Method of object recognition, apparatus of the same and recording medium therefor

• US 6,647,139 B1
• Filed: 02/17/2000
• Issued: 11/11/2003
• Est. Priority Date: 02/18/1999
• Status: Expired due to Term

• Current Assignee: Matsushita Electric Industrial Company Limited (Panasonic Holdings Corporation)
• Original Assignee: Matsushita Electric Industrial Company Limited (Panasonic Holdings Corporation)
• Inventors: Nagao, Kenji, Kunii, Mihoko
• Primary Examiner(s): AHMED, SAMIR ANWAR
• Application Number:
  US09/506,483
• Time in Patent Office: 1,363 Days
• Field of Search: 382/159, 382/171, 382/190, 382/195, 382/199, 382/206, 382/224, 700/47, 700/48, 706/16, 706/19, 706/25
• US Class Current: 382/159
• CPC Class Codes: G06F 18/2132 based on discrimination criteria, e.g. discriminant analysis
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
• CPC Class Codes: G06V 20/56 exterior to a vehicle by using sensors mounted on the vehicle
• CPC Class Codes: G06V 2201/08 Detecting or categorising vehicles

Abstract

According to a disclosed method, an image is learned beforehand, and an image of an object to be recognized is entered, then this object is recognized. An image including an object to be learned is entered, and it is divided into partial images. Further classifying into plural classes, a matrix for feature extraction is calculated in each class. A feature is calculated by using this matrix for feature extraction, and stored. Consequently, an image including an object to be recognized is entered, and it is divided into partial images. From the partial images, the feature of the object of recognition is calculated by using the obtained matrix for feature extraction, and the similarity measure of the both is calculated by using the data of the stored feature and the feature of the object of recognition, and the object is recognized.

Claims

1. An object recognition method comprising at least a learning step of learning a first entered image, and a recognition step of recognizing a second entered image, wherein said learning step includes:

a step of entering the first image including the object to be learned,

a step of dividing said entered image into a first partial image,

a step of classifying said first partial image into plural classes,

a step of calculating a matrix for feature extraction from said partial image classified into classes,

a step of calculating a first feature by using said matrix for feature extraction from said partial image classified into classes, and

a step of storing the data of said first feature, and

said recognition step includes;

a step of receiving a second image including the object to be recognized,

step of dividing said entered image into a second partial image,

a step of calculating a second feature by using a matrix for feature extraction from said second partial image,

a step of calculating the similarity measure of the first image and second image by using said stored first feature data and second feature,

a step of recognizing the object in said second image by using said similarity measure, and

a step of issuing said result of recognition,

wherein the step of entering the first image including the object to be learned is to enter plural pieces of image information mutually different in properties about one object to be learned, said learning step further includes a step of integrating said plural pieces of information and generating first integrated information, said first partial image includes said first integrated information, said step of entering the second image including the object to be recognized is to enter plural pieces of image information mutually different in properties about one object to be recognized, said recognizing step further includes a step of integrating said plural pieces of information and generating second integrated information, and said second partial image includes second integrated information.

2. The object recognition method of claim 1, wherein the step of dividing into first and second partial images is to limit a partial region of input image.

3. The object recognition method of claim 1, wherein said integrated information includes at least optical information and geometrical information.

4. The object recognition method of claim 3, wherein at least the optical information is contrast distribution information of image, or the geometrical information is edge information of image.

5. The object recognition method of claim 1, wherein the class is classified by elements including at least a pair of the object position and kind.

6. The object recognition method, of claim 5, wherein the object position is expressed by the distance to the object.

7. The object recognition method of claim 1, wherein the matrix for feature extraction is determined from the within-class covariance matrix and between-class covariance matrix.

8. The object recognition method of claim 7, wherein the matrix for feature extraction is determined so as to maximize the Fisher ratio calculated from the within-class covariance matrix and between-class covariance matrix.

9. The object recognition method of claim 1, wherein the similarity measure is judged by the normalized correlation in the feature space.

12. The object recognition method of claim 1, wherein the step of entering the first image and the step of dividing into a first partial image may be replaced by a step of creating by an artificial synthesizing technique.

13. The object recognition method of claim 1, wherein the learning step further includes a step of converting the scale of the first partial image in each class.

14. The object recognition method of claim 1, wherein the learning step further includes a step of cutting out into different shapes in every class when dividing the first partial image.

15. The object recognition method of claim 1, wherein the class is classified by the difference in the shape of said object to be learned.

16. The object recognition method of claim 1, wherein the class is classified by the difference in the size of said object to be learned.

10. An object recognition apparatus comprising at least learning means for learning a first entered image, and recognition means for recognizing a second entered image, wherein said learning means includes:

means for entering the first image including the object to be learned,

means for dividing said entered image into a first partial image,

means for classifying said first partial image into plural classes,

means for calculating a matrix for feature extraction from said partial images image classified into classes,

means for calculating a first feature by using said matrix for feature extraction from said partial image classified into classes, and

means for storing the data of said first feature, and

said recognition means includes;

means for receiving a second image including the object to be recognized,

means for dividing said entered image into a second partial image,

means for calculating a second feature by using a matrix for feature extraction from said partial image,

means for calculating the similarity measure of the first image and second image by using said stored first feature data and second feature,

means for recognizing the object in said second image by using said similarity measure, and

means for issuing said result of recognition,

wherein the means for entering the first image including the object to be learned is to enter plural pieces of image information mutually different in properties about one object to be learned, said learning means further includes a means for integrating said plural pieces of information and generating first integrated information, said first partial image includes said first integrated information, said means for entering the second image including the object to be recognized is to enter plural pieces of image information mutually different in properties about one object to be recognized, said recognizing means further includes a means for integrating said plural pieces of information and generating second integrated information, and said second partial image includes second integrated information.

17. The object recognition apparatus of claim 10, wherein the scale of partial image is converted in every class, and a feature vector is created.

18. The object recognition apparatus of claim 10, wherein the cutting shape of partial image is set in every class, and a feature vector is created.

19. The object recognition apparatus of claim 10, wherein the class is classified by the difference in the shape of partial image.

20. The object recognition apparatus of claim 10, wherein the class is classified by the difference in the size of partial image.

11. A recording medium in which a computer program is recorded, being a recording medium of which program presents a method, being read by the computer and installed, for recognizing a specific object, said method having at least a learning step of learning a first entered image, and a recognition step of recognizing a second entered image, wherein said learning step includes:

a step of entering the first image including the object to be learned,

a step of dividing said entered image into a first partial image,

a step of classifying said first partial image into plural classes,

a step of calculating a matrix for feature extraction from said partial image classified into classes,

a step of calculating a first feature by using said matrix for feature extraction from said partial image classified into classes, and

a step of storing the data of said first feature, and

said recognition step includes;

a step of receiving a second image including the object to be recognized,

a step of dividing said entered image into a second partial image,

a step of calculating a second feature by using a matrix for feature extraction from said second partial image,

a step of calculating the similarity measure of the first image and second image by using said stored first feature data and second feature,

a step of recognizing the object in said second image by using said similarity measure, and

a step of issuing said result of recognition,

wherein the step of entering the first image including the object to be learned is to enter plural pieces of image information mutually different in properties about one object to be learned, said learning step further includes a step of integrating said plural pieces of information and generating first integrated information, said first partial image includes said first integrated information, said step of entering the second image including the object to be recognized is to enter plural pieces of image information mutually different in properties about one object to be recognized, said recognizing step further includes a step of integrating said plural pieces of information and generating second integrated information, and said second partial image includes second integrated information.