CN103034861A - Identification method and device for truck brake shoe breakdown - Google Patents

Abstract

The invention discloses an identification method for truck brake shoe breakdown. The identification method for the truck brake shoe breakdown comprises extracting segmentation characteristics of three angles from a current image; conforming a characteristic region of a truck brake shoe according to the segmentation characteristics of the three angles; extracting truck brake shoe characteristics from the characteristic region of the truck brake shoe in the current image; utilizing the identification method of a support vector machine (SVM) for the truck brake shoe breakdown and calculating characteristics values of the truck brake shoe; judging whether the truck brake shoe failures or not according to the characteristic values and the presetting fault indentification values. The invention discloses an identification device for the truck brake shoe breakdown. Manual-acting faults can be avoided by utilizing the identification device for the truck brake shoe breakdown. The identification rate of the breakdown can be guaranteed. The accidents can be prevented timely. Therefore, operation safety can be guaranteed.

Description

A kind of recognition methods and device of lorry brake shoe fault

Technical field

The present invention relates to image processing field, relate in particular to a kind of recognition methods and device of lorry brake shoe fault.

Background technology

At present, the lorry Block brake is the tread brake mode that domestic railway freight-car generally adopts, brake shoe drill on the lorry brake shoe is mainly used to prevent that the lorry brake shoe from coming off, in case brake shoe drill is lost, the lorry brake shoe just very likely comes off in the lorry operational process, cause brake failure, if more seriously the lorry brake shoe just in time comes off on rail, then can cause derailment accident.In recent years, the Ministry of Railways widelys popularize lorry operation troubles image dynamic detection system (TFDS) each key position of operating train is carried out imaging, and finishes Fault Identification by artificial image browsing.The problems such as there is poor efficiency in this mode by manually carrying out Fault Identification, and discrimination is unstable can't satisfy the demand for development of safe train operation.

As seen, use TFDS to carry out the brake shoe Fault Identification in the prior art, too rely on manually-operated, therefore can't guarantee the discrimination of fault, and then Accident prevention occurs timely, thereby can't guarantee operation security.

Summary of the invention

In view of this, the object of the present invention is to provide a kind of recognition methods and device of lorry brake shoe fault, can avoid manually-operated error, guarantee the discrimination of fault, Accident prevention occurs timely, thereby guarantees operation security.

For achieving the above object, technical scheme of the present invention is achieved in that

The invention provides a kind of recognition methods of lorry brake shoe fault, the method comprises:

From present image, extract the feature of cutting apart of three angles;

The lorry brake shoe characteristic area that feature is determined present image of cutting apart according to described three angles;

From the lorry brake shoe characteristic area of present image, extract the feature of lorry brake shoe;

Use support vector machine (SVM, Support Vector Machine) algorithm that the feature calculation of lorry brake shoe is drawn the eigenwert of present image, judge according to described eigenwert and the Fault Identification value that presets whether the lorry brake shoe exists fault.

In the such scheme, the described feature of cutting apart of from present image, extracting three angles, comprising: periodically extract present image from TDFS, the Gray Projection that present image is carried out three angles obtains three drop shadow curves; All drop shadow curves are carried out filtering, with the maximal value in filtered each drop shadow curve all as the feature of cutting apart of three angles of present image.

In the such scheme, the described lorry brake shoe characteristic area that feature is determined present image of cutting apart according to three angles, comprise: the marginal information of using Canny operator extraction present image, be the feature of cutting apart of zero degree according to angle, determine the left margin of lorry brake shoe characteristic area and the coordinate figure of right margin; According to the cut apart feature of angle for-25 ° and 25 °, determine the coboundary of lorry brake shoe characteristic area and the coordinate figure of lower boundary.

In the such scheme, the described feature of from the lorry brake shoe characteristic area of present image, extracting the lorry brake shoe, comprise: use the method for estimating based on background area, the image segmentation in the lorry brake shoe characteristic area of present image is obtained lorry brake shoe zone binary picture; Use the element marking method, from the binary picture of described lorry brake shoe zone, extract largest connected zone; Use the Canny operator, extract lorry brake shoe edge contour in the binary picture from largest connected zone; Extract the feature of lorry brake shoe according to described lorry brake shoe edge contour.

In the such scheme, the feature of described lorry brake shoe comprises: the convex and concave feature value of the smooth features value of lorry brake shoe edge contour, lorry brake shoe edge contour, the jaggedness of lorry brake shoe edge contour, lorry brake shoe edge contour solid by the compactness of value, lorry brake shoe edge contour, the circle of lorry brake shoe edge contour, the length breadth ratio of lorry brake shoe edge contour, the area of lorry brake shoe edge contour and the girth of lorry brake shoe edge contour.

In the such scheme, described from present image, the extraction the cutting apart before the feature of three angles, the method also comprises: use the lorry brake shoe drill image that does not have fault and exist the lorry brake shoe drill image of fault to set up respectively the positive and negative samples training set; Extract respectively all Characteristic of Images in the positive and negative samples training set, form nine dimensional feature vectors corresponding to positive and negative samples training set, value corresponding to positive and negative samples of using the SVM algorithm to calculate, the value that the negative sample training set is corresponding is as the Fault Identification value.

The present invention also provides a kind of recognition device of lorry brake shoe fault, and this device comprises: characteristic extracting module and identification module; Wherein,

Characteristic extracting module, be used for extracting from present image the feature of cutting apart of three angles, the lorry brake shoe characteristic area that feature is determined present image of cutting apart according to described three angles, from the lorry brake shoe characteristic area of present image, extract the feature of lorry brake shoe, the feature of described lorry brake shoe is sent to identification module;

Identification module be used for to use the feature calculation of the described lorry brake shoe that the SVM algorithm sends characteristic extracting module to draw the eigenwert of present image, judges according to described eigenwert and the Fault Identification value that presets whether the lorry brake shoe exists fault.

In the such scheme, described characteristic extracting module, concrete being used for periodically extracted present image from the TDFS at place, and the Gray Projection that present image is carried out three angles obtains three drop shadow curves; All drop shadow curves are carried out filtering, with the maximal value in filtered each drop shadow curve all as the feature of cutting apart of three angles of present image.

In the such scheme, described characteristic extracting module, the concrete marginal information that is used for using Canny operator extraction present image is the feature of cutting apart of zero degree according to angle, determines the left margin of lorry brake shoe characteristic area and the coordinate figure of right margin; According to the cut apart feature of angle for-25 ° and 25 °, determine the coboundary of lorry brake shoe characteristic area and the coordinate figure of lower boundary.

In the such scheme, described characteristic extracting module, concrete for using the method for estimating based on background area, the image segmentation in the lorry brake shoe characteristic area of present image is obtained lorry brake shoe zone binary picture; Use the element marking method, from the binary picture of described lorry brake shoe zone, extract largest connected zone; Use the Canny operator, extract lorry brake shoe edge contour in the binary picture from largest connected zone; Extract the feature of lorry brake shoe according to described lorry brake shoe edge contour.

In the such scheme, described characteristic extracting module, the girth of the length breadth ratio of the solid compactness by value, lorry brake shoe edge contour of the convex and concave feature value of concrete smooth features value for extraction lorry brake shoe edge contour, lorry brake shoe edge contour, the jaggedness of lorry brake shoe edge contour, lorry brake shoe edge contour, the circle of lorry brake shoe edge contour, lorry brake shoe edge contour, the area of lorry brake shoe edge contour and lorry brake shoe edge contour is as the feature of lorry brake shoe.

In the such scheme, described identification module also is used for using the lorry brake shoe drill image that does not have fault and existing the lorry brake shoe drill image of fault to set up respectively the positive and negative samples training set; Extract respectively all Characteristic of Images in the positive and negative samples training set, form nine dimensional feature vectors corresponding to positive and negative samples training set, value corresponding to positive and negative samples of using the SVM algorithm to calculate, the value that the negative sample training set is corresponding is as the Fault Identification value.

Recognition methods and the device of lorry brake shoe fault provided by the present invention can be automatically extract the feature of cutting apart of three angles from present image; The lorry brake shoe characteristic area that feature is determined present image of cutting apart according to described three angles; From the lorry brake shoe characteristic area of present image, extract the feature of lorry brake shoe; And then according to the feature of described lorry brake shoe, and the Fault Identification condition determines whether the lorry brake shoe exists fault.So, just can avoid manually-operated error, guarantee the discrimination of fault, Accident prevention occurs timely, thereby guarantees operation security.

Description of drawings

Fig. 1 is the recognition methods schematic flow sheet of lorry brake shoe fault of the present invention;

Fig. 2 is that the recognition device of lorry brake shoe fault of the present invention forms structural representation;

Fig. 3 is table with test results.

Embodiment

Basic thought of the present invention is: the feature of cutting apart of extracting three angles from present image; The lorry brake shoe characteristic area that feature is determined present image of cutting apart according to described three angles; From the lorry brake shoe characteristic area of present image, extract the feature of lorry brake shoe; Use the SVM algorithm that the feature calculation of lorry brake shoe is drawn the eigenwert of present image, judge according to described eigenwert and the Fault Identification value that presets whether the lorry brake shoe exists fault.

The present invention is further described in more detail below in conjunction with drawings and the specific embodiments.

The recognition methods of lorry brake shoe fault of the present invention as shown in Figure 1, may further comprise the steps:

Step 101: the feature of cutting apart of from present image, extracting three angles.

Concrete, from TDFS, periodically extracting present image, the Gray Projection that present image is carried out three angles obtains three drop shadow curves; All drop shadow curves are carried out filtering, with the maximal value in filtered each drop shadow curve all as the feature of cutting apart of three angles of present image;

Here, described three angles-25 ° 0 ° and 25 ° of three angles for setting;

Described Gray Projection comprises: present image is expanded the image after being expanded; Image after the expansion is carried out projection calculate, the image that projection is calculated disperses and obtains drop shadow curve.

Describedly present image is expanded to mend gray scale in the image border be zero pixel, the error that cause the image border when avoiding projection, can use following formula to calculate:

$\&ForAll;(i,j)\&Element;{\mathrm{<figure-callout\; id="2"\; label="R"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">R</figure-callout>}}^2,$ $I^{\&prime;}(i,j)=\left\{\begin{array}{cc}I(i,j)& (i,j)\&Element;\mathrm{\&Omega;}\\ 0& (i,j)\&NotElement;\mathrm{\&Omega;}\end{array}\right.;$ Wherein, Ω represents infinity.

Wherein, described projection is calculated and can be used formula:

Wherein, $i=\sqrt{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">r</figure-callout>}}^2+{\mathrm{<figure-callout\; id="2"\; label="l"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">l</figure-callout>}}^2}\cos [\mathrm{\&theta;}+a\tan \left(\frac{l}{r}\right)],$ $j=\sqrt{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">r</figure-callout>}}^2+{\mathrm{<figure-callout\; id="2"\; label="l"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">l</figure-callout>}}^2}\sin [\mathrm{\&theta;}+a\tan \left(\frac{l}{r}\right)],$ Suppose to set up coordinate system, true origin is o, establishes present image I and highly is m, and width is n,

Be projecting direction, two-dimensional image I will be projected to projection line

Starting point is o,

With

Mutually vertical.R is any point on the projection line, and p was the r point, and with

Parallel straight line, l for any point (i, j) on it from projection line

Vertical range, I (i, j) is the grey scale pixel value that image is located at point (i, j), establishes projection line and x axle angle theta is the projecting direction angle, θ is-25 ° 0 ° or 25 °.

The described image that projection is calculated disperses and can use following formula to calculate:

Wherein,

θ is-25 ° 0 ° and 25 °,

Present image I highly is m, and width is n, δ _θ(r) Gray Projection of presentation video on the θ of projecting direction angle, it can reflect that image is in the grey scale change of projecting direction angle θ direction.

Describedly all drop shadow curves are carried out filtering can use formula:

$h\left(x\right)=\left\{\begin{array}{ccc}1& x_2\&le;x\&le;x_1& \\ -1& x_3\&le;x<x_1\&cup;& x_2<x\&le;x_4\end{array}\right.;$ Wherein, x ₁, x ₂, x ₃, x ₄Be constant, satisfy $\left\{\begin{array}{c}|x_1-x_2|=H\\ |x_3-x_4|=2H\end{array}\right.,$ H estimates for cutting apart characteristic width.

Step 102: according to the lorry brake shoe characteristic area that feature is determined present image of cutting apart of described three angles.

Be specially: using the marginal information of Canny operator extraction present image, is the feature of cutting apart of zero degree according to angle, determines the left margin of lorry brake shoe characteristic area and the coordinate figure of right margin; According to the cut apart feature of angle for-25 ° and 25 °, determine the coboundary of lorry brake shoe characteristic area and the coordinate figure of lower boundary.

Here, described is the feature of cutting apart of zero degree according to angle, determine that the left margin of lorry brake shoe characteristic area and the coordinate figure of right margin comprise: use angle is two number of sub images about cutting apart feature the marginal information of present image being divided into of zero degree, calculate respectively the grey scale pixel value sum of two number of sub images, the size of the grey scale pixel value sum of more described two number of sub images, if the grey scale pixel value sum of left side subimage is larger, then the coordinate figure of the right margin of lorry brake shoe characteristic area equals the abscissa value of cutting apart feature of zero degree, deduct the right margin distance value that presets, the coordinate figure that the coordinate figure of the left margin of lorry brake shoe characteristic area equals its right margin deducts the width value that presets; Otherwise the coordinate figure of the left margin of lorry brake shoe characteristic area equals the abscissa value of cutting apart feature of zero degree, adds the left margin distance value that presets, and the coordinate figure that the coordinate figure of the right margin of lorry brake shoe characteristic area equals its left margin adds the width value that presets;

Such as, the marginal information of supposing present image is I, angle is to be characterized as l among the figure cutting apart of zero degree ₂, two number of sub images were respectively I about the marginal information I of present image was divided into ₁And I ₂Calculate respectively two number of sub images I ₁And I ₂Grey scale pixel value sum edge1 and edge2; If edge1＞edge2, then B _Rigth=l ₂-h ₁, B _Left=B _Rigth-N; When edge1＜edge2, B _Left=l ₂+ h ₂, B _Right=B _Left+ N; Wherein, l ₂Be the abscissa value of cutting apart feature of zero degree, h ₁Be the right margin distance value that presets, h ₂Be the left margin distance value that presets, N is the width value that presets;

Wherein, $\mathrm{<figure-callout\; id="1"\; label="edge"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">edge</figure-callout>}1=\underset{j=0}{\overset{{\mathrm{<figure-callout\; id="1"\; label="N"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">N</figure-callout>}}_1}{\mathrm{\&Sigma;}}}\underset{i=0}{\overset{M_1}{\mathrm{\&Sigma;}}}{I}_1(i,j),$ $\mathrm{<figure-callout\; id="2"\; label="edge"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">edge</figure-callout>}2=\underset{j=0}{\overset{{\mathrm{<figure-callout\; id="2"\; label="N"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">N</figure-callout>}}_2}{\mathrm{\&Sigma;}}}\underset{i=0}{\overset{M_2}{\mathrm{\&Sigma;}}}{I}_2(i,j);$ I wherein ₁(i, j) is subimage I ₁At the grey scale pixel value that point (i, j) is located, M ₁, N ₁Be I ₁Height and width; I ₂(i, j) is subimage I ₂At the grey scale pixel value that point (i, j) is located, M ₂, N ₂Be subimage I ₂Height and width.

Describedly determine the coboundary of lorry brake shoe characteristic area and the coordinate figure of lower boundary according to the feature of cutting apart of angle for-25 ° and 25 °, comprising: establish angle and be-25 ° cut apart the corresponding straight line l of feature ₁With image coboundary intersection point abscissa value be top1, lower boundary intersection point abscissa value is bot1; Angle be 25 ° cut apart the corresponding straight line l of feature ₃With image coboundary intersection point abscissa value be top3, lower boundary intersection point abscissa value is bot3, calculating ratio ratio=| (top1-top3)/(bot1-bot3) |; When ratio＜1, lorry brake shoe drill zone is positioned at the tip position of present image; When ratio＞1, lorry brake shoe drill zone is positioned at the bottom position of present image; According to the physical construction characteristics in lorry brake shoe drill zone, extract the coordinate figure of coboundary and lower boundary according to lax principle.Wherein, described lax principle is prior art, is used for guaranteeing selected lorry brake shoe characteristic area energy perfect representation lorry brake shoe, and its implementation is not done here and given unnecessary details.

Step 103: the feature of from the lorry brake shoe characteristic area of present image, extracting the lorry brake shoe.

Concrete, use the method for estimating based on background area, the image segmentation in the lorry brake shoe characteristic area of present image is obtained lorry brake shoe zone binary picture; Use the element marking method, from the binary picture of described lorry brake shoe zone, extract largest connected zone; Use the Canny operator, extract lorry brake shoe edge contour in the binary picture from largest connected zone; Extract the feature of lorry brake shoe according to described lorry brake shoe edge contour.

Here, the method that described background area is estimated is prior art, does not do here and gives unnecessary details; The element marking method is prior art, does not do here and gives unnecessary details; The Canny operator is prior art, does not do here and gives unnecessary details.

The feature of described lorry brake shoe comprises: the convex and concave feature value of the smooth features value of lorry brake shoe edge contour, lorry brake shoe edge contour, the jaggedness of lorry brake shoe edge contour, lorry brake shoe edge contour solid by the compactness of value, lorry brake shoe edge contour, the circle of lorry brake shoe edge contour, the length breadth ratio of lorry brake shoe edge contour, the area of lorry brake shoe edge contour and the girth of lorry brake shoe edge contour.

Wherein, the calculating of the smooth features value of described lorry brake shoe edge contour comprises: calculate the central point of lorry brake shoe edge contour, centered by described central point lorry brake shoe edge contour is expressed as polar coordinates; The described lorry brake shoe edge contour that is expressed as polar form is carried out discretize calculate, obtain the lorry brake shoe edge contour of discretize; After the lorry brake shoe edge contour of described discretize carried out normalization and calculate, sort according to the rule that presets, again the result who obtains after the ordering is carried out wavelet decomposition and obtain wavelet coefficient, use wavelet coefficient to calculate the smooth features value;

The described central point that calculates lorry brake shoe edge contour can comprise: suppose that lorry brake shoe edge contour is x, then take relative arc length s as parameter, then lorry brake shoe edge contour can be expressed as:

0≤s≤1 wherein,

Be lorry brake shoe edge contour; Then the central point computing formula of lorry brake shoe edge contour is $\stackrel{\&RightArrow;}{x_0}={\&Integral;}_0^1\stackrel{\&RightArrow;}{x\left(s\right)}\mathrm{ds}=({\&Integral;}_0^{1}x\left(s\right)\mathrm{ds},{\&Integral;}_0^{1}y\left(s\right)\mathrm{ds}),$ 0≤s≤1 wherein.

Describedly centered by central point, lorry brake shoe edge contour is expressed as polar coordinates and can be expressed as:

$\left\{\begin{array}{c}\mathrm{\&theta;}\left(s\right)=\&angle;(\stackrel{\&RightArrow;}{x}\left(s\right)-\stackrel{\&RightArrow;}{x_0})\\ r\left(s\right)={\left|\right|\stackrel{\&RightArrow;}{x}\left(s\right)-\stackrel{\&RightArrow;}{x_0}\left|\right|}_2\\ \stackrel{\&RightArrow;}{z}\left(t\right)=(\mathrm{\&theta;}\left(s\right),\left(r\left(s\right)\right).\end{array}\right.,$ Wherein, ∠ represents the polar angle of vector under the polar coordinate system.

Describedly the described lorry brake shoe edge contour that is expressed as polar form is carried out discretize calculate and to be the polar coordinates (r (s that obtains each discrete point on the profile _i), θ (s _i)), wherein

I=0,1 ... N-1, N are the discrete point number.

Described normalization is calculated can use formula:

The rule that described basis presets sorts and can comprise: note θ ' (s _i) be θ ' (i), r ' (s _i) be r ' (i), r ' (i) is resequenced: R (i)=r ' (η (i)) so that right

θ (η (i))＜θ (η (j)) is arranged;

Describedly the result that obtains after the ordering is carried out wavelet decomposition obtain wavelet coefficient and can comprise: $d\left(i\right)=r^{\&prime;}\left(\mathrm{\&eta;}\left(i\right)\right)\&CircleTimes;\mathrm{\&psi;}\left(i\right);$

Described use wavelet coefficient calculates the smooth features value can be E=||d (i) || ₂

The formula that the calculating of the convex and concave feature value of described lorry brake shoe edge contour is used is:

Wherein, c (n), n=1, the sequence of points of 2N lorry brake shoe edge contour, c (n)=|| c (n)-c ₀||, n=1,2 ..., N, $c_0=\frac{1}{N}\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}c\left(n\right);$

The rectangular degree computing formula of described lorry brake shoe edge contour can for:

A wherein ₀The area of profile, A _MERBe the area of the minimum boundary rectangle of profile, acquisition methods is prior art, does not do here and gives unnecessary details.

Described lorry brake shoe edge contour admittedly by the property computing formula can for:

Wherein, A represents the area of profile, and CA is the area of its minimal convex polygon, and acquisition methods is prior art, does not do here and gives unnecessary details.

The compactness computing formula of described lorry brake shoe edge contour can for:

Wherein, P represents the girth of profile, and A represents area, and acquisition methods is prior art, does not do here and gives unnecessary details.

The circle computing formula of described lorry brake shoe edge contour can for:

Wherein,

μ _RIt is the mean distance from the regional barycenter to the frontier point. δ _RBe from the regional barycenter to the frontier point apart from mean square deviation.

The length breadth ratio of described lorry brake shoe edge contour is the ratio of wide and length of the minimum boundary rectangle of lorry brake shoe edge contour;

The area of described lorry brake shoe edge contour can be added up acquisition for the number of pixels in: the lorry brake shoe edge contour scope.

The girth of described lorry brake shoe edge contour can be the boundary length of lorry brake shoe edge contour.

Step 104: use the SVM algorithm that the feature calculation of lorry brake shoe is drawn the eigenwert of present image, judge according to described eigenwert and the Fault Identification value that presets whether the lorry brake shoe exists fault.

Here, definite method of described Fault Identification value is: use the lorry brake shoe image that does not have fault and exist the lorry brake shoe image of fault to set up respectively the positive and negative samples training set; Extract respectively all Characteristic of Images in the positive and negative samples training set, form nine dimensional feature vectors corresponding to positive and negative samples training set, value corresponding to positive and negative samples of using the SVM algorithm to calculate, the value that the negative sample training set is corresponding is as the Fault Identification value.Wherein, described SVM algorithm be embodied as prior art, do not do here and give unnecessary details.

Describedly judge according to eigenwert and the Fault Identification value that presets whether the lorry brake shoe exists fault to comprise: if the eigenwert of present image equals the Fault Identification value, determine that then the lorry brake shoe breaks down; Otherwise, determine that the lorry brake shoe does not break down.

As shown in Figure 2, the invention provides a kind of recognition device of lorry brake shoe fault, this device comprises: characteristic extracting module 21 and identification module 22; Wherein,

Characteristic extracting module 21, be used for extracting from present image the feature of cutting apart of three angles, the lorry brake shoe characteristic area that feature is determined present image of cutting apart according to described three angles, from the lorry brake shoe characteristic area of present image, extract the feature of lorry brake shoe, the feature of described lorry brake shoe is sent to identification module 22;

Identification module 22 be used for to use the feature calculation of the described lorry brake shoe that the SVM algorithm sends characteristic extracting module 21 to draw the eigenwert of present image, judges according to described eigenwert and the Fault Identification value that presets whether the lorry brake shoe exists fault.

Described characteristic extracting module 21, concrete being used for periodically extracted present image from the TDFS at place, the Gray Projection that present image is carried out three angles obtains three drop shadow curves, all drop shadow curves are carried out filtering, with the maximal value in filtered each drop shadow curve all as the feature of cutting apart of three angles of present image.Wherein, described three angles-25 ° 0 ° and 25 ° of three angles for setting.

Described characteristic extracting module 21, concrete being used for expanded present image the image after being expanded; Image after the expansion is carried out projection calculate, the image that projection is calculated disperses and obtains drop shadow curve.

Described characteristic extracting module 21, concrete is zero pixel for mend gray scale in the image border, the error that cause the image border when avoiding projection is expanded present image, can use following formula to calculate:

$\&ForAll;(i,j)\&Element;{\mathrm{<figure-callout\; id="2"\; label="R"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">R</figure-callout>}}^2,$ $I^{\&prime;}(i,j)=\left\{\begin{array}{cc}I(i,j)& (i,j)\&Element;\mathrm{\&Omega;}\\ 0& (i,j)\&NotElement;\mathrm{\&Omega;}\end{array}\right.;$ Wherein, Ω represents infinity.

Described characteristic extracting module 21, the concrete following formula of use that is used for carries out projection calculating:

Wherein, $i=\sqrt{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">r</figure-callout>}}^2+{\mathrm{<figure-callout\; id="2"\; label="l"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">l</figure-callout>}}^2}\cos [\mathrm{\&theta;}+a\tan \left(\frac{l}{r}\right)],$ $j=\sqrt{{\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">r</figure-callout>}}^2+{\mathrm{<figure-callout\; id="2"\; label="l"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">l</figure-callout>}}^2}\sin [\mathrm{\&theta;}+a\tan \left(\frac{l}{r}\right)],$ Suppose to set up coordinate system, true origin is o, establishes present image I and highly is m, and width is n,

Be projecting direction, two-dimensional image I will be projected to projection line

Starting point is o,

With

Mutually vertical.R is any point on the projection line, and p was the r point, and with

Parallel straight line, l for any point (i, j) on it from projection line

Vertical range, I (i, j) is the grey scale pixel value that image is located at point (i, j), establishes projection line and x axle angle theta is the projecting direction angle, θ is-25 ° 0 ° or 25 °.

Described characteristic extracting module 21, concrete being used for uses following formula that the image that projection calculates is dispersed:

Wherein,

θ is-25 ° 0 ° and 25 °,

Present image I highly is m, and width is n, δ _θ(r) Gray Projection of presentation video on the θ of projecting direction angle, it can reflect that image is in the grey scale change of projecting direction angle θ direction.

Described characteristic extracting module 21, the concrete following formula of use that is used for carries out filtering to all drop shadow curves:

$h\left(x\right)=\left\{\begin{array}{ccc}1& x_2\&le;x\&le;x_1& \\ -1& {\mathrm{<figure-callout\; id="3"\; label="x"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">x</figure-callout>}}_3\&le;x<x_1\&cup;& x_2<x\&le;x_4\end{array}\right.;$ Wherein, x ₁, x ₂, x ₃, x ₄Be constant, satisfy $\left\{\begin{array}{c}|x_1-x_2|=H\\ |x_3-x_4|=2H\end{array}\right.,$ H estimates for cutting apart characteristic width.

Described characteristic extracting module 21, the concrete marginal information that is used for using Canny operator extraction present image is the feature of cutting apart of zero degree according to angle, determines the left margin of lorry brake shoe characteristic area and the coordinate figure of right margin; According to the cut apart feature of angle for-25 ° and 25 °, determine the coboundary of lorry brake shoe characteristic area and the coordinate figure of lower boundary.

Described characteristic extracting module 21, the concrete use angle that is used for is two number of sub images about cutting apart feature the marginal information of present image being divided into of zero degree, calculate respectively the grey scale pixel value sum of two number of sub images, the size of the grey scale pixel value sum of more described two number of sub images, if the grey scale pixel value sum of left side subimage is larger, then the coordinate figure of the right margin of lorry brake shoe characteristic area equals the abscissa value of cutting apart feature of zero degree, deduct the right margin distance value that presets, the coordinate figure that the coordinate figure of the left margin of lorry brake shoe characteristic area equals its right margin deducts the width value that presets; Otherwise the coordinate figure of the left margin of lorry brake shoe characteristic area equals the abscissa value of cutting apart feature of zero degree, adds the left margin distance value that presets, and the coordinate figure that the coordinate figure of the right margin of lorry brake shoe characteristic area equals its left margin adds the width value that presets;

Such as, the marginal information of supposing present image is I, angle is to be characterized as l among the figure cutting apart of zero degree ₂, two number of sub images were respectively I about the marginal information I of present image was divided into ₁And I ₂Calculate respectively two number of sub images I ₁And I ₂Grey scale pixel value sum edge1 and edge2; If edge1＞edge2, then B _Rigth=l ₂-h ₁, B _Left=B _Rigth-N; When edge1＜edge2, B _Left=l ₂+ h ₂, B _Right=B _Left+ N; Wherein, l ₂Be the abscissa value of cutting apart feature of zero degree, h ₁Be the right margin distance value that presets, h ₂Be the left margin distance value that presets, N is the width value that presets;

Wherein, $\mathrm{<figure-callout\; id="1"\; label="edge"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">edge</figure-callout>}1=\underset{j=0}{\overset{{\mathrm{<figure-callout\; id="1"\; label="N"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">N</figure-callout>}}_1}{\mathrm{\&Sigma;}}}\underset{i=0}{\overset{M_1}{\mathrm{\&Sigma;}}}{I}_1(i,j),$ $\mathrm{<figure-callout\; id="2"\; label="edge"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">edge</figure-callout>}2=\underset{j=0}{\overset{{\mathrm{<figure-callout\; id="2"\; label="N"\; filenames="HDA00002586069100021.png"\; state="\{\{state\}\}">N</figure-callout>}}_2}{\mathrm{\&Sigma;}}}\underset{i=0}{\overset{M_2}{\mathrm{\&Sigma;}}}{I}_2(i,j);$ I wherein ₁(i, j) is subimage I ₁At the grey scale pixel value that point (i, j) is located, M ₁, N ₁Be I ₁Height and width; I ₂(i, j) is subimage I ₂At the grey scale pixel value that point (i, j) is located, M ₂, N ₂Be subimage I ₂Height and width.

Described characteristic extracting module 21, concrete be used for establishing angle for-25 ° cut apart the corresponding straight line l of feature ₁With image coboundary intersection point abscissa value be top1, lower boundary intersection point abscissa value is bot1; Angle be 25 ° cut apart the corresponding straight line l of feature ₃With image coboundary intersection point abscissa value be top3, lower boundary intersection point abscissa value is bot3, calculating ratio ratio=| (top1-top3)/(bot1-bot3) |; When ratio＜1, lorry brake shoe drill zone is positioned at the tip position of present image; When ratio＞1, lorry brake shoe drill zone is positioned at the bottom position of present image; According to the physical construction characteristics in lorry brake shoe drill zone, extract the coordinate figure of coboundary and lower boundary according to lax principle.

Described characteristic extracting module 21, concrete for using the method for estimating based on background area, the image segmentation in the lorry brake shoe characteristic area of present image is obtained lorry brake shoe zone binary picture; Use the element marking method, from the binary picture of described lorry brake shoe zone, extract largest connected zone; Use the Canny operator, extract lorry brake shoe edge contour in the binary picture from largest connected zone; Extract the feature of lorry brake shoe according to described lorry brake shoe edge contour.

Here, the method that described background area is estimated is prior art, does not do here and gives unnecessary details; The element marking method is prior art, does not do here and gives unnecessary details; The Canny operator is prior art, does not do here and gives unnecessary details.

Described characteristic extracting module 21, the concrete central point that is used for calculating lorry brake shoe edge contour is expressed as polar coordinates with lorry brake shoe edge contour centered by described central point; The described lorry brake shoe edge contour that is expressed as polar form is carried out discretize calculate, obtain the lorry brake shoe edge contour of discretize; After the lorry brake shoe edge contour of described discretize carried out normalization and calculate, sort according to the rule that presets, again the result who obtains after the ordering is carried out wavelet decomposition and obtain wavelet coefficient, use wavelet coefficient to calculate the smooth features value;

Wherein, the described central point that calculates lorry brake shoe edge contour can comprise: suppose that lorry brake shoe edge contour is x, then take relative arc length s as parameter, then lorry brake shoe edge contour can be expressed as:

0≤s≤1 wherein, Be lorry brake shoe edge contour; Then the central point computing formula of lorry brake shoe edge contour is $\stackrel{\&RightArrow;}{x_0}={\&Integral;}_0^1\stackrel{\&RightArrow;}{x\left(s\right)}\mathrm{ds}=({\&Integral;}_0^{1}x\left(s\right)\mathrm{ds},{\&Integral;}_0^{1}y\left(s\right)\mathrm{ds}),$ 0≤s≤1 wherein.

Describedly centered by central point, lorry brake shoe edge contour is expressed as polar coordinates and can be expressed as:

$\left\{\begin{array}{c}\mathrm{\&theta;}\left(s\right)=\&angle;(\stackrel{\&RightArrow;}{x}\left(s\right)-\stackrel{\&RightArrow;}{x_0})\\ r\left(s\right)={\left|\right|\stackrel{\&RightArrow;}{x}\left(s\right)-\stackrel{\&RightArrow;}{x_0}\left|\right|}_2\\ \stackrel{\&RightArrow;}{z}\left(t\right)=(\mathrm{\&theta;}\left(s\right),\left(r\left(s\right)\right).\end{array}\right.,$ Wherein, ∠ represents the polar angle of vector under the polar coordinate system.

Describedly the described lorry brake shoe edge contour that is expressed as polar form is carried out discretize calculate and to be the polar coordinates (r (s that obtains each discrete point on the profile _i), θ (s _i)), wherein I=0,1 ... N-1, N are the discrete point number.

Described normalization is calculated can use formula:

The rule that described basis presets sorts and can comprise: note θ ' (s _i) be θ ' (i), r ' (s _i) be r ' (i), r ' (i) is resequenced: R (i)=r ' (η (i)) so that right

θ (η (i))＜θ (η (j)) is arranged;

Describedly the result that obtains after the ordering is carried out wavelet decomposition obtain wavelet coefficient and can comprise: $d\left(i\right)=r^{\&prime;}\left(\mathrm{\&eta;}\left(i\right)\right)\&CircleTimes;\mathrm{\&psi;}\left(i\right);$

Described use wavelet coefficient calculates the smooth features value can be E=||d (i) || ₂

The formula that the calculating of the convex and concave feature value of described lorry brake shoe edge contour is used is:

Wherein, c (n), n=1, the sequence of points of 2N lorry brake shoe edge contour, c ' (n)=|| c (n)-c ₀||, n=1,2,, N, $c_0=\frac{1}{N}\underset{n=1}{\overset{N}{\mathrm{\&Sigma;}}}c\left(n\right);$

The rectangular degree computing formula of described lorry brake shoe edge contour can for:

A wherein ₀The area of profile, A _MERIt is the area of the minimum boundary rectangle of profile.

Described lorry brake shoe edge contour admittedly by the property computing formula can for:

Wherein, A represents the area of profile, and CA is the area of its minimal convex polygon.

The compactness computing formula of described lorry brake shoe edge contour can for:

Wherein, P represents the girth of profile, and A represents area.

The circle computing formula of described lorry brake shoe edge contour can for:

Wherein,

μ _RIt is the mean distance from the regional barycenter to the frontier point.

δ _RBe from the regional barycenter to the frontier point apart from mean square deviation.

The length breadth ratio of described lorry brake shoe edge contour is the ratio of wide and length of the minimum boundary rectangle of lorry brake shoe edge contour;

The area of described lorry brake shoe edge contour can be added up acquisition for the number of pixels in: the lorry brake shoe edge contour scope.

The girth of described lorry brake shoe edge contour can be the boundary length of lorry brake shoe edge contour.

Described identification module 22, the concrete preservation Fault Identification value that is used for.

Described identification module 22 equals the Fault Identification value if specifically be used for the eigenwert of present image, determines that then the lorry brake shoe breaks down; Otherwise, determine that the lorry brake shoe does not break down.

Described identification module 22 also is used for using the lorry brake shoe image that does not have fault and existing the lorry brake shoe image of fault to set up respectively the positive and negative samples training set; Extract respectively all Characteristic of Images in the positive and negative samples training set, form nine dimensional feature vectors corresponding to positive and negative samples training set, value corresponding to positive and negative samples of using the SVM algorithm to calculate, the value that the negative sample training set is corresponding is as the Fault Identification value.

Said apparatus can be used as in the management equipment that logic module is installed on TDFS.

The method that the application of the invention provides and device, 75 of lorry brake shoe fault sample images are chosen in test, 75 of non-fault sample images are set up sample set and are carried out the sorter training, adaptability for the test recognizer, (cross validation is called again the intersection contrast to the thought of employing cross validation, at first with training sample v five equilibrium, a copy of it is as test set, and v-1 part is as training set in addition.Rotate successively, until every increment has originally all been made test set one time, namely carried out the process of v training and prediction.Therefore, the accuracy rate of cross validation is v time mean value) carry out Fault Identification test, do altogether 10 experiments, the mean value of getting experimental result be final recognition result as shown in Figure 3.As seen, use the recognition methods and device of lorry brake shoe fault provided by the invention after, recognition time is lower than two seconds, obviously will be less than manually and distinguish by image, and loss, false drop rate is very low.

The above is preferred embodiment of the present invention only, is not for limiting protection scope of the present invention.

Claims (12)

1. the recognition methods of a lorry brake shoe fault is characterized in that the method comprises:

From present image, extract the feature of cutting apart of three angles;

The lorry brake shoe characteristic area that feature is determined present image of cutting apart according to described three angles;

From the lorry brake shoe characteristic area of present image, extract the feature of lorry brake shoe;

Use the support vector machines algorithm that the feature calculation of lorry brake shoe is drawn the eigenwert of present image, judge according to described eigenwert and the Fault Identification value that presets whether the lorry brake shoe exists fault.

2. method according to claim 1, it is characterized in that, the described feature of cutting apart of from present image, extracting three angles, comprising: periodically extract present image from fault picture dynamic detection system TDFS, the Gray Projection that present image is carried out three angles obtains three drop shadow curves; All drop shadow curves are carried out filtering, with the maximal value in filtered each drop shadow curve all as the feature of cutting apart of three angles of present image.

3. method according to claim 1, it is characterized in that, the described lorry brake shoe characteristic area that feature is determined present image of cutting apart according to three angles, comprise: the marginal information of using Canny operator extraction present image, be the feature of cutting apart of zero degree according to angle, determine the left margin of lorry brake shoe characteristic area and the coordinate figure of right margin; According to the cut apart feature of angle for-25 ° and 25 °, determine the coboundary of lorry brake shoe characteristic area and the coordinate figure of lower boundary.

4. method according to claim 1, it is characterized in that, the described feature of from the lorry brake shoe characteristic area of present image, extracting the lorry brake shoe, comprise: use the method for estimating based on background area, the image segmentation in the lorry brake shoe characteristic area of present image is obtained lorry brake shoe zone binary picture; Use the element marking method, from the binary picture of described lorry brake shoe zone, extract largest connected zone; Use the Canny operator, extract lorry brake shoe edge contour in the binary picture from largest connected zone; Extract the feature of lorry brake shoe according to described lorry brake shoe edge contour.

5. method according to claim 4, it is characterized in that the feature of described lorry brake shoe comprises: the convex and concave feature value of the smooth features value of lorry brake shoe edge contour, lorry brake shoe edge contour, the jaggedness of lorry brake shoe edge contour, lorry brake shoe edge contour solid by the compactness of value, lorry brake shoe edge contour, the circle of lorry brake shoe edge contour, the length breadth ratio of lorry brake shoe edge contour, the area of lorry brake shoe edge contour and the girth of lorry brake shoe edge contour.

6. method according to claim 1, it is characterized in that, described from present image, the extraction the cutting apart before the feature of three angles, the method also comprises: use the lorry brake shoe image that does not have fault and exist the lorry brake shoe image of fault to set up respectively the positive and negative samples training set; Extract respectively all Characteristic of Images in the positive and negative samples training set, form nine dimensional feature vectors corresponding to positive and negative samples training set, value corresponding to positive and negative samples of using the SVM algorithm to calculate, the value that the negative sample training set is corresponding is as the Fault Identification value.

7. the recognition device of a lorry brake shoe fault is characterized in that, this device comprises: characteristic extracting module and identification module; Wherein,

Characteristic extracting module, be used for extracting from present image the feature of cutting apart of three angles, the lorry brake shoe characteristic area that feature is determined present image of cutting apart according to described three angles, from the lorry brake shoe characteristic area of present image, extract the feature of lorry brake shoe, the feature of described lorry brake shoe is sent to identification module;

Identification module be used for to use the feature calculation of the described lorry brake shoe that the SVM algorithm sends characteristic extracting module to draw the eigenwert of present image, judges according to described eigenwert and the Fault Identification value that presets whether the lorry brake shoe exists fault.

8. device according to claim 7 is characterized in that, described characteristic extracting module, and concrete being used for periodically extracted present image from the TDFS at place, and the Gray Projection that present image is carried out three angles obtains three drop shadow curves; All drop shadow curves are carried out filtering, with the maximal value in filtered each drop shadow curve all as the feature of cutting apart of three angles of present image.

9. device according to claim 7, it is characterized in that described characteristic extracting module, the concrete marginal information that is used for using Canny operator extraction present image, be the feature of cutting apart of zero degree according to angle, determine the left margin of lorry brake shoe characteristic area and the coordinate figure of right margin; According to the cut apart feature of angle for-25 ° and 25 °, determine the coboundary of lorry brake shoe characteristic area and the coordinate figure of lower boundary.

10. device according to claim 7 is characterized in that, described characteristic extracting module is concrete for using the method for estimating based on background area, and the image segmentation in the lorry brake shoe characteristic area of present image is obtained lorry brake shoe zone binary picture; Use the element marking method, from the binary picture of described lorry brake shoe zone, extract largest connected zone; Use the Canny operator, extract lorry brake shoe edge contour in the binary picture from largest connected zone; Extract the feature of lorry brake shoe according to described lorry brake shoe edge contour.

11. device according to claim 7, it is characterized in that, described characteristic extracting module, the girth of the length breadth ratio of the solid compactness by value, lorry brake shoe edge contour of the convex and concave feature value of concrete smooth features value for extraction lorry brake shoe edge contour, lorry brake shoe edge contour, the jaggedness of lorry brake shoe edge contour, lorry brake shoe edge contour, the circle of lorry brake shoe edge contour, lorry brake shoe edge contour, the area of lorry brake shoe edge contour and lorry brake shoe edge contour is as the feature of lorry brake shoe.

12. device according to claim 7 is characterized in that,

Described identification module also is used for using the lorry brake shoe image that does not have fault and existing the lorry brake shoe image of fault to set up respectively the positive and negative samples training set; Extract respectively all Characteristic of Images in the positive and negative samples training set, form nine dimensional feature vectors corresponding to positive and negative samples training set, value corresponding to positive and negative samples of using the SVM algorithm to calculate, the value that the negative sample training set is corresponding is as the Fault Identification value.

Images