Method for automatic detection and classification of objects and patterns in low resolution environments

• US 7,970,212 B2
• Filed: 11/30/2006
• Issued: 06/28/2011
• Est. Priority Date: 12/08/2005
• Status: Active Grant

• Current Assignee: OSR Enterprises AG
• Original Assignee: Amir Zahavi, Shai Nissim
• Inventors: Zahavi, Amir, Nissim, Shai
• Primary Examiner(s): Strege; John B
• Application Number:
  US12/096,025
• Publication Number:
  US 20080285856A1
• Time in Patent Office: 1,671 Days
• Field of Search: 382/173, 382/190, 382/195, 382/207, 382/224, 382/240, 382/276, 382/277
• US Class Current: 382/207
• CPC Class Codes: G06V 10/421 by analysing segments intersecting the pattern

Abstract

The invention is a method of using Wavelet Transformation and Artificial Neural Network (ANN) systems for automatic detecting and classifying objects. To train the system in object recognition different images, which usually contain desired objects alongside other objects are used. These objects may appear at different angles. Different characteristics regarding the objects are extracted from the images and stored in a data bank. The system then determines the extent to which each inserted characteristic will be useful in future recognition and determines its relative weight. After the initial insertion of data, the operator tests the system with a set of new images, some of which contain the class objects and some of which contain similar and/or dissimilar objects of different classification. The system learns from the images containing similar objects of different classes as well as from the images containing the class objects, since each specific class characteristic needs to be set apart from other class characteristic. The system may be tested and trained again and again until the operator is satisfied with the system'"'"'s success rate of object recognition and classification.

Claims

1. A method of automatically detecting and classifying objects, the method comprising:

a. providing an original image comprising one or more objects;

b. creating a new image of each of said objects isolated from all other objects of said original image by performing a process of image separation;

c. extracting geometrical characteristics of the object from the new image, and storing them in a feature vector;

d. transforming at least one of said new images of said isolated object into a vector using at least one of erasing and shifting, wherein said transforming comprises reading the pixels of the new image vector by vector, where said vector contains information of said isolated object, and said vector comprises at least one consecutive background pixel for each vector of the at least one new image;

e. applying a 1 Dimensional Continuous Wavelet Transform to said vector;

f. extracting from said 1 Dimensional Continuous Wavelet Transform, both the scale, and coefficient values of said isolated object and storing said extracted values in said feature vector;

g. comparing said feature vector to at least one other feature vector from a data base; and

h. classifying each of said one or more objects responsive to said comparing of said feature vector.

2. A method according to claim 1, wherein the performed process of image separation comprises:

a. applying a 2 Dimensional Discrete Wavelet Transform to the original image;

b. detecting edges of at least one of the objects or providing a threshold to detect at least one of the objects;

c. producing a contrast image responsive to said detected edges or said provided threshold, wherein pixels of the at least one object have a different value from background pixels;

d. refining said contrast image by eliminating isolated pixels; and

e. assigning an individual value to each of the at least one object, where said pixels of the at least one object are assigned said individual value; and

f. producing a new image for each of the at least one object containing said pixels of the at least one object and background pixels.

3. A method according to claim 1, further comprising the step of:

reducing the resolution of the created new image.

4. A method according to claim 2, where the applied 2 Dimensional Discrete Wavelet Transform derives one of the approximation coefficient matrix obtained by the wavelet decomposition of the original image and the reconstruction matrix of said coefficients.

5. A method according to claim 2, where the 2 Dimensional Discrete Wavelet Transform derives one of the approximation coefficients matrix, the detail coefficient matrixes of the original image and the reconstruction matrixes of said coefficients.

6. A method according to claim 2, where the 2 Dimensional Discrete Wavelet Transform derives one of the vertical detail coefficients matrix of the original image and the reconstruction matrix of said coefficients.

7. A method according to claim 2, where the 2 Dimensional Discrete Wavelet Transform derives one of the horizontal detail coefficients matrix of the original image and the reconstruction matrix of said coefficients.

8. A method according to claim 2, where the 2 Dimensional Discrete Wavelet Transform derives one of the diagonal detail coefficients matrix of the original image and the reconstruction matrix of said coefficients.

9. A method according to claim 2, where the 2 Dimensional Discrete Wavelet Transform is applied in “

slices”

to a multi dimensional original image.

10. A method according to claim 2, where said produced new image contains the original object pixels from the original image.

11. A method according to claim 2, where pixels of all of the at least one object of said produced new image are assigned the same value.

12. A method according to claim 1, where the object is a pattern in an image.

13. A method according to claim 1, where the provided original image is a 2 dimensional image.

14. A method according to claim 1, where the provided original image has more than 2 dimensions.

15. A method according to claim 1, where the extracted features are wavelet scale values.

16. A method according to claim 1, where the extracted features are wavelet coefficients.

17. A method according to claim 1, where the information of said extracted at least one feature comprises at least one of:

a scale in which the maximum or minimum coefficient is located;

a value of the maximum coefficient; and

a value of the minimum coefficient.

18. A method according to claim 1, where the Wavelet used in the applied 1 Dimensional Continuous Wavelet Transform may belong to any one of the known Mother Wavelets.

19. A method according to claim 1, where the Wavelet used in the applied 1 Dimensional Continuous Wavelet Transform may be a newly created Mother Wavelet.

20. A method according to claim 1, where the transforming is accomplished by a partial rearrangement.

21. A method according to claim 1, where the comparing of the feature vector to at least one other feature vector is done by an Artificial Neural Network.