other edges in metope or ground etc., such as crack, road surface or metope japanning etc., these edges cause interference to the bright dark side of headlamp boundary line, make bright dark side boundary line irradiate in image discontinuous, bright dark side boundary line turning point at headlamp not obvious or have except headlamp other edge line bright dark side boundary line.Simultaneously, surround lighting is crossed bright or excessively secretly also can be made bright dark side boundary line owe clear, thereby causes interruption, has affected the detection in bright dark side boundary line.Based on this, method provided by the invention, form according to the bright dark side of the headlamp boundary line in vehicle front lighting light irradiation image, first segmentation detects to bright dark side boundary line, combine again, obtain the bright dark side of all candidates' headlamp boundary line in image, and utilize the headlight beam feature to wait set and filter out the most qualified bright dark side boundary line as final testing result from the bright dark side of candidate boundary line, effectively got rid of in outdoor environment due to jamming light source, the impact in the interference edge boundary line that the interference texture causes, go for complicated Outdoor Scene, detection sensitivity is high, accuracy is good, strong robustness.

Referring to Fig. 1, it is bright dark border line detecting method the first embodiment process flow diagram of the vehicle head lamp based on image provided by the invention.

S101, the irradiation image of acquisition vehicle head lamp.

During specific implementation, at vehicle front, in-vehicle camera is set, described in-vehicle camera is used for irradiating the irradiation image of vehicle head lamp, and to the bright dark border line detector of vehicle head lamp, sends the irradiation image of described vehicle head lamp.Wherein, the shooting direction of in-vehicle camera is identical with the direction of illumination of headlamp.The mode that obtains the vehicle head lamp direction of illumination comprises: set in advance the radiation modality of in-vehicle camera, take the irradiation image of vehicle head lamp on road surface.One or more that the radiation modality of described camera can comprise on schedule, the mode of predetermined mileage, predetermined period or pre-determined number is irradiated.

It will be appreciated by persons skilled in the art that method provided by the invention can be applied to outdoor environment, and no matter vehicle is in transport condition or halted state all can obtain the irradiation image of headlamp on road.

S102, carry out rim detection to described irradiation image, obtains edge image.

During specific implementation, the irradiation image obtained is carried out to the method that Image Edge-Detection can adopt the operator extraction edges, edge such as sobel, canny, the present invention does not limit this.

In a kind of specific implementation, the present invention specifically carries out Image Edge-Detection in the following ways.It will be appreciated by persons skilled in the art that followingly only for exemplary illustration, not to be considered as limitation of the present invention, those skilled in the art all belong to protection scope of the present invention not paying other implementations of obtaining under creative work.

Particularly, step S102 can realize by following steps:

S102A, carry out the filtering processing to irradiating image, obtains the mould value image of horizontal gradient image, VG (vertical gradient) image and gradient.

During specific implementation, can utilize filter template as shown in Figure 2 to carry out the filtering processing, obtain the mould value image of horizontal gradient image, VG (vertical gradient) image and gradient.

Wherein, described horizontal gradient image I 1 obtains in the following manner:

I ₁(u,v)＝I(u,v)*w ^T(u,v) （1）

Wherein, I ₁(u, v) means the horizontal gradient image, I (u, v) presentation video point coordinate, and * means convolution operation, and u is the picture point horizontal coordinate, and v is the picture point ordinate, and w is the matrix of Fig. 2 template representation, w ^TIt is the transpose of a matrix of Fig. 2 template representation.

Wherein, described VG (vertical gradient) image I ₂Obtain in the following manner:

I ₂(u,v)＝I(u,v)*w(u,v) （2）

Wherein, I ₂(u, v) means the VG (vertical gradient) image, I (u, v) presentation video point coordinate, and * means convolution operation, and u is the picture point horizontal coordinate, and v is the picture point ordinate, and w is the matrix of Fig. 2 template representation.

Wherein, the mould value image M o of gradient obtains in the following manner:

$\mathrm{Mo}(u,v)=\sqrt{I_1{(u,v)}^2+I_2{(u,v)}^2}---\left(3\right)$

Wherein, Mo (u, v) means the mould value image of gradient, I ₁(u, v) means horizontal gradient image, I ₂(u, v) means the VG (vertical gradient) image.

S102B, the mould value according to irradiating each pixel of image, judge whether described pixel is marginal point.

During specific implementation, the mould value that judges each pixel determines that whether described pixel is that a kind of possible implementation of marginal point is: whether the mould value that judges pixel is greater than predetermined threshold value; When the mould value of judgement pixel is greater than predetermined threshold value, determine that described pixel is marginal point; When the mould value of judgement pixel is not more than predetermined threshold value, determine that described pixel is non-marginal point.For example, can set in advance threshold value is 10, when the mould value of pixel is greater than 10, determines that described pixel is marginal point, otherwise is non-marginal point.

S102C, form edge image by the point that is defined as marginal point.

By step S102B, detect all marginal points, the figure consisted of all marginal points namely forms edge detection results figure, i.e. edge image.

S103, in described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition.

When carrying out the detection of dim head lights direction of illumination, need to detect the position in bright dark side boundary line.As shown in Figure 3, in vehicle head lamp irradiates image, the bright dark side of the headlamp boundary line schematic diagram formed by the bright dark side of level boundary line 1, oblique bright dark side boundary line 2, bright dark side boundary line turning point 3.

In this step, namely to obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction by detecting respectively, i.e. detection level bright dark side boundary line and oblique bright dark side boundary line combination respectively, to obtain all possible bright dark side boundary line.Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of described candidate boundary line in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form.The set in the corresponding bright dark side of level of the set of the first line segment boundary line, the set in the corresponding oblique bright dark side of the set of the second line segment boundary line.The boundary line set of the bright dark side of candidate is the set that all bright dark side of candidate boundary lines that consist of the bright dark side of a level boundary line (corresponding the first line segment) and an oblique bright dark side boundary line (corresponding the second line segment) form.

During specific implementation, step S103 specifically can realize by following steps:

S103A, detect and obtain the first line segment aggregate that meets first direction.

Wherein, step S103A can realize by following steps again:

For each marginal point in edge image, obtain the gradient direction angle of marginal point.Whether the difference that judges the angle that the gradient direction angle of marginal point is corresponding with first direction is less than the first predetermined threshold value, obtains the first judged result; When described the first judged result shows that the difference of the angle that the gradient direction angle of marginal point is corresponding with first direction is less than the first predetermined threshold value, preserve described marginal point to the first edge point set and close; The marginal point obtained is carried out to clustering processing, obtain corresponding line segment.

Wherein, the marginal point obtained is carried out to clustering processing, obtains corresponding line segment and comprise:

Obtain the corresponding polar coordinates straight-line equation of each marginal point parameter ρ in described marginal point set, by size parameter ρ is sorted, statistics obtains the number of the marginal point with identical parameters ρ, obtains and has the corresponding one or more parameter ρ of local maximum marginal point number and marginal point set corresponding to parameter ρ;

Marginal point in marginal point set corresponding to each parameter ρ is sorted;

Marginal point to after sequence, calculate the distance between adjacent two edges point, determines according to distance whether adjacent two edges point belongs to same line segment.

The edge point carries out clustering processing for the first time and namely passes through to obtain the corresponding polar coordinates straight-line equation of each marginal point parameter ρ in described marginal point set, by size parameter ρ is sorted, statistics obtains the number of the marginal point with identical parameters ρ, obtain the beneficial effect with the corresponding parameter ρ of local maximum marginal point number and marginal point set corresponding to this parameter ρ and be to get rid of in a large number the Clutter edge point in background, effectively improve the accuracy detected.

Specific implementation is: for each marginal point (u, v), if its gradient direction | θ-φ | and≤θ _ thre, through type (4) calculates the parameter ρ of its corresponding polar coordinates straight-line equation:

ρ＝ucos(θ)+vsin(θ) （4）

Wherein, φ is the orientation angle of appointment, can be set to 90 degree for the angle of first direction, and θ _ thre is the gradient direction error threshold, and θ means straight line gradient direction angle.

$\mathrm{\θ}=\arg \tan \frac{I_2(u,v)}{I_1(u,v)}---\left(5\right)$

To parameter ρ sequence, statistics obtains the number of the marginal point with identical parameters ρ by size, obtains and has the corresponding parameter ρ of local maximum marginal point number and marginal point set corresponding to this parameter ρ; After marginal point set corresponding to the parameter ρ that obtains maximum point, the marginal point set of each acquisition is carried out to clustering processing, obtain some the first line segments.Concrete, marginal point in the marginal point S set detected is sorted: the name placement of pressing in S set is obtained to ordered set S ', sequence needs to meet: for any point, if it has a left side (right side) adjoint point, it,, to the Euclidean distance of left (right side) adjoint point, is not more than its Euclidean distance to left other point of (right side) side.The little point of horizontal ordinate is in left side, if horizontal ordinate is identical, the point that ordinate is large is in left side.

After completing sequence, the marginal point to after sequence, calculate the distance between adjacent two edges point, determines according to distance whether adjacent two edges point belongs to same line segment.Concrete, obtain the distance between adjacent two edges point, judge whether the distance between adjacent two edges point is greater than the 3rd predetermined threshold value, obtain the 3rd judged result; When showing distance between adjacent two edges point, described the 3rd judged result while being greater than the 3rd predetermined threshold value, carries out the line segment dividing processing; When showing distance between adjacent two edges point, described the 3rd judged result while being less than the 3rd predetermined threshold value, determines that adjacent two edges point belongs to same line segment.The position that is greater than the 3rd threshold value D_thre by the distance of searching neighboring edge point is carried out benefit that line segment cuts apart and is to solve boundary line and is subject to the line segment discontinuous problem that environmental interference causes.Cut apart schematic diagram as shown in Figure 4.As shown in Figure 4, Fig. 4 has 21 marginal points, cuts apart and is divided into two line segments through line segment.6, left side marginal point has formed line segment 11, and 15, the right marginal point has formed line segment 22.Point A is line segment 11 starting points, and some F is line segment 11 terminals.Point G is line segment 22 starting points, and some U is line segment 22 terminals.Through calculating, the distance B between some F and some G is greater than the 3rd threshold value D_thre, therefore carries out from here line segment and cuts apart, and thinks that a F and some G do not belong to same line segment.

S103B, detect the second line segment with second direction.

During specific implementation, step S103B can realize by following steps again:

For each marginal point in edge image, obtain the gradient direction angle of marginal point.Whether the difference that judges the angle that the gradient direction angle of marginal point is corresponding with second direction is less than the second predetermined threshold value, obtains the second judged result; When described the second judged result shows that the difference of the angle that the gradient direction angle of marginal point is corresponding with second direction is less than the first predetermined threshold value, preserve described marginal point to the second edge point set and close; The marginal point obtained is carried out to clustering processing, obtain corresponding line segment.

Concrete, the marginal point obtained is carried out to clustering processing, obtain corresponding line segment and comprise:

Obtain the corresponding polar coordinates straight-line equation of each marginal point parameter ρ in described marginal point set, by size parameter ρ is sorted, statistics obtains the number of the marginal point with identical parameters ρ, obtains and has the corresponding parameter ρ of local maximum marginal point number and marginal point set corresponding to this parameter ρ;

Marginal point in marginal point set corresponding to each parameter ρ is sorted;

Marginal point to after sequence, calculate the distance between adjacent two edges point, determines according to distance whether adjacent two edges point belongs to same line segment.

The specific implementation process can be carried out with reference to step S103A, and wherein, the angle of the second direction of appointment can be set to 75 degree.

S103C, meet the bright dark side of pre-conditioned candidate boundary line based on described the first line segment aggregate and described the second line segment aggregate acquisition.

The first line segment detected in step S102 (the bright dark border line segment of level) and the second line segment (oblique bright dark border line segment) are combined, obtain complete bright dark side boundary line.Each combination consists of a bright dark border line segment of level and an oblique bright dark border line segment, can set combination condition according to actual conditions.For example require the end point (u of the bright dark border line segment of level in image ₁, v ₁) appear at the starting point (u of oblique line segment ₂, v ₂) lower left, and the some number on level bright dark side boundary line and oblique bright dark side boundary line is in the reasonable scope.During specific implementation, based on described the first line segment aggregate and described the second line segment aggregate, obtain and to meet the pre-conditioned bright dark side of a candidate boundary line and comprise: from described the first line segment aggregate, choosing a line segment as the first line segment, from described the second line segment aggregate, choosing a line segment as the second line segment; Judge whether the horizontal ordinate of the terminal of described the first line segment is less than the horizontal ordinate of the terminal of the second line segment, obtain the 4th judged result; Judge whether the ordinate of the terminal of described the first line segment is greater than the ordinate of the terminal of the second line segment, obtain the 5th judged result; Whether the number that judges the marginal point of the number of the marginal point on the first line segment and described the second line segment all is greater than the 4th predetermined threshold value, obtains the 6th judged result; When described the 4th judged result, described the 5th judged result and described the 6th judged result are while being, using the combination of described the first line segment, described the second line segment as the bright dark side of a candidate boundary line.Wherein, the 4th predetermined threshold value can preset.Repeat above-mentioned steps, namely can obtain the bright dark side of a plurality of candidates boundary line.

S104, according to the vehicle head lamp beam characteristics, determine bright dark side boundary line in the bright dark side of the candidate boundary line set obtained.

The purpose of step S104 is from the bright dark side of the some candidates boundary line obtained, selecting to meet most the bright dark side boundary line of headlamp feature.During specific implementation, step S104 can have multiple implementation, and step S104 specifically can comprise:

Obtain the response of the bright dark side of each candidate boundary line set, described response is sorted, the bright dark side boundary line using the bright dark side of the candidate of response maximum boundary line as vehicle head lamp.

Wherein, the response in the bright dark side of each candidate of described acquisition boundary line comprises:

The mean value of the gradient-norm value of the marginal point of the line segment of the mean value of the gradient-norm value of the marginal point of the line segment of first direction and second direction in the acquisition bright dark side of candidate boundary line;

Obtain the mean value of gradient-norm value of marginal point of line segment of described first direction and described second direction line segment marginal point the gradient-norm value mean value with the response of value as the bright dark side of this candidate boundary line.

In another implementation of the present invention, before the response in each bright dark side boundary line, described method also comprises in obtaining the combination of the bright dark side of candidate boundary line:

Utilize texture region feature and/or regional luminance feature, further screen the bright dark side of candidate boundary line.

That is to say, described method also comprises utilizes the texture region feature, the screening bright dark side of candidate boundary line in the bright dark side of the candidate boundary line set obtained; And/or

Utilize the regional luminance feature, the screening bright dark side of candidate boundary line in the bright dark side of the candidate boundary line set obtained; And/or

Below specific implementation is introduced.It will be appreciated by persons skilled in the art that and can only use the response feature response of all bright dark sides boundary line candidate combinations to be sorted to determine the final position in bright dark side boundary line.Also can first use textural characteristics, determine the position candidate in bright dark side boundary line, and then utilize the response feature to determine the final position in bright dark side boundary line.Also can first use the regional luminance feature to determine the position candidate in bright dark side boundary line, and then utilize the response feature to determine the final position in bright dark side boundary line.Certainly, it will be appreciated by persons skilled in the art that and also can utilize simultaneously textural characteristics and regional luminance feature to determine the position candidate in bright dark side boundary line, and then utilize the response feature to determine the final position in bright dark side boundary line.In addition, below be only exemplary illustration, those skilled in the art all belong to protection scope of the present invention not paying other implementations of obtaining under creative work.

Wherein, utilize the texture region feature, the screening bright dark side of candidate boundary line can comprise in the bright dark side of the candidate boundary line set obtained: in the bright dark side of the candidate boundary line set obtained, obtain texture region corresponding to bright dark side boundary line; Whether the entropy that judges texture region is less than the 5th predetermined threshold value; When the entropy of described texture region is less than the 5th predetermined threshold value, determine that described texture region is possible headlamp zone, the bright dark side boundary line that described texture region is corresponding is as the bright dark side of candidate boundary line.During specific implementation, the texture in headlamp zone should meet certain condition, and in the image based on road surface, the upper beam spot region belongs to weak texture region, if near bright dark side boundary line, texture strength is too large, may be barrier region, should get rid of.In practical operation: texture region scope corresponding to bright dark side boundary line can be determined by the boundary rectangle in bright dark side boundary line.The power of horizontal amount texture can be by regional entropy judgement.For example, whether the entropy that judges texture region is less than the 5th predetermined threshold value, when the entropy of described texture region is less than the 5th predetermined threshold value, determine that described texture region is the potential zone of headlamp, the bright dark side boundary line that described texture region is corresponding is as the bright dark side of candidate boundary line.

Wherein, utilize the regional luminance feature, screening follow-up bright dark side boundary line can comprise in the bright dark side of the candidate boundary line set obtained: the brightness value of the lower zone of the brightness value of the upper area of the first line segment and the first line segment in the boundary line set of the bright dark side of acquisition candidate; The brightness value of the right side area of the brightness value of the left field of the second line segment and the second line segment in the boundary line set of the bright dark side of acquisition candidate; Whether the brightness value that judges the lower zone of the first line segment is greater than the 6th predetermined threshold value, obtains the 7th judged result; Whether the brightness value that judges the right side area of the second line segment is greater than the 7th predetermined threshold value, obtains the 8th judged result; Judge that whether the difference between the brightness value of upper area of the brightness value of lower zone of the first line segment and the first line segment is greater than the 8th predetermined threshold value, obtains the 9th judged result; Judge that whether the difference between the brightness value in left zone of the brightness value of right side area of the second line segment and the second line segment is greater than the 9th predetermined threshold value, obtains the tenth judged result; When described the 7th judged result, described the 8th judged result, described the 9th judged result, described the tenth judged result are and are, using the combination of described the first line segment and the second line segment as the bright dark side of candidate boundary line.During specific implementation, the brightness of image contrast in upper and lower region territory, the bright dark side of headlamp level boundary line and left and right region territory, oblique bright dark side boundary line should be in the reasonable scope.Suppose that the first line segment is the bright dark side of level boundary line lower zone E ₁Interior average luminance of pixels value is I (E ₁), the first line segment is the bright dark side of level boundary line upper area E ₂Interior average luminance of pixels value is I (E ₂), the second line segment is oblique bright dark side boundary line right side area E ₃Interior average luminance of pixels value is I (E ₃), the second line segment is oblique bright dark side boundary line left field E ₄Interior average luminance of pixels value is I (E ₄).Zone E ₁, E ₂, E ₃, E ₄Shape and size can set according to actual conditions.The bright dark side of headlamp boundary line should meet following condition:

$\left\{\begin{array}{c}I\left(E_1\right)>{\mathrm{<figure-callout\; id="1"\; label="IE"\; filenames="HDA00003599170000012.png"\; state="\{\{state\}\}">IE</figure-callout>}}_1\_\mathrm{thre}\\ I\left(E_3\right)>{\mathrm{IE}}_3\_\mathrm{thre}\\ I\left(E_1\right)-I\left(E_2\right)>{\mathrm{IE}}_{12}\_\mathrm{thre}\\ I\left(E_3\right)-I\left(E_4\right)>{\mathrm{IE}}_{34}\_\mathrm{thre}\end{array}---\left(6\right)\right.$

Wherein, IE ₁_ thre is the 6th predetermined threshold value, IE ₃_ thre is the 7th predetermined threshold value, IE ₁₂_ thre is the 8th predetermined threshold value, IE ₃₄_ thre is the 9th predetermined threshold value, all can set according to actual conditions.

Wherein, described according to the vehicle head lamp beam characteristics, in the bright dark side of the candidate boundary line candidate collection obtained, determine that bright dark side boundary line comprises: the response that obtains the bright dark side of each candidate boundary line, described response is sorted, the bright dark side boundary line using the bright dark side of the candidate of response maximum boundary line as vehicle head lamp.During specific implementation: the mean value of the gradient-norm value of the marginal point of the second line segment of the mean value of the gradient-norm value of the marginal point of the first line segment of first direction and second direction in the boundary line set of the bright dark side of acquisition candidate; Obtain the mean value of gradient-norm value of marginal point of described the first line segment and described the second line segment marginal point the gradient-norm value mean value with the response of value as the bright dark side of this candidate boundary line; The position that the combination of the bright dark side boundary line of response maximum is corresponding is as the final position in bright dark side boundary line.

Wherein, the gradient-norm value of the marginal point of the first line segment obtains in the following manner:

$I\left(M_H\right)=\sqrt{I_1{(u,v)}^2+I_2{(u,v)}^2}---\left(7\right)$

Wherein, I (M _H) mean the mould value of the gradient of marginal point, I ₁(u, v) means the horizontal gradient image of the marginal point of the first line segment, I ₂(u, v) means the VG (vertical gradient) image of the marginal point of the first line segment.

Obtaining the first line segment is the mould value of all marginal points on the bright dark side of level boundary line, and calculating mean value has then namely obtained the mean value of gradient-norm value of the marginal point of the first line segment.

Wherein, the Grad mould value of the marginal point of the second line segment obtains in the following manner:

$I\left(M_S\right)=\sqrt{I_1{(u,v)}^2+I_2{(u,v)}^2}---\left(8\right)$

Wherein, I (M _S) mean the mould value of the gradient of marginal point, I ₁(u, v) means the horizontal gradient image of the marginal point of the second line segment, I ₂(u, v) means the VG (vertical gradient) image of the marginal point of the second line segment.

Wherein, response I (M) obtains in the following manner:

I(M)=I(M _H)+I(M _S) （9）

Response is that point is met to the value that bright dark side boundary line feature degree is described.To illustrate that more this lights dark variation stronger when the mould value, and this just meets the characteristics of image in bright dark side boundary line, a kind of form that therefore can be using the mould value as response.Other can describe the description that certain point belongs to the possibility of putting on bright dark side boundary line, can be as response.

In this embodiment of the present invention, the position that the combination of the bright dark side boundary line of response maximum is corresponding is as the final position in bright dark side boundary line.

It should be noted that, if first utilized textural characteristics or regional luminance feature further to screen the bright dark side of candidate boundary line, the bright dark side boundary line of calculated response value is the bright dark side of the described candidate boundary line filtered out.

Described method also comprises: the positional information of exporting bright dark side boundary line.

Wherein, the positional information in bright dark side boundary line comprises the positional information of the turning point that bright dark side boundary line is corresponding and/or forms the level bright dark side boundary line in this bright dark side boundary line and the positional information in oblique bright dark side boundary line.The positional information of the turning point that described bright dark side boundary line is corresponding is determined by the positional information of described the first line segment and the positional information of described the second line segment.Concrete, the positional information of the turning point that bright dark side boundary line is corresponding is specially the intersecting point coordinate in level bright dark side boundary line and oblique bright dark side boundary line.

Further, method provided by the invention can comprise:

Utilize the positional information of turning point corresponding to described bright dark side boundary line and/or form the level bright dark side boundary line in this bright dark side boundary line and the irradiating angle of the positional information calculation headlamp in oblique bright dark side boundary line.

In first embodiment of the invention, in method provided by the invention, owing to the bright dark side boundary line of the level to meeting first direction and oblique bright dark side boundary line detecting respectively, obtained the candidate combinations in all possible bright dark side boundary line, and utilize the headlight beam feature from the boundary line set of the bright dark side of candidate, filtering out the most qualified bright dark side boundary line as final testing result, effectively got rid of in outdoor environment due to jamming light source, the impact in the interference edge boundary line that the interference texture causes, go for complicated Outdoor Scene, detection sensitivity is high, accuracy is good, strong robustness.

Especially, for in complex scene, there is jamming light source, disturb texture etc. to cause the boundary line in the bright dark side of other similar headlamps boundary line to exist the bright dark side of the headlamp boundary line caused to detect inaccurate problem, the present invention detects and obtains all possible bright dark side boundary line and combine by segmentation, and according to the characteristics of image of headlamp, therefrom screen the most qualified boundary line, as the bright dark side of headlamp boundary line, effectively overcome interference, detection sensitivity is high, accuracy good, strong robustness.

Especially; in complex scene; owing to disturbing texture etc. can cause the bright dark side of the discontinuous headlamp the caused boundary line in the bright dark side of headlamp boundary line to detect inaccurate problem; in the present invention; the testing result in each bright dark side boundary line; all that mode by combination obtains, separated situation in allowing; And, when having again bright dark side boundary line marginal point to obtain bright dark border line segment, be provided with threshold value D_thre, and whether the distance that judges consecutive point is greater than setting threshold and limits and cut apart.Therefore, for every section line segment, when the edge interruption is less than D_thre, still can be linked to be the same line segment, can not disconnect.Therefore, when boundary line is discontinuous, still can accurately detect strong robustness.

Referring to Fig. 5, be bright dark border line detecting method the second embodiment process flow diagram of the vehicle head lamp based on image provided by the invention.

In this specific implementation, when utilizing the headlight beam feature to determine the final position in bright dark side boundary line, it is that texture region feature and/or brightness screen that the present invention considers the headlight beam feature, and in the candidate combinations filtered out, the final position using the position of the combination correspondence of response maximum as bright dark side boundary line.

S501, the irradiation image of acquisition vehicle head lamp.

Specific implementation please refer to the implementation of step S101.

S502, carry out rim detection to described irradiation image, obtains edge image.

Specific implementation please refer to the implementation of step S102.

S503, in described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition.

Specific implementation please refer to the implementation of step S103.

S504, utilize the texture region feature screening bright dark side of candidate boundary line in the bright dark side of the candidate boundary line set obtained.

During specific implementation, utilize the texture region feature, the screening bright dark side of candidate boundary line comprises in the bright dark side of the candidate boundary line set obtained: in the bright dark side of the candidate boundary line set obtained, obtain texture region corresponding to bright dark side boundary line; Whether the entropy that judges texture region is less than the 5th predetermined threshold value; When the entropy of described texture region is less than the 5th predetermined threshold value, determine that described texture region is the headlamp candidate region, the bright dark side boundary line that described texture region is corresponding is as the bright dark side of candidate boundary line.

S505, utilize the regional luminance feature, the screening bright dark side of candidate boundary line in the bright dark side of the candidate boundary line set obtained.

During specific implementation, utilize the regional luminance feature, the screening bright dark side of candidate boundary line comprises in the bright dark side of the candidate boundary line set obtained: the brightness value of the lower zone of the brightness value of the upper area of the first line segment and the first line segment in the boundary line set of the bright dark side of acquisition candidate; The brightness value of the right side area of the brightness value of the left field of the second line segment and the second line segment in the boundary line set of the bright dark side of acquisition candidate; Whether the brightness value that judges the lower zone of the first line segment is greater than the 6th predetermined threshold value, obtains the 7th judged result; Whether the brightness value that judges the right side area of the second line segment is greater than the 7th predetermined threshold value, obtains the 8th judged result; Judge that whether the difference between the brightness value of upper area of the brightness value of lower zone of the first line segment and the first line segment is greater than the 8th predetermined threshold value, obtains the 9th judged result; Judge that whether the difference between the brightness value in left zone of the brightness value of right side area of the second line segment and the second line segment is greater than the 9th predetermined threshold value, obtains the tenth judged result; When described the 7th judged result, described the 8th judged result, described the 9th judged result, described the tenth judged result are and are, using the combination of described the first line segment and the second line segment as the bright dark side of candidate boundary line.

It will be appreciated by persons skilled in the art that step S504 and step S505 do not have inevitable priority execution sequence, can carry out concurrently, also can carry out reversedly.Certainly, step S504 and step S505 also can select one applicable.

S506, in the bright dark side of the candidate who filters out boundary line, utilize the response feature, determines bright dark side boundary line.

In all bright dark side of candidate boundary lines that meet above-mentioned condition, the bright dark side of headlamp boundary line is the bright dark side boundary line with maximum response.For example can add up the mean value I (M of the gradient-norm value of putting on the bright dark side of the level boundary line of residue combination _H), the mean value I (M of the gradient-norm value of putting on oblique bright dark side boundary line _S).Calculated response value I (M):

I(M)＝I(M _H)+I(M _S) （10）

Then, get the final position of the position of the combination correspondence with maximum I (M) value as the bright dark side of headlamp boundary line.

S507, the positional information in the described bright dark side boundary line that output is determined.

Fig. 6 is the bright dark border line detector of the vehicle head lamp based on image provided by the invention schematic diagram.

The bright dark border line detector of a kind of vehicle head lamp based on image, described device comprises:

Image obtains unit 601, be used to obtaining the irradiation image of vehicle head lamp.

Edge detection unit 602, for described irradiation image is carried out to rim detection, obtain edge image.

Candidate combinations is chosen unit 603, be used at described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition; Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of described candidate boundary line in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form.

Determining unit 604, for according to the headlight beam feature, determine the bright dark side boundary line of vehicle head lamp in the bright dark side of the candidate boundary line set obtained.

Further, described device also comprises:

Output unit, be used to exporting the positional information in bright dark side boundary line.

Further, described candidate combinations is chosen unit and is comprised that line segment detecting unit and combination choose unit, wherein, described line segment detecting unit the first line segment detecting unit and the second line segment detecting unit, wherein, the first line segment detecting unit comprises:

The first gradient direction angle acquiring unit, for each marginal point for edge image, obtain the gradient direction angle of marginal point;

Whether the first judging unit, be less than the first predetermined threshold value be used to the difference of the gradient direction angle that the judges marginal point angle corresponding with first direction, obtains the first judged result;

The first storage unit, while for the difference of the gradient direction angle that shows marginal point when described the first judged result angle corresponding with first direction, being less than the first predetermined threshold value, preserving described marginal point to the first edge point set and close;

The clustering processing unit, carry out clustering processing for the marginal point to obtaining, and obtains corresponding line segment.

Further, described candidate combinations is chosen unit and is comprised that line segment detecting unit and combination choose unit, wherein, described line segment detecting unit the first line segment detecting unit and the second line segment detecting unit, wherein, the second line segment detecting unit comprises:

The second gradient direction angle acquiring unit, for each marginal point for edge image, obtain the gradient direction angle of marginal point;

Whether the second judging unit, be less than the second predetermined threshold value be used to the difference of the gradient direction angle that the judges marginal point angle corresponding with second direction, obtains the second judged result;

The second storage unit, while for the difference of the gradient direction angle that shows marginal point when described the second judged result angle corresponding with second direction, being less than the first predetermined threshold value, preserving described marginal point to the second edge point set and close;

The clustering processing unit, carry out clustering processing for the marginal point to obtaining, and obtains corresponding line segment.

Further, described clustering processing unit comprises:

Marginal point set acquiring unit, for obtaining the corresponding polar coordinates straight-line equation of each marginal point parameter ρ of described marginal point set, by size parameter ρ is sorted, statistics obtains the number of the marginal point with identical parameters ρ, obtains and has the corresponding parameter ρ of local maximum marginal point number and marginal point set corresponding to this parameter ρ;

Sequencing unit, sort for the marginal point to marginal point set corresponding to each parameter ρ;

The line segment processing unit, the marginal point for to after sequence, calculate the distance between adjacent two edges point, determines according to distance whether adjacent two edges point belongs to same line segment.

Further, described candidate combinations is chosen unit and is comprised that line segment detecting unit and combination choose unit, and wherein, described combination is chosen unit and comprised:

Choose unit, for from described the first line segment aggregate, choosing a line segment as the first line segment, from described the second line segment aggregate, choosing a line segment as the second line segment;

Whether the 4th judging unit, be less than the horizontal ordinate of the terminal of the second line segment be used to the horizontal ordinate of the terminal that judges described the first line segment, obtain the 4th judged result;

Whether the 5th judging unit, be greater than the ordinate of the terminal of the second line segment be used to the ordinate of the terminal that judges described the first line segment, obtain the 5th judged result;

Whether the 6th judging unit, all be greater than the 4th predetermined threshold value be used to the number of the marginal point of the number that judges the marginal point on the first line segment and described the second line segment, obtains the 6th judged result;

Assembled unit, while for when described the 4th judged result, described the 5th judged result and described the 6th judged result, being being, using the combination of described the first line segment, described the second line segment as the bright dark side of candidate boundary line.

Further, described determining unit specifically for:

Obtain the response in each bright dark side boundary line, described response is sorted, the bright dark side boundary line using the bright dark side of the candidate of response maximum boundary line as vehicle head lamp.

Further, described determining unit is specifically for the mean value of the gradient-norm value of the marginal point of the second line segment of the mean value of the gradient-norm value of the marginal point of the first line segment of obtaining the bright dark side of candidate boundary line set first direction and second direction; Obtain the mean value of gradient-norm value of marginal point of the first line segment of described first direction and described second direction the second line segment marginal point the gradient-norm value mean value with the response of value as bright dark side boundary line.

Further, described device also comprises:

The first determining unit, be used to utilizing the texture region feature, determine the bright dark side of candidate boundary line in the bright dark side of the candidate boundary line set obtained; And/or

The second determining unit, be used to utilizing the regional luminance feature, determine the bright dark side of candidate boundary line in the bright dark side of the candidate boundary line set obtained.Described determining unit, be used to utilizing the response feature, determine bright dark side boundary line in the bright dark side of the candidate boundary line obtained.

Further, described the first determining unit comprises:

Texture region obtains unit, and the bright dark side of the candidate boundary line set for obtaining, obtain texture region corresponding to bright dark side boundary line;

Whether the 6th judging unit, be less than the 5th predetermined threshold value be used to the entropy that judges texture region;

The primary importance determining unit, while for the entropy when described texture region, being less than the 5th predetermined threshold value, determine that described texture region is the headlamp candidate region, and the bright dark side boundary line that described texture region is corresponding is as the bright dark side of candidate boundary line.

Further, described the second determining unit comprises:

Brightness value obtains unit, for the brightness value of the lower zone of the brightness value of the upper area that obtains the bright dark side of candidate boundary line set the first line segment and the first line segment; The brightness value of the right side area of the brightness value of the left field of the second line segment and the second line segment in the boundary line set of the bright dark side of acquisition candidate;

Whether the 7th judging unit, be greater than the 6th predetermined threshold value be used to the brightness value of the lower zone that judges the first line segment, obtains the 7th judged result;

Whether the 8th judging unit, be greater than the 7th predetermined threshold value be used to the brightness value of the right side area that judges the second line segment, obtains the 8th judged result;

Whether the 9th judging unit, be greater than the 8th predetermined threshold value be used to the difference between the brightness value of the upper area of the brightness value of the lower zone that judges the first line segment and the first line segment, obtains the 9th judged result;

Whether the tenth judging unit, be greater than the 9th predetermined threshold value be used to the difference between the brightness value in the left zone of the brightness value of the right side area that judges the second line segment and the second line segment, obtains the tenth judged result;

Second place determining unit, for when described the 7th judged result, described the 8th judged result, described the 9th judged result, described the tenth judged result are and are, using the combination of described the first line segment and the second line segment as the bright dark side of candidate boundary line.

Further, described determining unit comprises:

Mould value acquiring unit, for the mean value of the gradient-norm value of the marginal point of the mean value of the gradient-norm value of the marginal point that obtains the bright dark side of candidate boundary line set the first line segment and the second line segment;

The response acquiring unit, be used to the mean value of the gradient-norm value of the marginal point of the mean value of the gradient-norm value of the marginal point that obtains described the first line segment and described the second line segment with the response of value as bright dark side boundary line;

The 3rd position determination unit, for the bright dark side boundary line of the bright dark side boundary line using the response maximum as vehicle head lamp.

Further, described output unit specifically for:

Export the positional information of turning point corresponding to bright dark side boundary line and/or form the level bright dark side boundary line in this bright dark side boundary line and the positional information in oblique bright dark side boundary line; The positional information of the turning point that described bright dark side boundary line is corresponding is determined by the positional information of described the first line segment and the positional information of described the second line segment;

Described device also comprises:

The irradiating angle computing unit, be used to the irradiating angle of the positional information calculation headlamp in the positional information of utilizing turning point corresponding to described bright dark side boundary line and/or the level bright dark side boundary line that forms this bright dark side boundary line and oblique bright dark side boundary line.

Fig. 7 is the bright dark side of the vehicle head lamp based on image provided by the invention boundary line detection system schematic diagram.

The bright dark side of a kind of vehicle head lamp based on image boundary line detection system, described system comprise in-vehicle camera and the bright dark border line detector of vehicle head lamp, wherein:

Described in-vehicle camera 701 is be used to irradiating the irradiation image of vehicle head lamp, and to the bright dark border line detector of described vehicle head lamp, sends the irradiation image of described vehicle head lamp;

The bright dark border line detector 702 of described vehicle head lamp is be used to the irradiation image of the vehicle head lamp that receives described in-vehicle camera and send; Described irradiation image is carried out to Image Edge-Detection, obtain edge image; In described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, based on described the first line segment aggregate and described the second line segment aggregate, meet the boundary line set of the bright dark side of pre-conditioned candidate; Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of a candidate boundary line in the boundary line set of the bright dark side of described candidate in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form; According to the vehicle head lamp beam characteristics, in the bright dark side of the candidate boundary line set obtained, determine the bright dark side boundary line of vehicle head lamp.

It will be appreciated by persons skilled in the art that the bright dark border line detector of vehicle head lamp that the bright dark side of vehicle head lamp of the present invention boundary line detection system comprises can be by device realization shown in Figure 6.

It should be noted that, in this article, relational terms such as the first and second grades only is used for an entity or operation are separated with another entity or operational zone, and not necessarily requires or imply between these entities or operation the relation of any this reality or sequentially of existing.And, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thereby make the process, method, article or the equipment that comprise a series of key elements not only comprise those key elements, but also comprise other key elements of clearly not listing, or also be included as the intrinsic key element of this process, method, article or equipment.In the situation that not more restrictions, the key element limited by statement " comprising ... ", and be not precluded within process, method, article or the equipment that comprises described key element and also have other identical element.

The present invention can describe in the general context of the computer executable instructions of being carried out by computing machine, for example program module.Usually, program module comprises the routine carrying out particular task or realize particular abstract data type, program, object, assembly, data structure etc.Also can in distributed computing environment, put into practice the present invention, in these distributed computing environment, be executed the task by the teleprocessing equipment be connected by communication network.In distributed computing environment, program module can be arranged in the local and remote computer-readable storage medium that comprises memory device.

The above is only the specific embodiment of the present invention; it should be pointed out that for those skilled in the art, under the premise without departing from the principles of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (12)

1. the bright dark border line detecting method of the vehicle head lamp based on image, is characterized in that, described method comprises:

Obtain the irradiation image of vehicle head lamp;

Described irradiation image is carried out to rim detection, obtain edge image;

In described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition; Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of candidate boundary line in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form;

According to the vehicle head lamp beam characteristics, in the bright dark side of the candidate boundary line set obtained, determine the bright dark side boundary line of vehicle head lamp.

2. method according to claim 1, is characterized in that, described detection respectively obtains the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction and comprises:

Detection obtains the first line segment aggregate that meets first direction and is specially: for each marginal point in edge image, obtain the gradient direction angle of marginal point; Whether the difference that judges the angle that the gradient direction angle of marginal point is corresponding with first direction is less than the first predetermined threshold value, obtains the first judged result; When described the first judged result shows that the difference of the angle that the gradient direction angle of marginal point is corresponding with first direction is less than the first predetermined threshold value, preserve described marginal point to the first edge point set and close; The marginal point obtained is carried out to clustering processing, obtain corresponding line segment;

The second line segment aggregate that detection meets second direction is specially: for each marginal point in edge image, obtain the gradient direction angle of marginal point; Whether the difference that judges the angle that the gradient direction angle of marginal point is corresponding with second direction is less than the second predetermined threshold value, obtains the second judged result; When described the second judged result shows that the difference of the angle that the gradient direction angle of marginal point is corresponding with second direction is less than the first predetermined threshold value, preserve described marginal point to the second edge point set and close; The marginal point obtained is carried out to clustering processing, obtain corresponding line segment.

3. method according to claim 2, is characterized in that, described the marginal point obtained carried out to clustering processing, obtains corresponding line segment and comprise:

Obtain the corresponding polar coordinates straight-line equation of each marginal point parameter ρ in described marginal point set, by size parameter ρ is sorted, statistics obtains the number of the marginal point with identical parameters ρ, obtains and has the corresponding parameter ρ of local maximum marginal point number and marginal point set corresponding to this parameter ρ;

Marginal point in marginal point set corresponding to each parameter ρ is sorted;

Marginal point to after sequence, calculate the distance between adjacent two edges point, determines according to distance whether adjacent two edges point belongs to same line segment.

4. method according to claim 1, is characterized in that, obtains and to meet the pre-conditioned bright dark side of a candidate boundary line and comprise based on described the first line segment aggregate and described the second line segment aggregate:

From described the first line segment aggregate, choosing a line segment as the first line segment, from described the second line segment aggregate, choosing a line segment as the second line segment;

Judge whether the horizontal ordinate of the terminal of described the first line segment is less than the horizontal ordinate of the terminal of the second line segment, obtain the 4th judged result;

Judge whether the ordinate of the terminal of described the first line segment is greater than the ordinate of the terminal of the second line segment, obtain the 5th judged result;

Whether the number that judges the marginal point of the number of marginal point of the first line segment and described the second line segment all is greater than the 4th predetermined threshold value, obtains the 6th judged result;

When described the 4th judged result, described the 5th judged result and described the 6th judged result are while being, using the combination of described the first line segment, described the second line segment as the bright dark side of a candidate boundary line.

5. method according to claim 1, is characterized in that, described according to the vehicle head lamp beam characteristics, in the bright dark side of the candidate boundary line set obtained, determines that bright dark side boundary line comprises:

Obtain the response in the bright dark side of each candidate boundary line, described response is sorted, the bright dark side boundary line using the bright dark side of the candidate of response maximum boundary line as vehicle head lamp.

Wherein, the response in the bright dark side of each candidate of described acquisition boundary line comprises:

The mean value of the gradient-norm value of the marginal point of the line segment of the mean value of the gradient-norm value of the marginal point of the line segment of first direction and second direction in the acquisition bright dark side of candidate boundary line;

Obtain the mean value of gradient-norm value of marginal point of line segment of described first direction and described second direction line segment marginal point the gradient-norm value mean value with the response of value as the bright dark side of this candidate boundary line.

6. method according to claim 5, is characterized in that, before the response in each bright dark side boundary line, described method also comprises in obtaining the combination of the bright dark side of candidate boundary line:

Utilize texture region feature and/or regional luminance feature, further screen the bright dark side of candidate boundary line.

7. method according to claim 1, is characterized in that, described method also comprises:

Export the positional information of turning point corresponding to bright dark side boundary line and/or form the level bright dark side boundary line in this bright dark side boundary line and the positional information in oblique bright dark side boundary line.

8. the bright dark border line detector of the vehicle head lamp based on image, is characterized in that, described device comprises:

Image obtains unit, be used to obtaining the irradiation image of vehicle head lamp;

Edge detection unit, for described irradiation image is carried out to rim detection, obtain edge image;

Candidate combinations is chosen unit, be used at described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition; Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of candidate boundary line in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form;

Determining unit, for according to the vehicle head lamp beam characteristics, determine the bright dark side boundary line of vehicle head lamp in the bright dark side of the candidate boundary line set obtained.

9. device according to claim 8, it is characterized in that, described candidate combinations is chosen unit and is comprised that line segment detecting unit and combination choose unit, and described line segment detecting unit comprises the first line segment detecting unit and the second line segment detecting unit, wherein, described the first line segment detecting unit comprises:

The first gradient direction angle acquiring unit, for each marginal point for edge image, obtain the gradient direction angle of marginal point;

Whether the first judging unit, be less than the first predetermined threshold value be used to the difference of the gradient direction angle that the judges marginal point angle corresponding with first direction, obtains the first judged result;

The first storage unit, while for the difference of the gradient direction angle that shows marginal point when described the first judged result angle corresponding with first direction, being less than the first predetermined threshold value, preserving described marginal point to the first edge point set and close;

The clustering processing unit, carry out clustering processing for the marginal point to obtaining, and obtains corresponding line segment;

Wherein, described the second line segment detecting unit comprises:

The second gradient direction angle acquiring unit, for each marginal point for edge image, obtain the gradient direction angle of marginal point;

Whether the second judging unit, be less than the second predetermined threshold value be used to the difference of the gradient direction angle that the judges marginal point angle corresponding with second direction, obtains the second judged result;

The second storage unit, while for the difference of the gradient direction angle that shows marginal point when described the second judged result angle corresponding with second direction, being less than the first predetermined threshold value, preserving described marginal point to the second edge point set and close;

The clustering processing unit, carry out clustering processing for the marginal point to obtaining, and obtains corresponding line segment.

10. device according to claim 9, is characterized in that, described clustering processing unit comprises:

Marginal point set acquiring unit, for obtaining the corresponding polar coordinates straight-line equation of each marginal point parameter ρ of described marginal point set, by size parameter ρ is sorted, statistics obtains the number of the marginal point with identical parameters ρ, obtains and has the corresponding parameter ρ of local maximum marginal point number and marginal point set corresponding to this parameter ρ;

Sequencing unit, sort for the marginal point to marginal point set corresponding to each parameter ρ;

The line segment processing unit, the marginal point for to after sequence, calculate the distance between adjacent two edges point, determines according to distance whether adjacent two edges point belongs to same line segment.

11. device according to claim 8, is characterized in that, described determining unit specifically for:

Obtain the response in the bright dark side of each candidate boundary line, described response is sorted, the bright dark side boundary line using the bright dark side of the candidate of response maximum boundary line as vehicle head lamp.

12. the bright dark side of the vehicle head lamp based on an image boundary line detection system is characterized in that described system comprises in-vehicle camera and the bright dark border line detector of vehicle head lamp, wherein:

Described in-vehicle camera is used for irradiating the irradiation image of vehicle head lamp, and to the bright dark border line detector of described vehicle head lamp, sends the irradiation image of described vehicle head lamp;

The bright dark border line detector of described vehicle head lamp is for the irradiation image of the vehicle head lamp that receives described in-vehicle camera and send; Described irradiation image is carried out to rim detection, obtain edge image; In described edge image, detect respectively and obtain the first line segment aggregate that meets first direction and the second line segment aggregate that meets second direction, meet the boundary line set of the bright dark side of pre-conditioned candidate based on described the first line segment aggregate and described the second line segment aggregate acquisition; Wherein, the corresponding bright dark side of level of described first direction boundary line direction, the corresponding oblique bright dark side of described second direction boundary line direction; The bright dark side of described candidate boundary line in described the first line segment aggregate a line segment and a line segment in described the second line segment aggregate form; According to the vehicle head lamp beam characteristics, in the bright dark side of the candidate boundary line set obtained, determine the bright dark side boundary line of vehicle head lamp.

Images