CN104299221A - Determination method and apparatus for traffic sign image - Google Patents

Abstract

The invention, which belongs to the automobile field, discloses a determination method and apparatus for a traffic sign image. The method comprises: connection zones of all predetermined colors in an RGB image needing to determine a traffic sign image are determined, wherein the predetermined colors consists of a red color, a blue color and a yellow color; according to the determined connection zones, candidate traffic sign images are extracted from the RGB image; edges of all candidate traffic sign images are extracted; the edge of each candidate traffic sign image is segmented into a plurality of line segments; a steering angel between two connected line segments and a steering angle between two disconnected adjacent line segments in an intersecting trend in each edge are calculated; and on the basis of the steering angle values of each edge, a traffic sign image is determined from the candidate traffic sign images. In addition, the apparatus is composed of a first determination module, a first extraction module, a second extraction module, a segmentation module, a calculation module and a second determination module. According to the invention, the color and the shape characteristics of the traffic sign are combined and utilized; and the effect of traffic sign image determination is good.

Description

A kind of defining method of Traffic Sign Images and device

Technical field

The present invention relates to automotive field, particularly a kind of defining method of Traffic Sign Images and device.

Background technology

At present in the research boom of advanced DAS (Driver Assistant System) and automatic driving car, how research identifies that the multiple external environmental information in vehicle running environment belongs to indispensable research contents.These external environmental informations comprise lane line, barrier, structuring or destructuring road surface and traffic sign etc.Wherein, the identification of traffic sign generally for be the traffic sign of speed limit, warning and instruction class.The method of existing identification traffic sign mainly first utilizes the information such as the shape of traffic sign itself to analyze, and is determined by Traffic Sign Images, and then specifically identifies Traffic Sign Images.

Realizing in process of the present invention, inventor finds that prior art at least exists following problem: traffic sign exists disturbed conditions such as being blocked, reflective and fuzzy, the defining method of existing traffic sign is under these disturbed conditions, loss is very many, Detection results is very poor, thus have impact on the identification to traffic sign.

Summary of the invention

The loss height existed in order to the defining method solving existing traffic sign and the problem of Detection results difference, embodiments provide a kind of defining method and device of Traffic Sign Images.Described technical scheme is as follows:

On the one hand, provide a kind of defining method of Traffic Sign Images, described method comprises:

Determine to determine the connected region of each predetermined color in the RGB image of Traffic Sign Images, described predetermined color comprises redness, blueness and yellow;

According to the connected region determined, from described RGB image, extract candidate's Traffic Sign Images;

Extract the edge of each described candidate's Traffic Sign Images;

Be some line segments by the edge segmentation of each described candidate's Traffic Sign Images;

Calculate between two line segments that are connected in edge described in each and do not connect but steering angle between the adjacent segments with crossing trend;

According to the size of steering angle described in edge described in each, from described candidate's Traffic Sign Images, determine Traffic Sign Images.

In the first embodiment, the described connected region determining to determine each predetermined color in the RGB image of Traffic Sign Images, comprising:

According to the transformation for mula of each predetermined color, described need are determined that the RGB image of Traffic Sign Images carries out colour switching respectively, obtains the image highlighting predetermined color; Wherein, the transformation for mula of described redness is C _red=max (0, min ((v _r-v _g)/A, (v _r-v _b)/A)), the transformation for mula of described blueness is C _blue=max (0, min ((v _b-v _r)/A, (v _b-v _g)/A)), the transformation for mula of described yellow is C _yellow=max (0, min ((v _g-v _b)/A, (v _r-v _b)/A)), C _redthe red image obtained after representing conversion, C _bluethe blue image obtained after representing conversion, C _yellowthe yellow image obtained after representing conversion, v _rrepresent R passage pixel value, v _grepresent G passage pixel value, v _brepresent channel B pixel value, A=v _r+ v _g+ v _b;

The image bianry image separately of predetermined color is highlighted described in acquisition;

Determine the connected region in each described bianry image.

In this second embodiment, the connected region that described basis is determined, from described RGB image, extract candidate's Traffic Sign Images, comprising:

The connected region determined described in acquisition area separately, width and height;

From the described connected region determined, determine to meet pre-conditioned connected region; Wherein, be describedly pre-conditionedly that the area of described connected region is in setting areal extent and the ratio of width to height of described connected region is in setting aspect ratio range and the dutycycle of described connected region is greater than setting dutycycle; The height of the width/described connected region of the ratio of width to height=described connected region of described connected region, area/(height of connected region described in the width * of described connected region) of the dutycycle=described connected region of described connected region;

The position of pre-conditioned connected region in described RGB image is met described in acquisition;

In described RGB image, described in extraction, meet the image of pre-conditioned connected region position.

In the third embodiment, the described size according to steering angle described in edge described in each, from described candidate's Traffic Sign Images, determine Traffic Sign Images, comprising:

When steering angles all in the edge at described candidate's Traffic Sign Images all following angular range wherein a kind of described angular range time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images, described angular range comprises [110 °, 130 °], [80 °, 100 °] and [0 °, 60 °].

In the 4th embodiment, the described size according to steering angle described in edge described in each, from described candidate's Traffic Sign Images, determine Traffic Sign Images, comprising:

As steering angle θ ∈ [110 ° all in the edge at described candidate's Traffic Sign Images, 130 °] and the length of three line segments gone out from the edge segmentation of described candidate's Traffic Sign Images is equal time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is equilateral triangle;

As steering angle θ ∈ [80 ° all in the edge at described candidate's Traffic Sign Images, 100 °] and the length of two line segments relative from four line segments that the edge segmentation of described candidate's Traffic Sign Images goes out is equal time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is rectangle;

As steering angle θ ∈ [0 ° all in the edge at described candidate's Traffic Sign Images, 60 °] and when the arc length that the line-fitting that the edge segmentation from described candidate's Traffic Sign Images goes out becomes bowlder to be formed is greater than 2/3rds of circumference, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is circle.

On the other hand, provide a kind of determining device of Traffic Sign Images, described device comprises:

First determination module, for determine to determine Traffic Sign Images RGB image in the connected region of each predetermined color, described predetermined color comprises redness, blueness and yellow;

First extraction module, for according to the connected region determined, from described RGB image, extracts candidate's Traffic Sign Images;

Second extraction module, for extracting the edge of each described candidate's Traffic Sign Images;

Segmentation module, for being some line segments by the edge segmentation of each described candidate's Traffic Sign Images;

Computing module, for calculating between two line segments that are connected in edge described in each and not connecting but steering angle between the adjacent segments with crossing trend;

Second determination module, for the size according to steering angle described in edge described in each, determines Traffic Sign Images from described candidate's Traffic Sign Images.

In the first embodiment, described first determination module comprises:

To described need, converter unit, for the transformation for mula according to each predetermined color, determines that the RGB image of Traffic Sign Images carries out colour switching respectively, obtains the image highlighting predetermined color; Wherein, the transformation for mula of described redness is C _red=max (0, min ((v _r-v _g)/A, (v _r-v _b)/A)), the transformation for mula of described blueness is C _blue=max (0, min ((v _b-v _r)/A, (v _b-v _g)/A)), the transformation for mula of described yellow is C _yellow=max (0, min ((v _g-v _b)/A, (v _r-v _b)/A)), C _redthe red image obtained after representing conversion, C _bluethe blue image obtained after representing conversion, C _yellowthe yellow image obtained after representing conversion, v _rrepresent R passage pixel value, v _grepresent G passage pixel value, v _brepresent channel B pixel value, A=v _r+ v _g+ v _b;

First acquiring unit, for highlighting the image bianry image separately of predetermined color described in obtaining;

First determining unit, for determining the connected region in each described bianry image.

In this second embodiment, described first extraction module comprises:

Second acquisition unit, for connected region area separately, width and the height determined described in obtaining;

Second determining unit, for from the described connected region determined, determines to meet pre-conditioned connected region; Wherein, be describedly pre-conditionedly that the area of described connected region is in setting areal extent and the ratio of width to height of described connected region is in setting aspect ratio range and the dutycycle of described connected region is greater than setting dutycycle; The height of the width/described connected region of the ratio of width to height=described connected region of described connected region, area/(height of connected region described in the width * of described connected region) of the dutycycle=described connected region of described connected region;

3rd acquiring unit, meets the position of pre-conditioned connected region in described RGB image described in obtaining;

Extraction unit, in described RGB image, meets the image of pre-conditioned connected region position described in extraction.

In the third embodiment, described second determination module is used for,

When steering angles all in the edge at described candidate's Traffic Sign Images all following angular range wherein a kind of described angular range time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images, described angular range comprises [110 °, 130 °], [80 °, 100 °] and [0 °, 60 °].

In the 4th embodiment, described second determination module is used for,

As steering angle θ ∈ [110 ° all in the edge at described candidate's Traffic Sign Images, 130 °] and the length of three line segments gone out from the edge segmentation of described candidate's Traffic Sign Images is equal time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is equilateral triangle;

As steering angle θ ∈ [80 ° all in the edge at described candidate's Traffic Sign Images, 100 °] and the length of two line segments relative from four line segments that the edge segmentation of described candidate's Traffic Sign Images goes out is equal time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is rectangle;

As steering angle θ ∈ [0 ° all in the edge at described candidate's Traffic Sign Images, 60 °] and when the arc length that the line-fitting that the edge segmentation from described candidate's Traffic Sign Images goes out becomes bowlder to be formed is greater than 2/3rds of circumference, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is circle.

The beneficial effect that the technical scheme that the embodiment of the present invention provides is brought is:

By the connected region of each predetermined color in the RGB image of determining to determine Traffic Sign Images, this predetermined color comprises redness, blueness and yellow, and according to the connected region determined, from the RGB image need determining Traffic Sign Images, extract candidate's Traffic Sign Images; Because this predetermined color contains the classification color of traffic sign, therefore make use of the color characteristics of traffic sign, make the candidate's Traffic Sign Images extracted compare closing to reality Traffic Sign Images; Extract the edge of each candidate's Traffic Sign Images; Be some line segments by the edge segmentation of each candidate's Traffic Sign Images; To calculate in some line segments that each edge segmentation becomes between two line segments that are connected and do not connect but steering angle between the adjacent segments with crossing trend; According to the size of steering angle in each edge, from candidate's Traffic Sign Images, determine Traffic Sign Images; Because steering angle can assess the shape at edge, therefore make use of the style characteristic of traffic sign, the Traffic Sign Images determined is more accurate; This method combines the CF characteristic that make use of traffic sign, and when determining Traffic Sign Images, effect is better, and loss is very low.

Accompanying drawing explanation

In order to be illustrated more clearly in the technical scheme in the embodiment of the present invention, below the accompanying drawing used required in describing embodiment is briefly described, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these accompanying drawings.

Fig. 1 is the process flow diagram of the defining method of a kind of Traffic Sign Images that the embodiment of the present invention provides;

Fig. 2 is the process flow diagram of the defining method of another Traffic Sign Images that the embodiment of the present invention provides;

Fig. 3 is the schematic diagram of the steering angle that the embodiment of the present invention provides;

Fig. 4 is the successional schematic diagram of the differential of the continuity point that the embodiment of the present invention provides;

Fig. 5 is the structural representation of the determining device of a kind of Traffic Sign Images that the embodiment of the present invention provides;

Fig. 6 is the structural representation of the determining device of another Traffic Sign Images that the embodiment of the present invention provides.

embodiment

For making the object, technical solutions and advantages of the present invention clearly, below in conjunction with accompanying drawing, embodiment of the present invention is described further in detail.

Embodiment one

Embodiments provide a kind of defining method of Traffic Sign Images, see Fig. 1, method flow comprises:

In a step 101, the connected region of each predetermined color in RGB (red, green, blue three Color Channels) image of Traffic Sign Images is determined to determine.

Wherein, this predetermined color comprises redness, blueness and yellow.

In a step 102, according to the connected region determined, from the RGB image need determining Traffic Sign Images, extract candidate's Traffic Sign Images.

In step 103, the edge of each candidate's Traffic Sign Images is extracted.

At step 104, be some line segments by the edge segmentation of each candidate's Traffic Sign Images.

In step 105, calculate between two line segments that are connected in each edge and do not connect but steering angle between the adjacent segments with crossing trend.

Wherein, the computing formula of steering angle is, θ=tan ^-1((V ₁× V ₂)/(|| V ₁|| × || V ₂||)).θ is steering angle, V ₁=P _c-P _p, V ₂=P _n-P _c.P _cbe the coordinate of the tie point of two line segments, P _pfor wherein a line segment removes P _cthe coordinate of another outer end points, P _nfor another line segment is except P _cthe coordinate of another outer end points.

In step 106, according to the size of steering angle in each edge, from candidate's Traffic Sign Images, Traffic Sign Images is determined.

In the first embodiment, when steering angles all in the edge at candidate's Traffic Sign Images all following angular range wherein a kind of angular range time, determine that this candidate's Traffic Sign Images is Traffic Sign Images, angular range comprises [110 °, 130 °], [80 °, 100 °] and [0 °, 60 °].

In this second embodiment, as steering angle θ ∈ [110 ° all in the edge at candidate's Traffic Sign Images, 130 °] and the length of three line segments gone out from the edge segmentation of candidate's Traffic Sign Images is equal time, determine that this candidate's Traffic Sign Images is Traffic Sign Images and the shape of this Traffic Sign Images is equilateral triangle.As steering angle θ ∈ [80 ° all in the edge at candidate's Traffic Sign Images, 100 °] and the length of two line segments relative from four line segments that the edge segmentation of candidate's Traffic Sign Images goes out is equal time, determine that this candidate's Traffic Sign Images is Traffic Sign Images and the shape of this Traffic Sign Images is rectangle.As steering angle θ ∈ [0 ° all in the edge at candidate's Traffic Sign Images, 60 °] and when the arc length that the line-fitting that the edge segmentation from candidate's Traffic Sign Images goes out becomes bowlder to be formed is greater than 2/3rds of circumference, determine that this candidate's Traffic Sign Images is Traffic Sign Images and the shape of this Traffic Sign Images is circle.

The embodiment of the present invention is by the connected region of each predetermined color in the RGB image of determining to determine Traffic Sign Images, this predetermined color comprises redness, blueness and yellow, and according to the connected region determined, from the RGB image need determining Traffic Sign Images, extract candidate's Traffic Sign Images; Because this predetermined color contains the classification color of traffic sign, therefore make use of the color characteristics of traffic sign, make the candidate's Traffic Sign Images extracted compare closing to reality Traffic Sign Images; Extract the edge of each candidate's Traffic Sign Images; Be some line segments by the edge segmentation of each candidate's Traffic Sign Images; To calculate in some line segments that each edge segmentation becomes between two line segments that are connected and do not connect but steering angle between the adjacent segments with crossing trend; According to the size of steering angle in each edge, from candidate's Traffic Sign Images, determine Traffic Sign Images; Because steering angle can assess the shape at edge, therefore make use of the style characteristic of traffic sign, the Traffic Sign Images determined is more accurate; This method combines the CF characteristic that make use of traffic sign, and when determining Traffic Sign Images, effect is better, and loss is very low.

Embodiment two

Embodiments provide a kind of defining method of Traffic Sign Images, see Fig. 2, method flow comprises:

In step 201, according to the transformation for mula of each predetermined color, to determining that the RGB image of Traffic Sign Images carries out colour switching respectively, obtain the image highlighting predetermined color.

Wherein, this predetermined color comprises redness, blueness and yellow.Red transformation for mula is C _red=max (0, min ((v _r-v _g)/A, (v _r-v _b)/A)), blue transformation for mula is C _blue=max (0, min ((v _b-v _r)/A, (v _b-v _g)/A)), yellow transformation for mula is C _yellow=max (0, min ((v _g-v _b)/A, (v _r-v _b)/A)).C _redthe red image obtained after representing conversion, C _bluethe blue image obtained after representing conversion, C _yellowthe yellow image obtained after representing conversion.V _rrepresent R passage pixel value, v _grepresent G passage pixel value, v _brepresent channel B pixel value, A=v _r+ v _g+ v _b.

By colour switching, will obtain three width and highlight the image of predetermined color, be highlighted red, blue and yellow image successively.Consider in practical application, if classified by traffic sign according to color, then traffic sign is divided into redness, blueness and yellow traffic sign.Therefore, highlight redness, blueness and yellow, will be conducive to obtaining Traffic Sign Images region.

Wherein, need determine that the RGB image of Traffic Sign Images can be, the outside road ambient image of vehicle front in vehicle travel process.This outside road ambient image can be taken by in-vehicle camera.This in-vehicle camera can be arranged on the rear side of rear-viewing mirror in vehicle.

In step 202., the image bianry image separately highlighting predetermined color is obtained.

Wherein, can be split by predetermined threshold value, obtain bianry image.Highlight in the bianry image of red image, red area will present white, and other color regions will present black.Similarly, highlighted in the bianry image of blueness or yellow image, blue region or yellow area will present white, and other color regions will present black.

In step 203, the connected region in each bianry image is determined.

Connected region refers to, there is in image same pixel value and position adjacent foreground pixel point composition image-region.In bianry image, foreground pixel point refers to that pixel value is the pixel of white.Two-Pass method or Seed-Filling se ed filling algorithm can be adopted to search connected region.

Achieved by step 201-step 203, determine to determine the connected region of each predetermined color in the RGB image of Traffic Sign Images.

In step 204, connected region area separately, width and the height determined is obtained.

Wherein, the area of connected region refers to that the number of pixels in this region, width refer to the number of pixels of this region in image coordinate shared by Y-direction, highly refers to the number of pixels of this region in image coordinate shared by X-direction.

In step 205, from the connected region determined, determine to meet pre-conditioned connected region.

Wherein, this is pre-conditioned can be, TA _l<CA _j<TA _hand TAR _l<CRAR _j<TAR _hand CRS _j>TCRS.CA _jfor the area of connected region, CRAR _jrepresent the ratio of width to height of connected region, CRAR _j=CW _j/ CH _j, CW _jfor the width of connected region, CH _jfor the height of connected region.CRS _jrepresent the dutycycle of connected region, CRS _j=CA _j/ (CW _j× CH _j).TA _hand TA _lbe respectively two end points of setting areal extent, TAR _land TAR _hbe respectively two end points of setting aspect ratio range, TCRS is setting dutycycle.

Wherein, determine meeting pre-conditioned connected region, the connected region that some can be belonged to chaff interference is got rid of, and adds the follow-up accuracy determining Traffic Sign Images.

In step 206, acquisition meets the position of pre-conditioned connected region in the RGB image need determining Traffic Sign Images.

Wherein, the coordinate of pre-conditioned connected region in bianry image is met identical with meeting the position of pre-conditioned connected region in the RGB image need determining Traffic Sign Images.For ease of extracting, the position that can will meet the boundary rectangle of pre-conditioned connected region in bianry image and cover, as meeting the position of pre-conditioned connected region in the RGB image need determining Traffic Sign Images.Suppose that i-th height meeting the boundary rectangle of pre-conditioned connected region and width are respectively H _iand W _i, the top left co-ordinate of this boundary rectangle is (X _i, Y _i).

In step 207, in the RGB image need determining Traffic Sign Images, extract the image meeting pre-conditioned connected region position.

Wherein, the image meeting pre-conditioned connected region position incites somebody to action alternatively Traffic Sign Images.Can extracting directly, be arranged in and meet pre-conditioned connected region and need determine the image of position of RGB image of Traffic Sign Images.

Can intactly occur in image after treatment in order to ensure candidate's Traffic Sign Images, when extracting, can extract in the lump together with the ambient image regions of candidate's Traffic Sign Images, such as the boundary rectangle meeting pre-conditioned connected region is expanded the length of 1/4th respectively at four direction up and down, and then extract.Suppose that the top left co-ordinate of the candidate's Traffic Sign Images extracted after expanding is respectively (EX _i, EY _i), height and width are respectively EH _iand EW _i, need determine that the height of the RGB image of Traffic Sign Images and width are respectively H and W.Then have,

$E{X}_i=\left\{\begin{array}{c}{X}_i-W_i/4,X_i-W_i/4>0\\ 1,\mathrm{otherwise}\end{array}\right.,{\mathrm{EY}}_i=\left\{\begin{array}{c}{Y}_i-H_i/4,Y_i-H_i/4>0\\ 1,\mathrm{otherwise}\end{array}\right.$

${\mathrm{EW}}_i=\left\{\begin{array}{c}1.5\&times;W_i,1.5\&times;W_i<W\\ W-{\mathrm{EX}}_i,\mathrm{otherwise}\end{array}\right.,{\mathrm{EH}}_i=\left\{\begin{array}{c}1.5\&times;H_i,1.5\&times;H_i<H\\ H-{\mathrm{EY}}_i,\mathrm{otherwise}\end{array}\right.$

Achieved by step 204-step 207, according to the connected region determined, from the RGB image need determining Traffic Sign Images, extract candidate's Traffic Sign Images.

In a step 208, the edge of each candidate's Traffic Sign Images is extracted.

Wherein, this edge refers to the outline edge of each candidate's Traffic Sign Images.Canny algorithm can be adopted to extract the marginal information of each candidate's Traffic Sign Images.

In step 209, be some line segments by the edge segmentation of each candidate's Traffic Sign Images.

Wherein, edge is that some line segments are formed, and in these line segments, some line segments can be tandem arrays, and some line segments can be discrete.According to the Gradient Features at edge, can be some line segments by edge segmentation, comprise, calculate the differential being positioned at each pixel on edge, if the differential value phase difference of continuous image vegetarian refreshments within the specific limits, then regard as these pixels on an approximate line segment.Also can adopt the method that other lines are split, be some line segments by edge segmentation.

It should be noted that, when splitting, the coordinate that can record each line segment two-end-point and the tie point coordinate of two line segments be connected.According to the trend at edge, can also be numbered for the line segment be partitioned into and sort.

In step 210, to calculate in some line segments that each edge segmentation becomes between two line segments that are connected and do not connect but steering angle between the adjacent segments with crossing trend.

Wherein, the computing formula of steering angle is, θ=tan ^-1((V ₁× V ₂)/(|| V ₁|| × || V ₂||)).Be steering angle see Fig. 3, θ, V ₁=P _c-P _p, V ₂=P _n-P _c.P _cbe the coordinate of the tie point of two line segments, P _pfor wherein a line segment removes P _cthe coordinate of another outer end points, P _nfor another line segment is except P _cthe coordinate of another outer end points.It should be noted that, for there is no the adjacent segments of tie point (being numbered serial number of adjacent segments), can according to the end points of line segment, obtain the straight-line equation of line segment place straight line, according to the straight-line equation of line segment place straight line, adjacent segments can be known by inference and whether intersect (whether there is crossing trend) at the bearing of trend of adjacent end points, and obtain the intersecting point coordinate intersected.This intersecting point coordinate is the coordinate of the tie point of adjacent segments.Calculate and do not connect but steering angle between the adjacent segments with crossing trend, can avoid because block the interference of factor, and neglect the steering angle between two line segments that should be connected.

Simply introduce the principle of the computing formula of this steering angle below.

First, on known curve, the curvature formulations of t point is,

κ(t)＝(x′(t)y″(t)-x″(t)y′(t))/[x ^′2(t)+y ^′2(t)] ^3/2

Wherein, k (t) is the curvature of t point on curve, x (t) is t point function in the x direction, y (t) is t point function in y-direction, " (t) is first order derivative and the second derivative of x (t) to x ' (t) respectively, and " (t) is first order derivative and the second derivative of y (t) to y ' (t) respectively with y with x.

Definition α ' (t)=(x ' (t), y ' (t)), α " (t)=(x " (t), y " (t)); according to curvature formulations, the signed curvature value K of t point can be defined _dthe symbol K of (t) and signed curvature _α(t), K _d(t) and K _αt () is respectively,

κ _d(t)＝(α′(t) _xα″(t) _y-α″(t) _xα′(t) _y)/||α′(t)|| ³

Secondly, see Fig. 4, according to the continuity of the differential between continuity point, (t) is respectively for acquisition α ' (t) and α " (t), α ' (t) and α ",

α′(t)＝α(t)-α(t-ε)

α″(t)＝α′(t)-α′(t-ε)＝α(t)-2α(t-ε)+α(t-2ε)

Wherein, α (t) is the coordinate of t point, the coordinate of the previous continuity point t-ε that α (t-ε) is t point, the coordinate of the previous continuity point t-2 ε that α (t-2 ε) is t-ε point.

Then, according to the continuity of the differential between continuity point, determine the two line segments vector be separately connected.

V ₁＝P _c-P _p，V ₂＝P _n-P _c

That is, V ₁and V ₂be respectively P _c(P _cprevious point be P _p) and P _n(P _nprevious point be P _c) differential.In the present embodiment, be regarded the end points of two line segments be connected as continuity point, then according to the continuity of the differential between continuity point, obtain V ₁and V ₂.

Finally, be combined with to curvature value K _dformula and the two line segments vector be separately connected determined of (t), the steering angle between two line segments that acquisition is connected.

This step 210 also comprises: to not connect in each edge but the adjacent segments with crossing trend is revised, and makes not connect in each edge but the adjacent segments with crossing trend extends to intersection point respectively.

For there is no the adjacent segments of tie point (being numbered serial number of adjacent segments), if having crossing trend, after acquisition intersecting point coordinate, adjacent segments is extended to this intersection point respectively.To not connect in each edge but the adjacent segments with crossing trend is revised, be that under the interference considering the factor of blocking, the length in the sideline in the outline edge of Traffic Sign Images can change.

In step 211, according to the size of steering angle in each edge, from candidate's Traffic Sign Images, Traffic Sign Images is determined.

As steering angle θ ∈ [110 ° all in the edge at candidate's Traffic Sign Images, 130 °] and the length of three line segments gone out from the edge segmentation of candidate's Traffic Sign Images is equal time, determine that this candidate's Traffic Sign Images is Traffic Sign Images and the shape of this Traffic Sign Images is equilateral triangle.

As steering angle θ ∈ [80 ° all in the edge at candidate's Traffic Sign Images, 100 °] and the length of two line segments relative from four line segments that the edge segmentation of candidate's Traffic Sign Images goes out is equal time, determine that this candidate's Traffic Sign Images is Traffic Sign Images and the shape of this Traffic Sign Images is rectangle.

As steering angle θ ∈ [0 ° all in the edge at candidate's Traffic Sign Images, 60 °] and when the arc length that the line-fitting that the edge segmentation from candidate's Traffic Sign Images goes out becomes bowlder to be formed is greater than 2/3rds of circumference, determine that this candidate's Traffic Sign Images is Traffic Sign Images and the shape of this Traffic Sign Images is circle.

Wherein, traffic sign, according to Shape Classification, can be divided into the traffic sign of equilateral triangle, rectangle and circle shape.For the Traffic Sign Images of equilateral triangle and rectangle, its steering angle calculated should be 120 ° and 90 °.But consider the precision of the steering angle calculated in actual computation process, think that the steering angle being positioned at [110 °, 130 °] is triangular apex, think that the steering angle being positioned at [80 °, 100 °] is rectangle summit.

Wherein, equilateral triangle is divided into again positive equilateral triangle and falls equilateral triangle.When realizing, can determine to continue after the shape of Traffic Sign Images is equilateral triangle to determine that this equilateral triangle is just equilateral or falls equilateral triangle.If positive equilateral triangle, then a summit is upper, and two summits are lower and horizontal coordinate difference is less; If fall equilateral triangle, then summit under, two summits are upper and horizontal coordinate difference is less; And just and fall the ordinate on a summit of equilateral triangle between the ordinate on two other summit.

Wherein, a square smallest circle approximating method can be adopted, the line-fitting Cheng Yuan gone out by the edge segmentation from candidate's Traffic Sign Images.Square smallest circle approximating method comprises:

First, the coordinate in the center of circle needing fitting circle is determined.

Defining method is as follows: suppose real space R ²in finite point set { (x _i, y _i), 0≤i<N}, definition,

$(\stackrel{\&OverBar;}{x},\stackrel{\&OverBar;}{y})=(\frac{1}{N}\underset{i}{\mathrm{\&Sigma;}}{x}_i,\frac{1}{N}\underset{i}{\mathrm{\&Sigma;}}{y}_i)$

Order aforementioned finite point set is transformed in (u, v) coordinate system.

The coordinate of the center of circle in (u, v) coordinate system is made to be (u _c, v _c), need matching radius of a circle to be r, by equation of a circle g (u, v)=(u-u _c) ²+ (v-v _c) ²the squared minimization of-β, obtains S=∑ _i(g (u _i, v _i)) ².Wherein, β=R ².

To S (β, u _c, v _c) in three parameters, ask for respectively β, u _cpartial differential, that is:

$\frac{\&PartialD;S}{\&PartialD;\mathrm{\&beta;}}=2\underset{i}{\mathrm{\&Sigma;}}g(u_i,v_i)\frac{\&PartialD;g}{\&PartialD;\mathrm{\&beta;}}(u_i,v_i)=-2\underset{i}{\mathrm{\&Sigma;}}g(u_i,v_i)$

$\begin{array}{c}\frac{\&PartialD;S}{\&PartialD;u_c}=2\underset{i}{\mathrm{\&Sigma;}}g(u_i,v_i)\frac{\&PartialD;g}{\&PartialD;u_c}(u_i,v_i)\\ =-4\underset{i}{\mathrm{\&Sigma;}}{u}_{i}g(u_i,v_i)+4u_c\underset{i}{\mathrm{\&Sigma;}}g(u_i,v_i)\end{array}$

If made and if only if ∑ _ig (u _i, v _i)=0 (1).

If made and if only if ∑ _iu _ig (u _i, v _i)=0 1..

1. formula is expanded, and hypothesis ${\mathrm{\&Sigma;}}_i{u}_i[u_i^2-2u_i{u}_c+u_c^2+v_i^2-2v_i{v}_c+v_c^2-\mathrm{\&beta;}]=0$ And define S 2., _u=∑ _iu _i, then 2. formula can be expressed as:

$S_{\mathrm{uuu}}-2u_c{S}_{\mathrm{uu}}+u_c^2{S}_u+S_{\mathrm{uvv}}-2v_c{S}_{\mathrm{uv}}+v_c^2{S}_u-\mathrm{\&beta;}{S}_u=0$ ③

Because S _u=0, then 3. formula can be reduced to:

$u_c{S}_{\mathrm{uu}}+v_c{S}_{\mathrm{uv}}=\frac{1}{2}(S_{\mathrm{uuu}}+S_{\mathrm{uvv}})$ ④

Again to v _cask local derviation, that is:

$\begin{array}{c}\frac{\&PartialD;S}{\&PartialD;u_v}=2\underset{i}{\mathrm{\&Sigma;}}g(u_i,v_i)\frac{\&PartialD;g}{\&PartialD;u_v}(u_i,v_i)\\ =-4\underset{i}{\mathrm{\&Sigma;}}{u}_{i}g(u_i,v_i)+4u_v\underset{i}{\mathrm{\&Sigma;}}g(u_i,v_i)\end{array}$

To make and if only if ∑ _iv _ig (u _i, v _i)=0 5..

Equally, 5. expanded type, defines S _v=0, then have

$u_c{S}_{\mathrm{uv}}+v_c{S}_{\mathrm{vv}}=\frac{1}{2}(S_{\mathrm{vvv}}+S_{\mathrm{vuu}})$ ⑥

Solution formula above 4. with formula 6., simultaneously given (u _c, v _c), be mapped in original coordinate system by these points, then, in original coordinate system, the coordinate in the center of circle is

Secondly, the radius in the center of circle needing fitting circle is determined.

Defining method is as follows, first expands above-mentioned formula (1), obtains:

$\underset{i}{\mathrm{\&Sigma;}}[u_i^2-2u_i{u}_c+u_c^2+v_i^2-2v_i{v}_c+v_c^2-\mathrm{\&beta;}]=0---\left(2\right)$

Make S _u=S _v=0, then (2) formula can be reduced to:

$N(u_c^2+v_c^2-\mathrm{\&beta;})+S_{\mathrm{uu}}+S_{\mathrm{vv}}=0---\left(3\right)$

(3) solution formula, obtains $\mathrm{\&beta;}=u_c^2+v_c^2+(S_{\mathrm{uu}}+S_{\mathrm{vv}})/N,$ Have simultaneously $r=\sqrt{\mathrm{\&beta;}}.$

Then, according to the coordinate in the center of circle and the radius that need fitting circle, the circumference position needing fitting circle is obtained.

Finally, from the line segment that the edge segmentation of candidate's Traffic Sign Images goes out, determine to be positioned at the line segment of the circumference position needing fitting circle.

Be not positioned at the line segment of the circumference position needing fitting circle, not think the edge belonging to circular traffic sign.Consider in actual process, there are some disturbing factors causes the actual line segment belonging to the edge of circular traffic sign not fit on circle, therefore, Helmholtz principle can be adopted judge the edge not being positioned at and needing the line segment of the circumference position of fitting circle whether to belong to same circular traffic sign.The principle of Helmholtz principle is as follows.

For the line segment that a length is n (the pixel sum on this line segment), calculate the horizontal line angle θ ' of each pixel on circular direction on this line segment, if the horizontal line angle θ ' between pixel within the specific limits, then think pixel marshalling.And the pixel number k of statistics marshalling on this line segment.If n and k meets preset formula, then think that this line segment belongs to the edge of same circular traffic sign.

Formwork calculation below the gradient magnitude g (x, y) at pixel (x, y) place and horizontal line orientation angle θ ' is available:

$\left\{\begin{array}{c}{g}_x(x,y)=(I(x+1,y)-I(x,y)+I(x+1,y+1)-I(x,y+1))/2\\ g_y(x,y)=(I(x,y+1)-I(x,y)+I(x+1,y+1)-I(x+1,y))/2\end{array}\right.$

$g(x,y)=\sqrt{g_x{(x,y)}^2+g_y{(x,y)}^2}$

θ＝∠g _xg _y＝arctan(-g _x(x,y)/g _y(x,y))

Wherein, I (x, y) is the brightness value of pixel (x, y).

Preset formula is as follows.

Wherein, N, p and be given value, i={k, k+1 ..., n}.

In the present embodiment, can also determine the shape of Traffic Sign Images after determining Traffic Sign Images, the identification for follow-up Traffic Sign Images provides and identifies basis accurately.

The embodiment of the present invention is by the connected region of each predetermined color in the RGB image of determining to determine Traffic Sign Images, this predetermined color comprises redness, blueness and yellow, and according to the connected region determined, from the RGB image need determining Traffic Sign Images, extract candidate's Traffic Sign Images; Because this predetermined color contains the classification color of traffic sign, therefore make use of the color characteristics of traffic sign, make the candidate's Traffic Sign Images extracted compare closing to reality Traffic Sign Images; Extract the edge of each candidate's Traffic Sign Images; Be some line segments by the edge segmentation of each candidate's Traffic Sign Images; To calculate in some line segments that each edge segmentation becomes between two line segments that are connected and do not connect but steering angle between the adjacent segments with crossing trend; According to the size of steering angle in each edge, from candidate's Traffic Sign Images, determine Traffic Sign Images; Because steering angle can assess the shape at edge, therefore make use of the style characteristic of traffic sign, the Traffic Sign Images determined is more accurate; This method combines the CF characteristic that make use of traffic sign, and when determining Traffic Sign Images, effect is better, and loss is very low.

Embodiment three

See Fig. 5, embodiments provide a kind of determining device of Traffic Sign Images, this device comprises the first determination module 301, first extraction module 302, second extraction module 303, segmentation module 304, computing module 305 and the second determination module 306.

First determination module 301 for, determine to determine the connected region of each predetermined color in the RGB image of Traffic Sign Images, this predetermined color comprises redness, blueness and yellow.

First extraction module 302 for, according to the connected region determined, from RGB image, extract candidate Traffic Sign Images.

Second extraction module 303 for, extract the edge of each candidate's Traffic Sign Images.

Segmentation module 304 for, be some line segments by the edge segmentation of each candidate's Traffic Sign Images.

Computing module 305 for, calculate between two line segments that are connected in each edge and do not connect but steering angle between the adjacent segments with crossing trend.

Second determination module 306 for, according to the size of steering angle in each edge, from candidate's Traffic Sign Images, determine Traffic Sign Images.

The embodiment of the present invention is by the connected region of each predetermined color in the RGB image of determining to determine Traffic Sign Images, this predetermined color comprises redness, blueness and yellow, and according to the connected region determined, from the RGB image need determining Traffic Sign Images, extract candidate's Traffic Sign Images; Because this predetermined color contains the classification color of traffic sign, therefore make use of the color characteristics of traffic sign, make the candidate's Traffic Sign Images extracted compare closing to reality Traffic Sign Images; Extract the edge of each candidate's Traffic Sign Images; Be some line segments by the edge segmentation of each candidate's Traffic Sign Images; To calculate in some line segments that each edge segmentation becomes between two line segments that are connected and do not connect but steering angle between the adjacent segments with crossing trend; According to the size of steering angle in each edge, from candidate's Traffic Sign Images, determine Traffic Sign Images; Because steering angle can assess the shape at edge, therefore make use of the style characteristic of traffic sign, the Traffic Sign Images determined is more accurate; This method combines the CF characteristic that make use of traffic sign, and when determining Traffic Sign Images, effect is better, and loss is very low.

Embodiment four

See Fig. 6, embodiments provide a kind of determining device of Traffic Sign Images, this device comprises the first determination module 401, first extraction module 402, second extraction module 403, segmentation module 404, computing module 405 and the second determination module 406.

First determination module 401 for, determine to determine the connected region of each predetermined color in the RGB image of Traffic Sign Images, this predetermined color comprises redness, blueness and yellow.

In the first embodiment, the first determination module 401 comprises converter unit 4011, first acquiring unit 4012 and the first determining unit 4013.

Converter unit 4011 for, according to the transformation for mula of each predetermined color, to determining that the RGB image of Traffic Sign Images carries out colour switching respectively, obtain the image highlighting predetermined color; Wherein, red transformation for mula is C _red=max (0, min ((v _r-v _g)/A, (v _r-v _b)/A)), blue transformation for mula is C _blue=max (0, min ((v _b-v _r)/A, (v _b-v _g)/A)), yellow transformation for mula is C _yellow=max (0, min ((v _g-v _b)/A, (v _r-v _b)/A)), C _redthe red image obtained after representing conversion, C _bluethe blue image obtained after representing conversion, C _yellowthe yellow image obtained after representing conversion, v _rrepresent R passage pixel value, v _grepresent G passage pixel value, v _brepresent channel B pixel value, A=v _r+ v _g+ v _b.

First acquiring unit 4012 for, obtain and highlight the image bianry image separately of predetermined color.

First determining unit 4013 for, determine the connected region in each bianry image.

First extraction module 402 for, according to the connected region determined, from the RGB image need determining Traffic Sign Images, extract candidate Traffic Sign Images.

In this second embodiment, the first extraction module 402 comprises second acquisition unit 4021, second determining unit 4022, the 3rd acquiring unit 4023 and extraction unit 4024.

Second acquisition unit 4021 for, obtain determine connected region area separately, width and height.

Second determining unit 4022 for, from the connected region determined, determine to meet pre-conditioned connected region; Wherein, this is pre-conditioned is, the area of described connected region is in setting areal extent and the ratio of width to height of described connected region is in setting aspect ratio range and the dutycycle of described connected region is greater than setting dutycycle; The height of the width/described connected region of the ratio of width to height=described connected region of described connected region, area/(height of connected region described in the width * of described connected region) of the dutycycle=described connected region of described connected region.

3rd acquiring unit 4023 for, obtain meet the position of pre-conditioned connected region in the RGB image need determining Traffic Sign Images.

Extraction unit 4024 for, in the RGB image need determining Traffic Sign Images, extract and meet the image of pre-conditioned connected region position.

Second extraction module 403 for, extract the edge of each candidate's Traffic Sign Images.

Segmentation module 404 for, be some line segments by the edge segmentation of each candidate's Traffic Sign Images.

Computing module 405 for, calculate the steering angle between two line segments that are connected in some line segments that each edge segmentation becomes.

Second determination module 406 for, according to the size of steering angle in each edge, from candidate's Traffic Sign Images, determine Traffic Sign Images.

In the third embodiment, second determination module 406 for, when steering angles all in the edge at candidate's Traffic Sign Images all following angular range wherein a kind of angular range time, determine that this candidate's Traffic Sign Images is Traffic Sign Images, angular range comprises [110 °, 130 °], [80 °, 100 °] and [0 °, 60 °].

In the 4th embodiment, second determination module 406 for, as steering angle θ ∈ [110 ° all in the edge at candidate's Traffic Sign Images, 130 °] and the length of three line segments gone out from the edge segmentation of candidate's Traffic Sign Images is equal time, determine that this candidate's Traffic Sign Images is Traffic Sign Images and the shape of this Traffic Sign Images is equilateral triangle; As steering angle θ ∈ [80 ° all in the edge at candidate's Traffic Sign Images, 100 °] and the length of two line segments relative from four line segments that the edge segmentation of candidate's Traffic Sign Images goes out is equal time, determine that this candidate's Traffic Sign Images is Traffic Sign Images and the shape of this Traffic Sign Images is rectangle; As steering angle θ ∈ [0 ° all in the edge at candidate's Traffic Sign Images, 60 °] and when the arc length that the line-fitting that the edge segmentation from candidate's Traffic Sign Images goes out becomes bowlder to be formed is greater than 2/3rds of circumference, determine that this candidate's Traffic Sign Images is Traffic Sign Images and the shape of this Traffic Sign Images is circle.

The embodiment of the present invention is by the connected region of each predetermined color in the RGB image of determining to determine Traffic Sign Images, this predetermined color comprises redness, blueness and yellow, and according to the connected region determined, from the RGB image need determining Traffic Sign Images, extract candidate's Traffic Sign Images; Because this predetermined color contains the classification color of traffic sign, therefore make use of the color characteristics of traffic sign, make the candidate's Traffic Sign Images extracted compare closing to reality Traffic Sign Images; Extract the edge of each candidate's Traffic Sign Images; Be some line segments by the edge segmentation of each candidate's Traffic Sign Images; To calculate in some line segments that each edge segmentation becomes between two line segments that are connected and do not connect but steering angle between the adjacent segments with crossing trend; According to the size of steering angle in each edge, from candidate's Traffic Sign Images, determine Traffic Sign Images; Because steering angle can assess the shape at edge, therefore make use of the style characteristic of traffic sign, the Traffic Sign Images determined is more accurate; This method combines the CF characteristic that make use of traffic sign, and when determining Traffic Sign Images, effect is better, and loss is very low.

It should be noted that: the determining device of the Traffic Sign Images that above-described embodiment provides is when determining Traffic Sign Images, only be illustrated with the division of above-mentioned each functional module, in practical application, can distribute as required and by above-mentioned functions and be completed by different functional modules, inner structure by equipment is divided into different functional modules, to complete all or part of function described above.In addition, the determining device of the Traffic Sign Images that above-described embodiment provides and the determination embodiment of the method for Traffic Sign Images belong to same design, and its specific implementation process refers to embodiment of the method, repeats no more here.

The invention described above embodiment sequence number, just to describing, does not represent the quality of embodiment.

One of ordinary skill in the art will appreciate that all or part of step realizing above-described embodiment can have been come by hardware, the hardware that also can carry out instruction relevant by program completes, described program can be stored in a kind of computer-readable recording medium, the above-mentioned storage medium mentioned can be ROM (read-only memory), disk or CD etc.

The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1. a defining method for Traffic Sign Images, is characterized in that, described method comprises:

Determine to determine the connected region of each predetermined color in the RGB image of Traffic Sign Images, described predetermined color comprises redness, blueness and yellow;

According to the connected region determined, from described RGB image, extract candidate's Traffic Sign Images;

Extract the edge of each described candidate's Traffic Sign Images;

Be some line segments by the edge segmentation of each described candidate's Traffic Sign Images;

Calculate between two line segments that are connected in edge described in each and do not connect but steering angle between the adjacent segments with crossing trend;

According to the size of steering angle described in edge described in each, from described candidate's Traffic Sign Images, determine Traffic Sign Images.

2. method according to claim 1, is characterized in that, the described connected region determining to determine each predetermined color in the RGB image of Traffic Sign Images, comprising:

According to the transformation for mula of each predetermined color, described need are determined that the RGB image of Traffic Sign Images carries out colour switching respectively, obtains the image highlighting predetermined color; Wherein, the transformation for mula of described redness is C _red=max (0, min ((v _r-v _g)/A, (v _r-v _b)/A)), the transformation for mula of described blueness is C _blue=max (0, min ((v _b-v _r)/A, (v _b-v _g)/A)), the transformation for mula of described yellow is C _yellow=max (0, min ((v _g-v _b)/A, (v _r-v _b)/A)), C _redthe red image obtained after representing conversion, C _bluethe blue image obtained after representing conversion, C _yellowthe yellow image obtained after representing conversion, v _rrepresent R passage pixel value, v _grepresent G passage pixel value, v _brepresent channel B pixel value, A=v _r+ v _g+ v _b;

The image bianry image separately of predetermined color is highlighted described in acquisition;

Determine the connected region in each described bianry image.

3. method according to claim 1, is characterized in that, the connected region that described basis is determined, from described RGB image, extracts candidate's Traffic Sign Images, comprising:

The connected region determined described in acquisition area separately, width and height;

From the described connected region determined, determine to meet pre-conditioned connected region; Wherein, be describedly pre-conditionedly that the area of described connected region is in setting areal extent and the ratio of width to height of described connected region is in setting aspect ratio range and the dutycycle of described connected region is greater than setting dutycycle; The height of the width/described connected region of the ratio of width to height=described connected region of described connected region, area/(height of connected region described in the width * of described connected region) of the dutycycle=described connected region of described connected region;

The position of pre-conditioned connected region in described RGB image is met described in acquisition;

In described RGB image, described in extraction, meet the image of pre-conditioned connected region position.

4. method according to claim 1, is characterized in that, the described size according to steering angle described in edge described in each, determines Traffic Sign Images, comprising from described candidate's Traffic Sign Images:

When steering angles all in the edge at described candidate's Traffic Sign Images all following angular range wherein a kind of described angular range time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images, described angular range comprises [110 °, 130 °], [80 °, 100 °] and [0 °, 60 °].

5. method according to claim 1, is characterized in that, the described size according to steering angle described in edge described in each, determines Traffic Sign Images, comprising from described candidate's Traffic Sign Images:

As steering angle θ ∈ [110 ° all in the edge at described candidate's Traffic Sign Images, 130 °] and the length of three line segments gone out from the edge segmentation of described candidate's Traffic Sign Images is equal time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is equilateral triangle;

As steering angle θ ∈ [80 ° all in the edge at described candidate's Traffic Sign Images, 100 °] and the length of two line segments relative from four line segments that the edge segmentation of described candidate's Traffic Sign Images goes out is equal time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is rectangle;

As steering angle θ ∈ [0 ° all in the edge at described candidate's Traffic Sign Images, 60 °] and when the arc length that the line-fitting that the edge segmentation from described candidate's Traffic Sign Images goes out becomes bowlder to be formed is greater than 2/3rds of circumference, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is circle.

6. a determining device for Traffic Sign Images, is characterized in that, described device comprises:

First determination module, for determine to determine Traffic Sign Images RGB image in the connected region of each predetermined color, described predetermined color comprises redness, blueness and yellow;

First extraction module, for according to the connected region determined, from described RGB image, extracts candidate's Traffic Sign Images;

Second extraction module, for extracting the edge of each described candidate's Traffic Sign Images;

Segmentation module, for being some line segments by the edge segmentation of each described candidate's Traffic Sign Images;

Computing module, for calculating between two line segments that are connected in edge described in each and not connecting but steering angle between the adjacent segments with crossing trend;

Second determination module, for the size according to steering angle described in edge described in each, determines Traffic Sign Images from described candidate's Traffic Sign Images.

7. device according to claim 6, is characterized in that, described first determination module comprises:

To described need, converter unit, for the transformation for mula according to each predetermined color, determines that the RGB image of Traffic Sign Images carries out colour switching respectively, obtains the image highlighting predetermined color; Wherein, the transformation for mula of described redness is C _red=max (0, min ((v _r-v _g)/A, (v _r-v _b)/A)), the transformation for mula of described blueness is C _blue=max (0, min ((v _b-v _r)/A, (v _b-v _g)/A)), the transformation for mula of described yellow is C _yellow=max (0, min ((v _g-v _b)/A, (v _r-v _b)/A)), C _redthe red image obtained after representing conversion, C _bluethe blue image obtained after representing conversion, C _yellowthe yellow image obtained after representing conversion, v _rrepresent R passage pixel value, v _grepresent G passage pixel value, v _brepresent channel B pixel value, A=v _r+ v _g+ v _b;

First acquiring unit, for highlighting the image bianry image separately of predetermined color described in obtaining;

First determining unit, for determining the connected region in each described bianry image.

8. device according to claim 6, is characterized in that, described first extraction module comprises:

Second acquisition unit, for connected region area separately, width and the height determined described in obtaining;

Second determining unit, for from the described connected region determined, determines to meet pre-conditioned connected region; Wherein, be describedly pre-conditionedly that the area of described connected region is in setting areal extent and the ratio of width to height of described connected region is in setting aspect ratio range and the dutycycle of described connected region is greater than setting dutycycle; The height of the width/described connected region of the ratio of width to height=described connected region of described connected region, area/(height of connected region described in the width * of described connected region) of the dutycycle=described connected region of described connected region;

3rd acquiring unit, meets the position of pre-conditioned connected region in described RGB image described in obtaining;

Extraction unit, in described RGB image, meets the image of pre-conditioned connected region position described in extraction.

9. device according to claim 6, is characterized in that, described second determination module is used for,

When steering angles all in the edge at described candidate's Traffic Sign Images all following angular range wherein a kind of described angular range time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images, described angular range comprises [110 °, 130 °], [80 °, 100 °] and [0 °, 60 °].

10. device according to claim 6, is characterized in that, described second determination module is used for,

As steering angle θ ∈ [110 ° all in the edge at described candidate's Traffic Sign Images, 130 °] and the length of three line segments gone out from the edge segmentation of described candidate's Traffic Sign Images is equal time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is equilateral triangle;

As steering angle θ ∈ [80 ° all in the edge at described candidate's Traffic Sign Images, 100 °] and the length of two line segments relative from four line segments that the edge segmentation of described candidate's Traffic Sign Images goes out is equal time, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is rectangle;

As steering angle θ ∈ [0 ° all in the edge at described candidate's Traffic Sign Images, 60 °] and when the arc length that the line-fitting that the edge segmentation from described candidate's Traffic Sign Images goes out becomes bowlder to be formed is greater than 2/3rds of circumference, determine that described candidate's Traffic Sign Images is described Traffic Sign Images and the shape of described Traffic Sign Images is circle.