Object detection apparatus and method therefor

• US 8,913,782 B2
• Filed: 01/11/2010
• Issued: 12/16/2014
• Est. Priority Date: 01/13/2009
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Yashiro, Satoshi, Tojo, Hiroshi, Guo, Muling
• Primary Examiner(s): Akhavannik, Hadi
• Assistant Examiner(s): RASHIDIAN, MOHAMMAD M
• Application Number:
  US12/685,521
• Publication Number:
  US 20100226538A1
• Time in Patent Office: 1,800 Days
• Field of Search: 382/100, 382103-104, 382106-107, 382/115, 382/118, 382/181, 382164-165, 382/167, 382/218, 382/232, 382/250, 382/284, 348/722.1, 348/143, 348/149, 348/150, 348/169, 348/14.16, 348/154, 348/77, 711/209, 700/86, 700/94, 706/62, 706/45, 706/47, 706/59, 706/12, 706/25
• US Class Current: 382/103
• CPC Class Codes: G06F 18/2148 characterised by the process organisation or structure, e.g. boosting cascade
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/30201 Face
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/143 involving probabilistic approaches, e.g. Markov random field [MRF] modelling
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters
• CPC Class Codes: G06V 40/167 using comparisons between temporally consecutive images

Abstract

An image processing apparatus includes a moving image input unit configured to input a moving image, an object likelihood information storage unit configured to store object likelihood information in association with a corresponding position in an image for each object size in each frame included in the moving image, a determination unit configured to determine a pattern clipping position where a pattern is clipped out based on the object likelihood information stored in the object likelihood information storage unit, and an object detection unit configured to detect an object in an image based on the object likelihood information of the pattern clipped out at the pattern clipping position determined by the determination unit.

Claims

1. An image processing apparatus comprising:

a moving image input unit configured to input a moving image;

a storage unit configured to store object likelihood information in association with a corresponding position in a frame in the moving image;

a setting unit configured to sequentially set a plurality of collation patterns in a frame with respective positions based on the object likelihood information stored in the storage unit; and

a determination unit configured to derive object reliability information for each of the collation patterns, determine whether each of the collation patterns represents an object based on the derived object reliability information, and update the object likelihood information stored in the storage unit in association with a position of each of the collation patterns.

2. The image processing apparatus according to claim 1, wherein the setting unit determines the positions of the collation patterns based on a position associated with object likelihood information that exceeds a predetermined threshold among pieces of the object likelihood information stored in the storage unit.

3. The image processing apparatus according to claim 1, wherein the determination unit determines object likelihood information at a position other than the positions of the collation patterns derived for a former frame based on the object likelihood information at a position other than the position, and stores the derived object likelihood information in the storage unit.

4. The image processing apparatus according to claim 1, wherein the determination unit generates motion vector information based on a plurality of frames, determines object likelihood information at a position other than the positions of the collation patterns derived for a former frame based on the object likelihood information at a position determined based on the motion vector information, and stores the derived object likelihood information in the storage unit.

5. The image processing apparatus according to claim 1, wherein the setting unit determines an amount of movement, for moving to sequentially set the collation patterns based on the object likelihood information stored in the storage unit.

8. The image processing apparatus according to claim 1, further comprises a generation unit configured to generate frames of a plurality of sizes by sequentially reducing a frame in the moving image,

wherein the setting unit sets the collation patterns in a frame generated by the generation unit.

9. The image processing apparatus according to claim 8, wherein the storage unit stores object likelihood information in association with a position in a frame and a size of the frame.

6. A method for image processing in an image processing apparatus, the method comprising:

inputting a moving image;

sequentially setting a plurality of collation patterns in a frame with respective positions based on object likelihood information stored in a storage unit configured to store the object likelihood information in association with a corresponding position in a frame in the moving image; and

deriving object reliability information for each of the collation patterns, determining whether each of the collation patterns represents an object based on the object reliability information obtaining the object likelihood information based on the object reliability information for each of the collation patterns, and updating the object likelihood information stored in the storage unit in association with a position of each of the collation patterns.

7. A computer-readable storage medium storing a program that causes a computer to execute a method, the method comprising:

inputting a moving image;

sequentially setting a plurality of collation patterns in a frame with respective positions based on object likelihood information stored in a storage unit configured to store the object likelihood information in association with a corresponding position in a frame in the moving image; and

deriving object reliability information for each of the collation patterns, determining whether each of the collation patterns represents an object based on the object likelihood information obtaining the object likelihood information based on the object reliability information and updating the object likelihood information stored in the storage unit in association with a position of each of the collation patterns.