Image processing apparatus, method, and program

• US 20070160294A1
• Filed: 12/13/2006
• Published: 07/12/2007
• Est. Priority Date: 12/15/2005
• Status: Active Grant

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Asano, Yasuharu
• Granted Patent: US 7,907,751 B2
• Time in Patent Office: Days
• US Class Current: 382/190
• CPC Class Codes: G06V 10/768 using context analysis, e.g. recognition aided by known co-occurring patterns
• CPC Class Codes: G06V 40/172 Classification, e.g. identification

Abstract

The present invention provides an image processing apparatus, including extraction means, parameter retaining means, context retaining means, and decision means. The extraction means is configured to extract a characteristic amount of a region in which a recognition object may possibly be included from within an image of a processing object. The parameter retaining means is configured to retain a parameter regarding the recognition object. The context retaining means is configured to retain a context regarding the recognition object. The decision means is configured to decide based on the characteristic amount extracted by the extraction means, the parameter retained in the parameter retaining means, and a result of arithmetic operation performed using the context retained in the context retaining means whether or not an image in the region is the recognition object.

Claims

1. An image processing apparatus, comprising:

extraction means configured to extract a characteristic amount of a region in which a recognition object may possibly be included from within an image of a processing object;

parameter retaining means configured to retain a parameter regarding the recognition object;

context retaining means configured to retain a context regarding the recognition object; and

decision means configured to decide based on the characteristic amount extracted by said extraction means, the parameter retained in said parameter retaining means, and a result of arithmetic operation performed using the context retained in said context retaining means whether or not an image in the region is the recognition object.

2. The image processing apparatus according to claim 1, wherein the context is a cooccurrence probability of a plurality of recognition objects.

3. The image processing apparatus according to claim 1, wherein the context is a linkage probability in order of the time between a plurality of recognition objects.

4. The image processing apparatus according to claim 1, wherein said decision means decides whether or not there exists a region in which a score calculated using the characteristic amount and the parameter is higher than a predetermined threshold value, and

performs, where it is decided that a region in which the score is higher than the predetermined threshold value exists, arithmetic operation regarding the recognition object corresponding to an image other than an image in the region using the characteristic amount, the parameter, and the context.

5. The image processing apparatus according to claim 1, wherein, when a new recognition object is set, an image in which the new recognition object exists is read out from a plurality of stored images, and

it is decided whether or not the read out image includes a different recognition object, and then

a cooccurrence probability is calculated based on a result of the decision and a context regarding the new recognition object retained in said context retaining means is updated based on the cooccurrence probability.

6. The image processing apparatus according to claim 1, wherein, when a new recognition object is set, an image in which the new recognition object exists is read out from a plurality of stored images, and

a linkage probability in order of the time regarding the new recognition object is calculated from the read out image, and then

a context regarding the new recognition object held in said context retaining means is updated based on the linkage probability.

7. The image processing apparatus according to claim 6, wherein, when the linkage probability in order of the time is calculated, a difference between time points at which the read out images are picked up is calculated and weighting is performed in accordance with the time difference.

8. An image processing method, comprising the steps of:

extracting a characteristic amount of a region in which a recognition object may possibly be included from within an image of a processing object;

retaining a parameter regarding the recognition object;

retaining a context regarding the recognition object; and

deciding based on the characteristic amount extracted by the process at the extraction step, the parameter retained by the process at the parameter retaining step, and a result of arithmetic operation performed using the context retained by the process at the context retaining step whether or not an image in the region is the recognition object.

9. A program for causing a computer to execute a process comprising the steps of:

extracting a characteristic amount of a region in which a recognition object may possibly be included from within an image of a processing object;

retaining a parameter regarding the recognition object;

retaining a context regarding the recognition object; and

deciding based on the characteristic amount extracted by the process at the extraction step, the parameter retained by the process at the parameter retaining step, and a result of arithmetic operation performed using the context retained by the process at the context retaining step whether or not an image in the region is the recognition object.

10. An image processing apparatus, comprising:

an extraction section configured to extract a characteristic amount of a region in which a recognition object may possibly be included from within an image of a processing object;

a parameter retaining section configured to retain a parameter regarding the recognition object;

a context retaining section configured to retain a context regarding the recognition object; and

a decision section configured to decide based on the characteristic amount extracted by said extraction section, the parameter retained in said parameter retaining section and a result of arithmetic operation performed using the context retained in said context retaining section whether or not an image in the region is the recognition object.