CN104751171A - Method of classifying Naive Bayes scanned certificate images based on feature weighting - Google Patents

Abstract

The invention discloses a naive Bayesian scanning certificate image classification method based on feature weighting, which extracts the color features of the HSV space of the circular stamp area by using the Hough transform to perform circular seal positioning, segmentation and size adjustment on the preprocessed certificate image vector and image aspect ratio; establish a certificate image database, process each certificate image in the database according to the above steps, and obtain the round stamp HSV color feature vector and image aspect ratio of each scanned certificate image in the database, according to the obtained The eigenvectors calculate the probability of different data combinations in the certificate image database, and save the data after weighting processing; calculate the most likely image category of the image to be classified according to the naive Bayesian algorithm and the probability of different data combinations in the certificate image database, and If the probability meets the set threshold requirement, the classification of the picture is judged; the method can simply and quickly classify the certificate image, and improve the efficiency of certificate image retrieval.

Description

Naive Bayesian Image Classification Method for Scanning Certificates Based on Feature Weighting

technical field

The invention relates to an image classification method, in particular to a scanning certificate image classification method.

Background technique

In recent years, image retrieval is a very popular topic, and its retrieval objects include objects swimming in the sea, flying in the sky and walking on the ground. Image classification is a preprocessing process of image retrieval, which can effectively improve the accuracy of image retrieval. Although there are many image classification and retrieval systems for different kinds of image datasets, less attention has been paid to the classification and retrieval of scanned certificate images, which are often important auxiliary materials for applications for awards or company expansion. In order to ensure the legitimate use of such certificate images and avoid multiple use of the same certificate, it is very important for some retrieval systems to check the scanned images in the special scanned certificate data set, which is similar to the similarity of documents examine. At present, the image features suitable for the popular content-based image classification retrieval system include color, texture, shape and spatial position relationship, but the image quality of the scanned certificate is low, there are many types, and the layout forms are diverse, including image signs with specific meanings, At the same time, it contains a concise description of the award-winning situation. Therefore, it is difficult to find whether there is an image file similar to the certificate to be tested from the massive image library by using the existing algorithm. Therefore, we have to specifically analyze the features of the scanned image and select features that can better describe the features of the certificate image. How to use computer technology to quickly and accurately conduct similarity detection on the attached certification materials—scanned images—is an urgent problem to be solved in the review of national science and technology awards.

Contents of the invention

The invention provides a scanning certificate image classification method, which can quickly and effectively classify certificate images, and can significantly improve the accuracy of certificate image retrieval.

To achieve the above object, the technical scheme of the present invention is as follows:

A feature-weighted naive Bayesian scanning certificate image classification method, comprising the following steps:

Step 1: Build a likelihood probability index for different data combinations of scanned certificate images;

Step 2: Read the scanned certificate image to be classified and perform preprocessing;

Step 3: Use the Hough transform to locate the seal on the preprocessed certificate image, obtain the circumscribed rectangular area of the seal, and extract the HSV color feature vector of the seal area;

Step 4: weighting the salient feature items of the HSV color feature vector;

Step 5: Calculate and record the probability of occurrence of different data combinations in the HSV color feature vector of the extracted circle stamp area;

Step 6: According to the HSV color feature vector of the image to be classified, the prior probability of each type of scanned certificate image and the likelihood probability index of different data combinations of the scanned certificate image obtained during the training process, the naive Bayesian algorithm is used to calculate the probability of the image to be classified In case of classification, return scanned certificate images that meet the set threshold requirements as classification results. The beneficial effects of the present invention are: the present invention is based on the feature-weighted Naive Bayesian scanning certificate image classification method, and the HSV of the circular stamp area is extracted by using the Hough transform on the preprocessed certificate image to perform seal positioning, segmentation, and size adjustment Space color feature vector and image aspect ratio; establish a certificate image database, process each certificate image in the database according to the above steps, and obtain the round stamp HSV color feature vector and image length and width of each scanned certificate image in the database According to the obtained eigenvectors, the probability of different data combinations in the certificate image database is calculated, and the data is saved after weighting processing; the most likely probability of the image to be classified is calculated according to the naive Bayesian algorithm and the probability of different data combinations in the certificate image database. image category, and the probability meets the set threshold requirements, the classification of the picture is judged; through this classification method, the certificate image can be classified simply and quickly, and the efficiency of certificate image retrieval can be effectively improved.

Description of drawings

FIG. 1 is a flowchart of an image classification method according to an embodiment of the present invention.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and examples.

Referring to Fig. 1, the Naive Bayesian scanning certificate image classification method based on feature weighting in this embodiment contains the following steps: A kind of feature weighting-based Naive Bayesian scanning certificate image classification method includes the following steps:

A: Input the scanned certificate image to be classified for preprocessing;

B: Use the Hough transform to locate the seal on the preprocessed certificate image, obtain the circumscribed rectangular area of the seal, and extract the HSV color feature vector of the seal area;

C: Weighting the significant feature items of the HSV color feature vector;

D: Calculate and record the probability of different data combinations appearing in the HSV color feature vector of the extracted circle stamp area;

Each certificate image in the certificate image database is processed according to the above steps A to D, and the prior probability of each type of scanned certificate image in the database and the probability of different data combinations in the HSV color feature vector extracted from the stamp area are calculated and recorded , that is, to establish a likelihood probability index of different data combinations of scanned certificate images;

E: According to the HSV color feature vector of the image to be classified, the prior probability of each type of scanned certificate image and the likelihood probability index of different data combinations of the scanned certificate image obtained during the training process, the classification of the image to be classified is calculated using the naive Bayesian algorithm case, return the scanned certificate image that meets the set threshold requirements as the classification result;

The naive Bayesian algorithm used in this method is as follows:

${\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; NB</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arg</span>}\mathrm{<span\; class="notranslate">\; max</span>}\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; \&Pi;</span>}}\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; a</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>$

$\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; \&Pi;</span>}}\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>$

The goal of this classification method is to obtain the most probable category of the certificate image according to the feature vector of the seal of the image to be classified, P(v _j ) is the prior probability, and it is enough to calculate the frequency of each category appearing in the certificate image database. v _NB represents the target value output by the Naive Bayesian classifier. In summary, based on their probabilities on the training data, Naive Bayesian learning methods need to estimate different P(v _j ) and P(a _i |v _j ) terms corresponding to the hypotheses to be learned, and then use The rules proposed by Naive Bayes to classify. The difference between the Naive Bayesian algorithm we use and other classification algorithms is that it only needs to simply calculate the frequency of occurrence of different data combinations in the training samples, without searching.

(L _k0 ,L _k1 ...L _k16 ) are the HSV color feature vectors and aspect ratios of the round stamp area of the image to be queried, and (L _i0 ,L _i2 ...L _i16 ) are the scanned certificate images in the database The HSV color feature vector and image aspect ratio of the stamp area.

The preprocessing in the step A is to use the existing noise filtering and tilt correction method for preprocessing;

In the step B, the preprocessed certificate image is utilized for the existing seal positioning method, and the circumscribed rectangle where the seal obtained by positioning is segmented and extracted to obtain the seal area, and the HSV color feature vector of the seal area is extracted ;

The specific operation steps are as follows:

1) Utilize the method for existing badge location, segment and extract the circumscribed rectangle where the badge obtained by positioning obtains the badge area;

2) The three components of chroma H, saturation S and brightness V are non-uniformly quantized into 8 parts, 4 parts and 4 parts respectively:

$\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 315,23</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 24,50</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 2</span>}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 51,75</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 3</span>}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 76,155</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 4</span>}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 156,195</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 5</span>}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 196,275</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 6</span>}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 276,290</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 7</span>}& \mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 290,316</span>}<span\; class="notranslate">\; ]</span>\end{array}\right.\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; 0</span>}& <span\; class="notranslate">\; S</span><span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 0,0.08</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0.08,0.4</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 2</span>}& \mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0.4,0.67</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 3</span>}& \mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0.67,1.0</span>}<span\; class="notranslate">\; ]</span>\end{array}\right.\mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 0,0.08</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0.08,0.4</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 2</span>}& \mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0.4,0.67</span>}<span\; class="notranslate">\; ]</span>\\ \mathrm{<span\; class="notranslate">\; 3</span>}& \mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0.67,1.0</span>}<span\; class="notranslate">\; ]</span>\end{array}\right.<span\; class="notranslate">\; ;</span>$

In this way, the HSV space of the round stamp area is divided into L _H + L _S + L _V intervals, L _H , L _S , and L _V are the quantization series of H, S, and V respectively, so we get a sixteen-dimensional color feature vector, plus the aspect ratio of the scanned image, and finally extract a seventeen-dimensional feature vector;

3) The naive Bayesian method is to count each data that appears, and to count the frequency of its occurrence. In order to facilitate the calculation, after trial and error, the best effect can be obtained by extracting single-digit integers for all eigenvalues. The seventeen-dimensional feature selected by this method is represented by (L _k0 , L _k1 ... L _k16 ), and the value range is an integer between [0, 9].

In the step C, the significant feature items of the feature vector are weighted.

The distribution of image features has such characteristics: in the same image category, if the statistical distribution of a certain feature is relatively dense and the degree of dispersion is relatively small, then this feature is relatively dominant with this category and is an important feature. On the contrary, if the statistics of a certain feature are scattered and the degree of dispersion is relatively high, it is an unimportant feature. The standard deviation of the data can well describe the discrete situation of the data. This method uses standard deviation to measure the image feature weights. w _i ={w _ko ,w _k1 ...w _k16 } represents the weight of the feature vector. The standard deviation σ _i of the i-th dimension of category j in the sample set, its calculation formula is:

${\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\sqrt{\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; L</span>}}_{\mathrm{<span\; class="notranslate">\; the\; ki</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; no</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>}$

n _j is the number of samples of class j, L _ki is the i-th dimension feature value of the k-th sample of image category j, is the average value of the feature of this dimension. Use e _i to represent feature importance, e _i ∈ [0,1] is the formula: Thus, the calculation method for the weighting of each dimension feature of each sample is: $w_{\mathrm{the\; ki}}=e_i/\underset{i=0}{\overset{16}{\mathrm{\&Sigma;}}}{e}_i.$

Among them, the probability of occurrence of different data combinations in the feature vector of the extracted circle stamp area is calculated and recorded, and the specific operation steps are as follows:

1) The probability of occurrence of different data in the statistical feature vector, for example, the probability of occurrence of 4 in the second dimension of the first category is 30%;

2) The obtained probability value is multiplied by the weight calculated in step C, and stored as the probability of occurrence of different data combinations.

Naive Bayesian scanning certificate image classification method based on feature weighting, the specific operation steps are as follows:

1) According to the probability of occurrence of different data combinations obtained in step D and the naive Bayesian algorithm, calculate the probability that the certificate image to be classified is an image of each type. For example, assuming that image A is the first type of image, and the number 4 appears in the second dimension, find the corresponding probability value in the probability saved in step D, and search and calculate all the data combinations that appear according to the probability of step D;

2) Obtain the probability that the certificate belongs to each category, and the maximum value is greater than the threshold, then it is judged that the certificate belongs to the category with the highest probability. The threshold was set at 0.048.

The classification results of scanned certificate images in this embodiment are shown in the following table.

Number of test pictures The number of correct classification Misclassified sheets Accuracy A class of software copyright scan certificate image 10 10 0 100% Class II software copyright scanning certificate image 10 10 0 100% Patent scan certificate image 10 10 0 100% other interfering images 10 9 1 90%

Claims (7)

1. a naive Bayesian scanning certificate image classification method for feature based weighting, is characterized in that, comprise the steps:

Step 1: set up the likelihood probability index that a scanning certificate graphs combines as different pieces of information;

Step 2: read scanning certificate graphs picture to be sorted, carry out pre-service;

Step 3: justify Zhang Dingwei to through pretreated certificate imagery exploitation Hough transform, obtains circle chapter circumscribed rectangular region, extracts the hsv color proper vector in circle chapter region;

Step 4: hsv color proper vector notable feature item is weighted;

Step 5: calculate and record the probability that in the hsv color proper vector extracting circle chapter region, different pieces of information combination occurs;

Step 6: the likelihood probability index that the scanning certificate graphs obtained according to prior probability and the training process of the hsv color proper vector of image to be classified, every class scanning certificate graphs picture combines as different pieces of information, utilize NB Algorithm to calculate the classification situation of image to be classified, return the result of scanning certificate graphs picture as classification of the threshold requirement meeting setting.

2. the naive Bayesian scanning certificate image classification method of feature based weighting according to claim 1, it is characterized in that, each the width certificate graphs picture in certificate image data base carries out processing obtaining to 5 according to step 2 by the likelihood probability index that step 1 foundation scanning certificate graphs combines as different pieces of information.

3. the naive Bayesian scanning certificate image classification method of feature based weighting according to claim 1, it is characterized in that, in described step 2, pre-service utilizes existing noise filtering and sloped correcting method.

4. the naive Bayesian scanning certificate image classification method of feature based weighting according to claim 1, it is characterized in that, the concrete operation step of described step 3 is as follows:

1) utilize the method for existing round Zhang Dingwei, segmentation is carried out to the boundary rectangle of locating the round chapter place obtained and extracts, obtain circle chapter region;

2) by colourity H, saturation degree S and brightness V tri-components respectively non-uniform quantizing be 8 parts, 4 parts and 4 parts:

$H=\left\{\begin{array}{cc}0& H\&Element;\left[\mathrm{315,23}\right]\\ 1& H\&Element;\left[\mathrm{24,50}\right]\\ 2& H\&Element;\left[\mathrm{51,75}\right]\\ 3& H\&Element;\left[\mathrm{76,155}\right]\\ 4& H\&Element;\left[\mathrm{156,195}\right]\\ 5& H\&Element;\left[\mathrm{196,275}\right]\\ 6& H\&Element;\left[\mathrm{276,290}\right]\\ 7& H\&Element;\left[\mathrm{290,316}\right]\end{array}\right.$ $S=\left\{\begin{array}{cc}0& S\&Element;\left[\mathrm{0,0.08}\right]\\ 1& S\&Element;\left(\mathrm{0.08,0.4}\right]\\ 2& S\&Element;\left(\mathrm{0.4,0.67}\right]\\ 3& S\&Element;\left(\mathrm{0.67,1.0}\right]\end{array}\right.$ $V=\left\{\begin{array}{cc}0& V\&Element;\left[\mathrm{0,0.08}\right]\\ 1& V\&Element;\left(\mathrm{0.08,0.4}\right]\\ 2& V\&Element;\left(\mathrm{0.4,0.67}\right]\\ 3& V\&Element;\left(\mathrm{0.67,1.0}\right]\end{array}\right.;$

The HSV space in so round chapter region is divided into L _h+ L _s+ L _vindividual interval, L _h, L _s, L _vbe the quantification progression of H, S and V respectively, obtain the color feature vector of ten 6 DOFs, add scan image picture length breadth ratio, final extraction ten 7 degree of freedom proper vectors;

3) the ten 7 degree of freedom feature (L extracted _k0, L _k1... L _k16) represent, span is the integer between [0,9].

5. the naive Bayesian scanning certificate image classification method of feature based weighting according to claim 1, it is characterized in that, the described step 4 pair concrete operation step that proper vector notable feature item is weighted is: adopt standard deviation to weigh characteristics of image weight, w _i={ w _ko, w _k1... w _k16the weight of representation feature vector, in sample set, classification is the standard deviation sigma of i-th dimension of j _i, its computing formula is:

${\mathrm{\&sigma;}}_i=\sqrt{\underset{k=1}{\overset{n_j}{\mathrm{\&Sigma;}}}(L_{\mathrm{ki}}-\stackrel{\&OverBar;}{x_i})/(n_j-1)}$

N _jfor j class sample number, L _kibe the i-th dimensional feature value of a kth sample of j for image category, for the mean value of this dimensional feature, use e _irepresentation feature importance, e _i∈ [0,1] is formula: thus the computing method obtaining the every dimensional feature weighting of each sample are: $w_{\mathrm{ki}}=e_i/\underset{i=0}{\overset{16}{\mathrm{\&Sigma;}}}{e}_i.$

6. the naive Bayesian scanning certificate image classification method of feature based weighting according to claim 1, it is characterized in that, described step 5 calculates and records the concrete operation step of probability that in the proper vector extracting circle chapter region, different pieces of information combination occurs and is: the probability that in statistical nature vector, different pieces of information occurs; The probable value obtained is multiplied by the weight calculated in step 4, and the probability occurred as different pieces of information combination is preserved.

7. the naive Bayesian scanning certificate image classification method of feature based weighting according to claim 1, it is characterized in that, described step 6 is specially: the probability occurred according to the different pieces of information combination obtained in step 5 and NB Algorithm, calculate the probability that certificate graphs picture to be sorted is every class image; Obtain the probability that certificate is each class, and maximal value is greater than threshold value, then judge that certificate is the classification of maximum probability, threshold value is set as 0.048.