Object detection apparatus and control method thereof, and storage medium

• US 9,202,126 B2
• Filed: 08/09/2013
• Issued: 12/01/2015
• Est. Priority Date: 08/22/2012
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Tojo, Hiroshi
• Primary Examiner(s): BHATNAGAR, ANAND P
• Application Number:
  US13/963,690
• Publication Number:
  US 20140056473A1
• Time in Patent Office: 844 Days
• US Class Current: 1/1
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation
• CPC Class Codes: G06T 7/73 using feature-based methods
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes

Abstract

The object detection apparatus prevents or eliminates detection errors caused by changes of an object which frequently appears in a background. To this end, an object detection apparatus includes a detection unit which detects an object region by comparing an input video from a video input device and a background model, a selection unit which selects a region of a background object originally included in a video, a generation unit which generates background object feature information based on features included in the background object region, and a determination unit which determines whether or not the object region detected from the input video is a background object using the background object feature information.

Claims

1. An object detection apparatus comprising:

a video input unit configured to input a frame of a video;

an object region detection unit configured to detect at least one object region by comparing a feature amount of the input frame and a feature amount of at least one background in the frame represented by background information in a background model;

a selection unit configured to select a background object region including a background object which is originally included in the background, from the detected at least one object region by using at least one first condition to be satisfied by the background object;

a discrimination unit configured to discriminate between a portion corresponding to the background object and a portion other than the background object in the background object region by using at least one second condition to be satisfied by a region which does not include the background object;

a generation unit configured to generate background object feature information representing a feature amount of the background object based on feature amounts of the background object region and a result of discrimination by said discrimination unit; and

an addition unit configured to add the background object feature information to the background model, thereby forcing said object region detection unit not to detect the background object as the object region from a succeeding frame.

2. The apparatus according to claim 1, wherein the background object feature information is a statistical amount based on feature amounts extracted from the background object region.

3. The apparatus according to claim 2, wherein the feature amounts are feature amounts according to a scene to be applied.

4. The apparatus according to claim 1, wherein one of the first conditions is based on a duration since the object region appeared in the video.

5. The apparatus according to claim 1, further comprising a moving object detecting unit configured to detect a moving object from the background object region,

wherein one of the second conditions regards a moving object region detected by said moving object detecting unit as the region which does not include the background object.

6. The apparatus according to claim 1, further comprising a translation/rotation determination unit configured to determine whether the object region is detected as a result of parallel translation of an object or is detected as a result of out-of-plane rotation of the object,

wherein when said translation/ rotation determination unit determines that the object region is detected as a result of parallel translation, one of the second conditions regards the object region as a partial background region which appears as a result of parallel translation.

7. The apparatus according to claim 1, wherein said determination unit includes a unit configured to update the background model based on the determination result.

8. The apparatus according to claim 1, further comprising a weight adjusting unit configured to adjust weights for feature amounts of the background object region by relatively lessening a weight for a region other than the background object in the background object region, and

wherein said generation unit is configured to generate the background object feature information representing a feature amount of the background object based on feature amounts of the background object region and the adjusted weights thereof.

9. The apparatus according to claim 1, wherein the background model includes, for each position in the frame, at least one background feature amount and time information indicating the time of occurrence of the at least one background feature amount,

wherein the apparatus further comprises an update unit configured to, if a feature amount of a position of interest is within an acceptable error regarding one of the background feature amounts, update the background feature amount of the one with the feature amount of the position of interest, but not update the information indicating the time of occurrence.

10. The apparatus according to claim 9, wherein said update unit is configured to, if the feature amount of the position of interest is not within the acceptable error regarding anyone of the background feature amounts, add the feature amount of the position of interest as a new background feature amount corresponding to the position of interest and the time of the current frame as the information indicating the time of occurrence of the new background feature amount, into the background model.

11. A control method of controlling an object detection apparatus which comprises a video input unit configured to input a frame of a video, comprising:

an object region detection step of detecting at least one object region by comparing a feature amount of the input frame from the video input unit and a feature amount of at least one background in the frame represented by background information in a background model;

a selection step of selecting a background object region including a background object which is originally included in the background, from the detected at least one object region by using at least one first condition to be satisfied by the background object;

a discrimination step of discriminating between a portion corresponding to the background object and a portion other than the background object in the background object region by using at least one second condition to be satisfied by a region which does not include the background object;

a generation step of generating background object feature information representing a feature amount of the background object based on feature amounts of the background object region and a result of discrimination in said discrimination step; and

an addition step of adding the background object feature information to the background model, thereby forcing said object region detection step not to detect the background object as the background region from a succeeding frame.

12. A non-transitory computer-readable storage medium storing a program for controlling a computer, which includes a video input unit configured to input a video, to execute respective steps of a method according to claim 11 when the computer loads and executes the program.