CN102968622A - station caption identification method and device - Google Patents

Abstract

The invention discloses a station caption identification method. The method comprises the following steps of: extracting shape features expressed by first two invariant moments in the shape features of a station caption, wherein the shape features of the station caption are expressed based on seven invariant moments; extracting four discrete shape features of the station caption, wherein the four discrete shape features of the station caption are described on the basis of gaussian descriptors; and indentifying the station caption according to the shape features expressed by the first two invariant moments and the four discrete shape features described on the basis of the gaussian descriptors and based on a station caption knowledge base. Correspondingly, the invention also discloses a station caption identification device. By using the method and the device, the station caption can be simply and efficiently identified.

Description

A kind of TV station symbol recognition method and TV station symbol recognition device

Technical field

The present invention relates to the computer image processing technology field, more specifically, relate to a kind of TV station symbol recognition method.

Background technology

Station symbol, namely the sign of a TV station is to be different from other channels, embodies the external manifestation symbol of doing platform theory and image.The basic function that station symbol has identification, distinguishes and beautifies, it also is one of important semantic source of realizing video analysis, understanding, retrieval.Along with the develop rapidly of TV tech, the program of each province, city, Television Station of County Level reaches covers up to a hundred.Usually, in case there is illegal signals to invade, TV station's station symbol will disappear, so real-time station symbol detection and Identification technology is the key that the monitoring illegal signals is invaded, has very important practical value.

Usually after detecting station symbol, carry out feature extraction and identification to station symbol.Having in the station symbol feature extraction major part now all is to adopt to carry shape, color, textural characteristics, and the feature of extraction is counted often; And in extracting shape facility, mostly be to adopt to extract not 7 features of bending moment of Hu.The method of this station symbol Shape Feature Extraction, calculated amount is large, consuming time; And because the feature of extracting is counted often, relating to for later sorter has also increased complexity.

Summary of the invention

For the problem in the correlation technique, the present invention proposes a kind of TV station symbol recognition method and apparatus, it can identify the station symbol that extracts fast.

According to an aspect of the present invention, provide a kind of TV station symbol recognition method, described method comprises:

Extraction based on 7 not bending moment represent front 2 shape facilities of representing of bending moment not in the shape facility of described station symbol;

Extraction is based on 4 discrete shape facilities describing described station symbol of Gaussian descriptors;

According to described front 2 shape facilities of representing of bending moment and identify described station symbol based on the station symbol knowledge base based on described 4 discrete shape facilities of Gaussian descriptors not.

In optional embodiment, described method also comprises the color characteristic that extracts described station symbol, then according to according to described front 2 shape facilities of representing of bending moment and further comprise based on the step that the station symbol knowledge base identifies described station symbol based on described 4 discrete shape facilities of Gaussian descriptors not: according to according to described front 2 not the shape facility that represents of bending moment, identify described station symbol based on described 4 discrete shape facilities of Gaussian descriptors and the color characteristic of described station symbol based on the station symbol knowledge base.

In optional embodiment, the color characteristic that extracts described station symbol comprises: the RGB color space conversion of described station symbol is become the hsv color space; The color characteristic that comprises saturation degree and brightness that is used for describing described station symbol in described hsv color spatial extraction.

In optional embodiment, before the step of extracting for the shape facility of describing described station symbol, also comprise: receive the signal that comprises the station symbol image information, from described signal, extract station symbol.

In optional embodiment, according to described front 2 not the shape facility that represents of bending moment and identify the step of described station symbol based on described 4 discrete shape facilities of Gaussian descriptors based on the station symbol knowledge base before, also comprise: the station symbol gray processing after will extracting.

According to another aspect of the present invention, also provide a kind of TV station symbol recognition device, described device comprises:

The Shape Feature Extraction module, be used for to extract based on 7 not bending moment represent front 2 shape facilities of representing of bending moment not of the shape facility of described station symbol; And extraction is based on 4 discrete shape facilities describing described station symbol of Gaussian descriptors;

The TV station symbol recognition module is according to described front 2 shape facilities of representing of bending moment and identify described station symbol based on the station symbol knowledge base based on described 4 discrete shape facilities of Gaussian descriptors not.

In optional embodiment, described device also comprises the color characteristic extraction module, is used for extracting the color characteristic of described station symbol; Described TV station symbol recognition module also is used for according to based on described front 2 shape facilities of bending moment not, identify described station symbol based on described 4 discrete shape facilities of Gaussian descriptors and the color characteristic of described station symbol based on the station symbol knowledge base.

In optional embodiment, the color characteristic that described color characteristic extraction module extracts described station symbol comprises: the RGB color space conversion of described station symbol is become the hsv color space; And the color characteristic that comprises saturation degree and brightness that is used for describing described station symbol in described hsv color spatial extraction.

In optional embodiment, described device also comprises the station symbol extraction module, is used for receiving the signal that comprises the station symbol image information, extracts station symbol from described signal.

In optional embodiment, described device comprises: the station symbol knowledge base that stores a plurality of station symbol SVM that train.

The present invention is by extracting calculated amount and the time that can greatly dwindle the station symbol feature extraction based on front 2 shape facilities in 7 shape facilities of bending moment not, can guarantee that at 4 shape facilities that extract based on the station symbol of Gaussian descriptors front 2 moment characteristics that extract have more robustness.Adopt TV station symbol recognition method of the present invention, the TV station symbol recognition rate also can further improve.

Description of drawings

Fig. 1 is the process flow diagram according to a kind of TV station symbol recognition method of the embodiment of the invention.

Fig. 2 is for the synoptic diagram of explaining the Gaussian descriptors principle.

Fig. 3 is the block diagram according to a kind of TV station symbol recognition device of the embodiment of the invention.

Embodiment

Below in conjunction with accompanying drawing the embodiment of the invention is described in detail.

Fig. 1 is the process flow diagram according to a kind of TV station symbol recognition method of the embodiment of the invention.As described in Figure 1, the method comprises S101, extract based on 7 not bending moment represent front 2 shape facilities of representing of bending moment not in the shape facility of described station symbol.

Wherein, step S101 comprises substep S101-1: the station symbol gray processing after will extracting.The background of the station symbol behind the gray processing is black.Step S101 also comprises substep S101-2: the station symbol behind the gray processing is extracted normalized not bending moment, wherein only extract front 2 eigenwerts in 7 eigenwerts of bending moment not.

The moment characteristics main geometric properties of image-region, be called again geometric moment.Because it has the invariant features of the characteristics such as rotation, translation, yardstick, so be called again not bending moment.In image was processed, geometric invariant moment can be used as an important feature and represents object, and feature is come the image operation such as classify accordingly.Geometric moment is to be proposed in 1962 by Hu (Visualpattern recognition by momentinvariants, IRE TRANSA CTIONS ON INFORMA TION THEORY):

Be provided with consecutive image two-dimensional function f (x, y), its (p+q) rank square is defined as follows:

$m_{\mathrm{pq}}=\underset{x}{\mathrm{\Σ}}\underset{y}{\mathrm{\Σ}}{x}^p{y}^{q}f(x,y)$ p，q＝0，1，2...(1)

Above-mentioned moment characteristics amount is carried out the center square that place normalization obtains image f (x, y) is

${\mathrm{\μ}}_{\mathrm{pq}}=\underset{x}{\mathrm{\Σ}}\underset{y}{\mathrm{\Σ}}{(x-\stackrel{\‾}{x})}^p{(y-\stackrel{\‾}{y})}^{q}f(x,y)$ p，q＝0，1，2...(2)

Wherein

Be the barycenter of image,

Again the center square is carried out size normalization, definition normalization center square is:

r＝(p+q)/2+1p+q＝2，3，4...(3)

Utilize second order and third central moment to construct 7 square invariants with translation, convergent-divergent and invariable rotary shape, expression is as follows:

$\left\{\begin{array}{c}\\ {\mathrm{\φ}}_1={\mathrm{\η}}_{20}+{\mathrm{\η}}_{02}\\ {\mathrm{\φ}}_2={({\mathrm{\η}}_{20}-{\mathrm{\η}}_{02})}^2+4{\mathrm{\η}}_{11}^2\\ {\mathrm{\φ}}_3={({\mathrm{\η}}_{30}-3{\mathrm{\η}}_{12})}^2+{(3{\mathrm{\η}}_{21}-{\mathrm{\η}}_{03})}^2\\ {\mathrm{\φ}}_4={({\mathrm{\η}}_{30}+{\mathrm{\η}}_{12})}^2+{({\mathrm{\η}}_{21}+{\mathrm{\η}}_{03})}^2\\ {\mathrm{\φ}}_5=({\mathrm{\η}}_{30}-3{\mathrm{\η}}_{12})({\mathrm{\η}}_{30}+{\mathrm{\η}}_{12})[{({\mathrm{\η}}_{30}+{\mathrm{\η}}_{12})}^2-3{({\mathrm{\η}}_{21}+{\mathrm{\η}}_{03})}^2]+\\ (3{\mathrm{\η}}_{21}-{\mathrm{\η}}_{03})({\mathrm{\η}}_{21}+{\mathrm{\η}}_{03})[3{({\mathrm{\η}}_{30}+{\mathrm{\η}}_{12})}^2-{({\mathrm{\η}}_{21}+{\mathrm{\η}}_{03})}^2]\\ {\mathrm{\φ}}_6=({\mathrm{\η}}_{20}-{\mathrm{\η}}_{02})[{({\mathrm{\η}}_{30}+{\mathrm{\η}}_{12})}^2-{({\mathrm{\η}}_{12}-{\mathrm{\η}}_{03})}^2]+4{\mathrm{\η}}_{11}({\mathrm{\η}}_{30}+{\mathrm{\η}}_{12})({\mathrm{\η}}_{21}+{\mathrm{\η}}_{03})\\ {\mathrm{\φ}}_7=(3{\mathrm{\η}}_{21}-{\mathrm{\η}}_{03})({\mathrm{\η}}_{30}+{\mathrm{\η}}_{21})[{({\mathrm{\η}}_{30}+{\mathrm{\η}}_{12})}^2-3{({\mathrm{\η}}_{21}+{\mathrm{\η}}_{03})}^2]+\\ (3{\mathrm{\η}}_{21}-{\mathrm{\η}}_{30})({\mathrm{\η}}_{21}+{\mathrm{\η}}_{03})[3{({\mathrm{\η}}_{30}+{\mathrm{\η}}_{12})}^2-{({\mathrm{\η}}_{21}+{\mathrm{\η}}_{03})}^2]\end{array}\right.---\left(4\right)$

Moment characteristics has obvious physical significance, the null matrix m of target ₀₀Therefore the area that has reflected target, first moment have reflected the centroid position of target, utilize these two square amounts just can avoid because of the impact on characteristics of image of article size and change in displacement.

Because 7 of same image bending moment φ not _k(k=1,2 ..., 7) the order of magnitude differ greatly, for the ease of the identification of successive image, can utilize the method for taking from right logarithm to carry out data area compression, consider that simultaneously the situation of negative value might appear in bending moment not, therefore the actual not bending moment that adopts is:

Φ _k＝|ln|φ _k|| k＝1，2，...，7 (5)

In the embodiment of the invention, extract φ based on formula (5) ₁And φ ₂The shape facility that is used for the expression station symbol.

S102 extracts 4 the discrete shape facilities describing described station symbol based on Gaussian descriptors.

Wherein, step S102 comprises substep S102-1: the edge that extracts station symbol.Edge extracting is the common technology of image processing field.For example carry out edge extracting with wavelet transformation.Step S102 also comprises substep S102-2: the total arc length l of computer board target _TotalAnd substep S102-3 calculates based on 4 discrete values of Gaussian descriptors proper vector with composing images.

The below introduces the principle of Gaussian descriptors and describes the enforcement of step S102 in detail.

If curve C is by curve C ₁, C ₂..., C _n(n≤1) forms.By the relevant knowledge of infinitesimal analysis as can be known, the center-of-mass coordinate of curve C can be expressed as with the 1st type curvilinear integral

$\stackrel{\‾}{x}=\frac{{\∫}_c\mathrm{xds}}{{\∫}_c\mathrm{ds}},$ $\stackrel{\‾}{y}=\frac{{\∫}_c\mathrm{yds}}{{\∫}_c\mathrm{ds}}---\left(6\right)$

The degree of scatter of having a few on the curve C can be used variances sigma ²Describe:

${\mathrm{\σ}}^2=\frac{{\∫}_c[{(x-\stackrel{\‾}{x})}^2+{(y-\stackrel{\‾}{y})}^2]\mathrm{ds}}{{\∫}_c\mathrm{ds}}---\left(7\right)$

With r represent to have a few on the curve C and barycenter between the mean value of distance:

$r=\frac{{\∫}_c\sqrt{{(x-\stackrel{\‾}{x})}^2+{(y-\stackrel{\‾}{y})}^2}\mathrm{ds}}{{\∫}_c\mathrm{ds}}---\left(8\right)$

Title with Be the center of circle, the circle take r as radius is the reference circle of curve C.

To λ ∈ arbitrarily (0 ,+∞), the note curve C ^(λ)=(x, y) | (x, y) ∈ C, and ${(x-\stackrel{\‾}{x})}^2+{(y-\stackrel{\‾}{y})}^2\≤{\left(\mathrm{\λr}\right)}^2\}.$

To arbitrarily θ ∈ [0,2 π] and λ ∈ (0 ,+∞), defined function

$f(\mathrm{\θ},\mathrm{\λ})=\frac{{\∫}_{c^{\left(\mathrm{\λ}\right)}}{e}^{-\frac{{(x-\stackrel{\‾}{x}-\mathrm{\λ}r\cos \mathrm{\θ})}^2+{(y-\stackrel{\‾}{y}-\mathrm{\λ}r\sin \mathrm{\θ})}^2}{2{\mathrm{\σ}}^2}\mathrm{ds}}}{{\∫}_c\mathrm{ds}}---\left(9\right)$

The meaning of function f (θ, λ) as shown in Figure 2.

Among Fig. 1, the part edge curve is positioned at round inner arc AmB and arc CnD, i.e. curve C ^(λ)=AmB+CnD.To C ^(λ)On any point Q, establish its coordinate for (x, y), what then put distance between P and the Q square is:

${\left|\mathrm{PQ}\right|}^2={(x-\stackrel{\‾}{x}-\mathrm{\λ}r\cos \mathrm{\θ})}^2+{(y-\stackrel{\‾}{y}-\mathrm{\λ}r\sin \mathrm{\θ})}^2$

" potential function " on f (θ, λ) the analog Neo-Confucianism just replaces simple inverse proportion function with 2 dimension Gaussian functions here, and the point (particle in the analog Neo-Confucianism, point charge etc.) on the circle outward flange curve is masked.F (θ, λ) is C ^(λ)The summation of upper all-pair point P " effect ".∫ in the denominator _cThe effect of ds is to carry out the yardstick standardization.Usually function f (θ, λ) is called the Gauss potential function.For given θ and λ, f (θ, λ) is constant about translation and yardstick.

Defined function

$l\left(\mathrm{\λ}\right)=\frac{{\∫}_0^{2\mathrm{\π}}f(\mathrm{\θ},\mathrm{\λ})\mathrm{d\θ}}{2\mathrm{\π}}---\left(10\right)$

As λ fixedly the time, Gauss potential function f (θ, λ) is the continuous function of θ.I (λ) can be interpreted as intuitively the mean value of f (θ, λ) on interval [0,2 π].Can say that also I (λ) is the mean value of " gesture " of each point on the circumference.So I (λ) has portrayed edge image and has been positioned at Be the center of circle, the shape facility of part in the circle take λ r as radius.I (λ) is called the Gaussian descriptors of curve C.

For the edge of station symbol image, with formula (6) to formula (10) discretize.For this reason, provide first total arc length ∫ of calculated curve C _cDs (is designated as l _Total) the discretize formula.Utilize the definition of the 1st type curvilinear integral, can obtain:

$l_{\mathrm{total}}=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\mathrm{\Δ}{s}_i---\left(11\right)$

Formula (6) is corresponding following various respectively to the discretize result of formula (9): $\stackrel{\‾}{x}=\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{x}_i\mathrm{\Δ}{s}_i}{l_{\mathrm{total}}},$ $\stackrel{\‾}{y}=\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{y}_i\mathrm{\Δ}{s}_i}{l_{\mathrm{total}}},$ (x _i，y _i)∈c (12) ${\mathrm{\σ}}^2=\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}[{(x_i-\stackrel{\‾}{x})}^2+{(y_i-\stackrel{\‾}{y})}^2]\mathrm{\Δ}{s}_i}{l_{\mathrm{total}}},$ (x _i，y _i)∈c

(13) $r=\frac{\underset{i=1}{\overset{N}{\mathrm{\Σ}}}\sqrt{{(x_i-\stackrel{\‾}{x})}^2+{(y_i-\stackrel{\‾}{y})}^2}\mathrm{\Δ}{s}_i}{l_{\mathrm{total}}},$ (x _i，y _i)∈c

(14) $f(\mathrm{\θ},\mathrm{\λ})=\frac{\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{e}^{-\frac{{(x_i-\stackrel{\‾}{x}-\mathrm{\λ}r\cos \mathrm{\θ})}^2+{(y_i-\stackrel{\‾}{y}-\mathrm{\λ}r\sin \mathrm{\θ})}^2}{2{\mathrm{\σ}}^2}\mathrm{\Δ}{s}_i}}{l_{\mathrm{total}}},$ (x _i，y _i)∈C ^(λ)

(15)

In above formula, Δ s _i" distance " between two neighbor pixels on the contour curve: if two neighbor pixels are that horizontal direction or vertical direction are arranged, Δ s so _i=1; If two neighbor pixels are 45 ° or 135 ° of directions arrangements, so

Interval [0,2 π] divided equally part for K, and then formula (5) can obtain according to the correlation theory of definite integral:

$I\left(\mathrm{\λ}\right)\≈\frac{\frac{2\mathrm{\π}}{K}\underset{i=1}{\overset{K}{\mathrm{\Σ}}}f(\frac{2\mathrm{\π}}{K}i,\mathrm{\λ})}{2\mathrm{\π}}=\frac{\underset{i=1}{\overset{K}{\mathrm{\Σ}}}(\frac{2\mathrm{\π}}{K}i,\mathrm{\λ})}{K}---\left(16\right)$

The result shows, can select the K value between 3≤K≤8 according to actual conditions, and K is integer.In embodiments of the present invention, the K value gets 8.

Portray the shape facility of object in order to obtain enough information, thereby improve identification/recall ratio, get 4 different λ values, for example: λ ₁=0.25r, λ ₃=0.75r, λ ₅=1.25r, λ ₇=1.75r.Like this, correspondingly can obtain 4 discrete values of Gaussian descriptors.Then consist of based on 4 shape facility vectors of Gaussian descriptors with the identification that is used for station symbol/retrieval with these 4 discrete values.

Step 103 is according to according to described front 2 shape facilities of representing of bending moment and identify station symbol based on the station symbol knowledge base based on described 4 discrete shape facilities of Gaussian descriptors not.

In one embodiment, adopt SVM technology (support vector machine) to identify station symbol.SVM is on the Statistical Learning Theory basis, and development a kind of new learning machine structure out is theoretical.In order to solve this multiclass problem of TV station symbol recognition, adopt equally the one-to-many strategy, namely for SVM of each television channel training, make each SVM can distinguish each channel station symbol.

From the station symbol Sample Storehouse, take out all samples of the first channel, it is designated as class I, all samples of other all channels are designated as class II, to all samples of classification have been marked as the input sample, train a SVM, obtain corresponding support vector and corresponding (broad sense) optimal classification face, this SVM sequence number is designated as A, show that A SVM is used for distinguishing first channel and remaining channel.

From the station symbol Sample Storehouse, take out all samples of second channel, it is designated as class I, all sample points of other all channels are designated as class II, following step is the same with the operation of previous step, at last the SVM sequence number that trains is designated as B, this SVM is used for distinguishing second channel and other all channels.

Repeat above step, until all channels in the traversal Sample Storehouse.Obtained at last m SVM.

Set up the station symbol knowledge base, it can comprise the station symbol SVM that trains, the knowledge such as TV station's title, background information and program orientation.

If m channel only arranged in the Sample Storehouse to be known, and each channel has the SVM that has trained of a correspondence.When new sample to be known of input, when simultaneously this sample to be known being classified by m SVM: if i SVM assigns to this sample among the class I, remaining SVM is classified among the class II, judges that then this sample belongs to i channel in the Sample Storehouse; If have simultaneously a plurality of SVM that this sample is assigned among the class I, and remaining SVM assigns to it among class II, then be judged as wrong minute; If all SVM assign to this sample among the class II, then also be judged as wrong minute.So in case a certain sample is identified, just corresponding related content is carried out semantic tagger to it in the available station symbol knowledge base.

In the TV station symbol recognition method of the optional embodiment of the present invention, also comprise: the color characteristic that extracts station symbol.Then, according to based on described front 2 shape facilities of bending moment not, identify described station symbol based on described 4 discrete shape facilities of Gaussian descriptors and the color characteristic of described station symbol based on the station symbol knowledge base.

Wherein, the color characteristic that extracts described station symbol comprises: the RGB color space conversion of described station symbol is become the hsv color space.The hsv color space is a kind of mode of describing the color characteristic of station symbol.HSV is expressed as three kinds of attributes to colour signal: tone H, saturation degree S and brightness V.

Then, be used for describing the color characteristic that comprises saturation degree and brightness of described station symbol in described hsv color spatial extraction.

Fig. 3 is the block diagram of a kind of TV station symbol recognition device of providing of the embodiment of the invention.Device 10 comprises: Shape Feature Extraction module 110 is used for extracting front 2 shape facilities of 7 shape facilities that are used for describing described station symbol based on bending moment not; And extraction is based on 4 discrete shape facilities that are used for describing described station symbol of Gaussian descriptors; And TV station symbol recognition module 130, according to identifying described station symbol based on described front 2 shape facilities of bending moment not with based on described 4 discrete shape facilities of Gaussian descriptors based on station symbol knowledge base 140.

In optional embodiment, described device 10 also comprises color characteristic extraction module 120, is used for extracting the color characteristic of described station symbol.TV station symbol recognition module 130 also is used for according to based on described front 2 shape facilities of bending moment not, identify described station symbol based on described 4 discrete shape facilities of Gaussian descriptors and the color characteristic of described station symbol based on station symbol knowledge base 140.

The color characteristic that described color characteristic extraction module 120 extracts described station symbol comprises: the RGB color space conversion of described station symbol is become the hsv color space; And the color characteristic that comprises saturation degree and brightness that is used for describing described station symbol in described hsv color spatial extraction.

In optional embodiment, described device also comprises the station symbol extraction module, is used for receiving the signal that comprises the station symbol image information, extracts station symbol from described signal.

In optional embodiment, station symbol knowledge base 140 stores a plurality of station symbol SVM that train.

Traditional based on Hu not the Shape Feature Extraction of bending moment entirely be based on 7 not bending moments, through a large amount of experimental verifications, 7 constant Hu squares all participate in the extraction of feature, and the error that sometimes produces is larger, and time-consuming, when carrying out the identification of back, sometimes also can produce misclassification.For front 2 the Hu moment characteristics that guarantee to extract have more robustness, the present invention has introduced again Gaussian descriptors, utilize Gaussian descriptors to extract 4 shape facilities of station symbol, it is little that Gaussian descriptors has calculated amount, to edge change and the insensitive for noise characteristics of appropriateness.In order to embody the strikingly color feature of station symbol, the present invention extracts 2 color characteristics in addition.

By the present invention, can dwindle widely calculated amount and the time of station symbol feature extraction, and discrimination also there is further raising.

The above only is preferred embodiment of the present invention, and is in order to limit the present invention, within the spirit and principles in the present invention not all, any modification of doing, is equal to replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (10)

1. TV station symbol recognition method, described method comprises:

Extraction based on 7 not bending moment represent front 2 shape facilities of representing of bending moment not in the shape facility of described station symbol;

Extraction is based on 4 discrete shape facilities describing described station symbol of Gaussian descriptors;

According to described front 2 shape facilities of representing of bending moment and identify described station symbol based on the station symbol knowledge base based on described 4 discrete shape facilities of Gaussian descriptors not.

2. TV station symbol recognition method according to claim 1 is characterized in that, described method also comprises:

Extract the color characteristic of described station symbol;

According to described front 2 shape facilities of representing of bending moment and further comprise based on the step that the station symbol knowledge base identifies described station symbol based on described 4 discrete shape facilities of Gaussian descriptors not:

According to described front 2 not the shape facility that represents of bending moment, identify described station symbol based on described 4 discrete shape facilities of Gaussian descriptors and the color characteristic of described station symbol based on the station symbol knowledge base.

3. TV station symbol recognition method according to claim 2 is characterized in that, the color characteristic that extracts described station symbol comprises:

The RGB color space conversion of described station symbol is become the hsv color space;

The color characteristic that comprises saturation degree and brightness that is used for describing described station symbol in described hsv color spatial extraction.

4. TV station symbol recognition method according to claim 1 is characterized in that, before the step of extracting for the shape facility of describing described station symbol, also comprises: receive the signal that comprises the station symbol image information, extract station symbol from described signal.

5. TV station symbol recognition method according to claim 4 is characterized in that, extract based on 7 not bending moment represent also to comprise in the shape facility of described station symbol front 2 not before the step of the shape facility that represents of bending moment:

With the station symbol gray processing after extracting.

6. TV station symbol recognition device, described device comprises:

The Shape Feature Extraction module, be used for to extract based on 7 not bending moment represent front 2 shape facilities of representing of bending moment not of the shape facility of described station symbol; And extraction is based on 4 discrete shape facilities describing described station symbol of Gaussian descriptors;

The TV station symbol recognition module is according to described front 2 shape facilities of representing of bending moment and identify described station symbol based on the station symbol knowledge base based on described 4 discrete shape facilities of Gaussian descriptors not.

7. TV station symbol recognition device according to claim 6 is characterized in that, described device also comprises the color characteristic extraction module, is used for extracting the color characteristic of described station symbol;

Described TV station symbol recognition module, also be used for according to described front 2 not the shape facility that represents of bending moment, identify described station symbol based on described 4 discrete shape facilities of Gaussian descriptors and the color characteristic of described station symbol based on the station symbol knowledge base.

8. TV station symbol recognition device according to claim 7 is characterized in that, the color characteristic that described color characteristic extraction module extracts described station symbol comprises:

The RGB color space conversion of described station symbol is become the hsv color space;

The color characteristic that comprises saturation degree and brightness that is used for describing described station symbol in described hsv color spatial extraction.

9. TV station symbol recognition device according to claim 6 is characterized in that, described device also comprises the station symbol extraction module, is used for receiving the signal that comprises the station symbol image information, extracts station symbol from described signal.

10. TV station symbol recognition device according to claim 6 is characterized in that, described device also comprises the station symbol knowledge base that stores a plurality of station symbol SVM that train.

Images