METHOD AND SYSTEM FOR RECOGNIZING IMAGES

• US 20130156301A1
• Filed: 03/28/2012
• Published: 06/20/2013
• Est. Priority Date: 12/19/2011
• Status: Active Grant

• Current Assignee: Industrial Technology Research Institute
• Original Assignee: Industrial Technology Research Institute
• Inventors: Kang, Chung-Chia, Chen, Jian-Ren, Chen, Chih-Wei, Chang, Leii H.
• Granted Patent: US 8,818,050 B2
• Time in Patent Office: Days
• US Class Current: 382/159
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06V 10/772 Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries
• CPC Class Codes: G06V 40/168 Feature extraction; Face representation

Abstract

A method and a system for recognizing at least one testing image according to classes are provided, wherein each of the classes includes sample images. The method includes generating an average image of each class according to the sample images, generating a feature enhancement mask according to differences between the average images of the classes, enhancing the sample images of each class by using the feature enhancement mask to generate a plurality of enhanced sample images corresponding to each class, and training a classifier according to the enhanced sample images of each class. The method also includes enhancing the at least one testing image by using the feature enhancement mask to generate an enhanced testing image, classifying the enhanced testing image into one of the classes by using the classifier, and recognizing that the testing image belongs to the classified class. Thereby, this method can effectively recognize the testing image.

Claims

1. An image recognizing method, recognizing at least one testing image according to a plurality of classes, wherein each of the classes comprises a plurality of sample images, the image recognizing method comprising:

generating an average image of each of the classes according to the sample images;

generating a feature enhancement mask according to differences between the average images of the classes;

enhancing the sample images of each of the classes by using the feature enhancement mask to generate a plurality of enhanced sample images corresponding to the class;

training a classifier according to the enhanced sample images corresponding to each of the classes;

enhancing the at least one testing image by using the feature enhancement mask to generate at least one enhanced testing image;

classifying the at least one enhanced testing image into one of the classes by using the classifier; and

recognizing that the at least one testing image belongs to the one of the classes.

2. The image recognizing method according to claim 1, wherein each of the sample images has a plurality of pixels arranged into an array, each of the pixels is corresponding to a pixel value, and the average image of each of the classes has a plurality of average pixel values corresponding to the pixels of the sample images,

wherein the average pixel value in an i^th column and a j^th row of the average image of the m^th class is obtained by dividing a sum of the pixel values corresponding to the pixels in the i^th columns and the j^th rows of the sample images of the m^th class by the number of the sample images of the m^th class.

3. The image recognizing method according to claim 2, wherein the feature enhancement mask has a plurality of feature enhancement values arranged into the array, and the feature enhancement values are obtained according to a plurality of absolute differences between the average pixel values of the average images of the classes.

4. The image recognizing method according to claim 3, wherein a number of the classes is 2,

wherein the feature enhancement value in an i^th column and a j^th row of the feature enhancement mask is obtained by dividing the absolute difference in the i^th column and the j^th row among the absolute differences by a maximum absolute difference among the absolute differences,

wherein the absolute difference in the i^th column and the j^th row among the absolute differences is equal to an absolute value of a difference between the average pixel value in the i^th column and the j^th row of the average image of the 1^st class and the average pixel value in the i^th column and the j^th row of the average image of the 2^nd class.

5. The image recognizing method according to claim 3, wherein a number of the classes is greater than 2,

wherein the feature enhancement value in an i^th column and a j^th row of the feature enhancement mask is obtained by dividing the absolute difference in the i^th column and the j^th row among the absolute differences by a maximum absolute difference among the absolute differences,

wherein the absolute difference in the i^th column and the j^th row among the absolute differences is equal to an absolute value of a difference between a maximum average pixel value among the average pixel values in the i^th column and the j^th row of the average images of the classes and a minimum average pixel value among the average pixel values in the i^th column and the j^th row of the average images of the classes.

6. The image recognizing method according to claim 4, wherein each of the enhanced sample images has a plurality of pixels arranged into the array, and each of the pixels of each of the enhanced sample images is corresponding to a pixel value;

wherein the pixel value corresponding to the pixel in an i^th column and a j^th row of the k^th enhanced sample image of the m^th class is obtained by multiplying the feature enhancement value in an i^th column and a j^th row of the feature enhancement mask by the pixel value corresponding to the pixel in an i^th column and a j^th row of the k^th sample image of the m^th class.

7. The image recognizing method according to claim 5, wherein each of the enhanced sample images has a plurality of pixels arranged into the array, and each of the pixels of each of the enhanced sample images is corresponding to a pixel value;

wherein the pixel value corresponding to the pixel in an i^th column and a j^th row of the k^th enhanced sample image of the m^th class is obtained by multiplying the feature enhancement value in an i^th column and a j^th row of the feature enhancement mask by the pixel value corresponding to the pixel in an i^th column and a j^th row of the k^th sample image of the m^th class.

8. The image recognizing method according to claim 1, wherein the step of training the classifier according to the enhanced sample images corresponding to each of the classes comprises:

training the classifier by using a K-means-clustering algorithm, a linear discriminant analysis (LDA), an adaptive boosting (AdaBoost) algorithm, a neural network, or a support vector machine (SVM).

9. The image recognizing method according to claim 4, wherein the enhanced testing image has a plurality of pixels arranged into the array, and each of the pixels of the enhanced testing image is corresponding to a pixel value,

wherein the pixel value corresponding to the pixel in an i^th column and a j^th row of the enhanced testing image is obtained by multiplying the feature enhancement value in the i^th column and the j^th row of the feature enhancement mask by the pixel value corresponding to the pixel in the i^th column and the j^th row of the corresponding testing image.

10. The image recognizing method according to claim 5, wherein the enhanced testing image has a plurality of pixels arranged into the array, and each of the pixels of the enhanced testing image is corresponding to a pixel value,

wherein the pixel value corresponding to the pixel in an i^th column and a j^th row of the enhanced testing image is obtained by multiplying the feature enhancement value in the i^th column and the j^th row of the feature enhancement mask by the pixel value corresponding to the pixel in the i^th column and the j^th row of the corresponding testing image.

11. The image recognizing method according to claim 10, further comprising:

performing a histogram equalization on the sample images and the testing image.

12. The image recognizing method according to claim 11, further comprising:

reducing a dimension of each of the enhanced sample images through a principal components analysis (PCA) and a LDA; and

reducing a dimension of the enhanced testing image through the PCA and the LDA.

13. An image recognizing system, recognizing at least one testing image according to a plurality of classes, wherein each of the classes comprises a plurality of sample images, the image recognizing system comprising:

an image capturing unit, capturing the at least one testing image;

an image recognizing unit, coupled to the image capturing unit; and

a classifier training unit, coupled to the image capturing unit and the image recognizing unit;

wherein the classifier training unit generates an average image of each of the classes according to the sample images;

wherein the classifier training unit generates a feature enhancement mask according to differences between the average images of the classes,

wherein the classifier training unit enhances the sample images of each of the classes by using the feature enhancement mask to generate a plurality of enhanced sample images corresponding to the class,

wherein the classifier training unit trains a classifier according to the enhanced sample images corresponding to each of the classes,

wherein the image recognizing unit enhances the at least one testing image by using the feature enhancement mask to generate at least one enhanced testing image,

wherein the image recognizing unit classifies the at least one enhanced testing image into one of the classes by using the classifier;

wherein the image recognizing unit recognizes that the at least one testing image which is belonged to the one of the classes.

14. The image recognizing system according to claim 13, wherein each of the sample images has a plurality of pixels arranged into an array, each of the pixels is corresponding to a pixel value, and the average image of each of the classes has a plurality of average pixel values corresponding to the pixels of the sample images,

wherein the average pixel value in an i^th column and a j^th row of the average image of the m^th class is obtained by dividing a sum of the pixel values corresponding to the pixels in the i^th columns and the j^th rows of the sample images of the m^th class by the number of the sample images of the m^th class.

15. The image recognizing system according to claim 14, wherein the feature enhancement mask has a plurality of feature enhancement values arranged into the array, and the feature enhancement values are obtained according to a plurality of absolute differences between the average pixel values of the average images of the classes.

16. The image recognizing system according to claim 15, wherein a number of the classes is 2,

wherein the feature enhancement value in an i^th column and a j^th row of the feature enhancement mask is obtained by dividing the absolute difference in the i^th column and the j^th row among the absolute differences by a maximum absolute difference among the absolute differences,

wherein the absolute difference in the i^th column and the j^th row among the absolute differences is equal to an absolute value of a difference between the average pixel value in the i^th column and the j^th row of the average image of the 1^st class and the average pixel value in the i^th column and the j^th row of the average image of the 2^nd class.

17. The image recognizing system according to claim 15, wherein a number of the classes is greater than 2,

wherein the feature enhancement value in an i^th column and a j^th row of the feature enhancement mask is obtained by dividing the absolute difference in the i^th column and the j^th row among the absolute differences by a maximum absolute difference among the absolute differences,

wherein the absolute difference in the i^th column and the j^th row among the absolute differences is equal to an absolute value of a difference between a maximum average pixel value among the average pixel values in the i^th column and the j^th row of the average images of the classes and a minimum average pixel value among the average pixel values in the i^th column and the j^th row of the average images of the classes.

18. The image recognizing system according to claim 16, wherein each of the enhanced sample images has a plurality of pixels arranged into the array, and each of the pixels of each of the enhanced sample images is corresponding to a pixel value,

wherein the pixel value corresponding to the pixel in an i^th column and a j^th row of the k^th enhanced sample image of the m^th class is obtained by multiplying the feature enhancement value in an i^th column and a j^th row of the feature enhancement mask by the pixel value corresponding to the pixel in an i^th column and a j^th row of the k^th sample image of the m^th class.

19. The image recognizing system according to claim 17, wherein each of the enhanced sample images has a plurality of pixels arranged into the array, and each of the pixels of each of the enhanced sample images is corresponding to a pixel value,

wherein the pixel value corresponding to the pixel in an i^th column and a j^th row of the k^th enhanced sample image of the m^th class is obtained by multiplying the feature enhancement value in an i^th column and a j^th row of the feature enhancement mask by the pixel value corresponding to the pixel in an i^th column and a j^th row of the k^th sample image of the m^th class.

20. The image recognizing system according to claim 13, wherein the classifier training unit trains the classifier by using a K-means-clustering algorithm, a LDA, an AdaBoost algorithm, a neural network, or a SVM.

21. The image recognizing system according to claim 16, wherein the enhanced testing image has a plurality of pixels arranged into the array, and each of the pixels of the enhanced testing image is corresponding to a pixel value,

wherein the pixel value corresponding to the pixel in an i^th column and a j^th row of the enhanced testing image is obtained by multiplying the feature enhancement value in the i^th column and the j^th row of the feature enhancement mask by the pixel value corresponding to the pixel in the i^th column and the j^th row of the corresponding testing image.

22. The image recognizing system according to claim 17, wherein the enhanced testing image has a plurality of pixels arranged into the array, and each of the pixels of the enhanced testing image is corresponding to a pixel value,

wherein the pixel value corresponding to the pixel in an i^th column and a j^th row of the enhanced testing image is obtained by multiplying the feature enhancement value in the i^th column and the j^th row of the feature enhancement mask by the pixel value corresponding to the pixel in the i^th column and the j^th row of the corresponding testing image.

23. The image recognizing system according to claim 22, wherein the image recognizing system further comprises a storage unit coupled to the classifier training unit and the image recognizing unit,

wherein the storage unit stores the classifier and the feature enhancement mask.

24. The image recognizing system according to claim 23 further comprising a normalization unit coupled to the image capturing unit, the classifier training unit, and the image recognizing unit, wherein the normalization unit performs a histogram equalization on the sample images and the testing image.

25. The image recognizing system according to claim 24, wherein the normalization unit reduces a dimension of each of the enhanced sample images through a PCA and a LDA;

and

the normalization unit reduces a dimension of the enhanced testing image through the PCA and the LDA.