Moving object detection apparatus and method

• US 5,991,428 A
• Filed: 08/15/1997
• Issued: 11/23/1999
• Est. Priority Date: 09/12/1996
• Status: Expired due to Fees

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Taniguchi, Yasuhiro
• Primary Examiner(s): Chang, Jon
• Assistant Examiner(s): Patel, Jayanti K.
• Application Number:
  US08/912,172
• Time in Patent Office: 830 Days
• Field of Search: 382/107, 382/232, 382/236, 382/103, 382/130, 382/153, 382/154, 382/171-172, 382/173, 382/291, 345/113, 345/114, 348/586, 348/154, 348/155-220, 348/208, 348/413, 348/416, 348/699-700, 396/3, 704/260, 352/53, 352/87, 368/229
• US Class Current: 382/107
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 7/223 using block-matching
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes

Abstract

A moving object detection apparatus includes a movable input section to input a plurality of images in a time series, in which a background area and a moving object are included. A calculation section divides each input image by unit of predetermined area, and calculates the moving vector between two images in a time series and a corresponding confidence value of the moving vector by unit of the predetermined area. A background area detection section detects a group of the predetermined areas, each of which moves almost equally as the background area from the input image according to the moving vector and the confidence value by unit of the predetermined area. A moving area detection section detects the area other than the background area as the moving area from the input image according to the moving vector of the background area.

Claims

1. A moving object detection apparatus, comprising:

image input means for inputting a plurality of images including a background and a moving object in time series, wherein said image input means is moving in parallel with movement of the background in the images;

a calculation means for dividing one image into units of predetermined area, for calculating an evaluation value between the predetermined area of the one image and each neighboring area of nex image, for calculating a moving vector between the predetermined area of the one image and corresponding area of the next image, the evaluation value of the corresponding area is maximum among the evaluation values of the each neighboring area of the next image, and for calculating a confidence value of the moving vector for each unit of predetermined area on the one image, the confidence value is a distance between the corresponding area and the neighboring area whose evaluation value is above a threshold based on the maximum on the next image;

a background area detection means for calculating a value of normal distribution using the moving vector and the confidence value for each unit of predetermined area on the one image, for adding the value of normal distribution to a position of the moving vector on a moving vector space for each unit of predetermined area on the one image, and for detecting one moving vector whose value of normal distribution is maximum on the moving vector space as a moving vector of the background area; and

a moving area detection means for shifting the next image as the one moving vector, and for subtracting the shifted image from the one image to detect the moving area.

2. The moving object detection apparatus according to claim 1,

wherein said calculation means matches the predetermined area of one image with the predetermined area of the next image by shifting as one pixel unit, calculates the difference of pixel values between the predetermined area of the one image and each matched area of the next image, and calculates an evaluation value of the difference of the pixel value.

3. The moving object detection apparatus according to claim 2,

wherein said calculation means selects a maximum of the evaluation value as a pair of the predetermined area of the one image and corresponding area of the next image, and calculates a confidence value of the maximum of the evaluation value using the evaluation value between the predetermined area of the one image and a non-corresponding area of the next image.

4. The moving object detection apparatus according to claim 3,

wherein said calculation means calculates a relative position between the predetermined area of the one image and the corresponding area of the next image as the moving vector, and calculates a value of normal distribution using the relative position and the confidence degree.

5. The moving object detection apparatus according to claim 4,

wherein said background area detection means evaluates the moving vector and the value of normal distribution of each predetermined area of the one image for a moving vector space, and detects one moving vector whose value of normal distribution is maximum from voted result on the moving vector space.

6. The moving object detection apparatus according to claim 5,

wherein said move area detection means shifts the next image as the one moving vector detected by said background area detection means, and subtracts the one image from the shifted next image to detect first overlap part of the moving area.

7. The moving object detection apparatus according to claim 6,

wherein said moving area detection means shifts the previous image as the one moving vector detected by said background area detection means, and subtracts the one image from shifted previous image to detect second overlap part of the move area.

8. The moving object detection apparatus according to claim 7,

wherein said moving area detection means produces the first overlap part and the second overlap part to detect the moving area from the one image.

9. A method for detecting a moving object, comprising the steps of:

inputting a plurality of images including a background and a moving object in time series to an image input section, wherein the image input section is moving in parallel with movement of the background in the images;

dividing one image into units of predetermined area;

calculating an evaluation value between the predetermined area of the one image and calculating an evaluation value between the predetermined area of the one image and each neighboring area of next image;

calculating a moving vector between the predetermined area of the one image and corresponding area of the next image, the evaluation value of the corresponding area is maximum among the evaluation value of the each neighboring area of the next image;

calculating a confidence value of the moving vector for each unit of predetermined area on the one image, the confidence value is a distance between the corresponding area and the neighboring area whose evaluation value is above a threshold based on the maximum on the next image;

calculating a value of normal distribution using the moving vector and the confidence value for each unit of predetermined area on the one image;

adding the value of normal distribution to a position of the moving vector on a moving vector space for each unit of predetermined area on the one image;

detecting one moving vector whose value of normal distribution is maximum on the moving vector space as a moving vector of the background area;

shifting the next image as the one moving vector; and

subtracting the shifted image from the one image to detect the moving area.

10. A computer readable memory containing computer readable instructions, comprising:

an instruction means for causing a computer to receive a plurality of images including a background and a moving object in time series by an image input section, wherein the image input section is moving in parallel with movement of the background in the images;

an instruction means for causing a computer to divide one image into units of predetermined area;

an instruction means for causing a computer to calculate an evaluation value between the predetermined area of the one image and each neighboring area of next image;

an instruction means for causing a computer to calculate a moving vector between the predetermined area of the one image and corresponding area of the next image, the evaluation value of the corresponding area is maximum among the evaluation values of the each neighboring area of the next image;

an instruction means for causing a computer to calculate a confidence value of the moving vector for each unit of predetermined area on the one image, the confidence value is a distance between the corresponding area and the neighboring area whose evaluation value is above a threshold based on the maximum on the next image;

an instruction means for causing a computer to calculate a value of normal distribution using the moving vector and the confidence value for each unit of predetermined area on the one image;

an instruction means for causing a computer to add the value of normal distribution to a position of the moving vector on a moving vector space for each unit of predetermined area on the one image;

an instruction means for causing a computer to detect one moving vector whose value of normal distribution is maximum on the moving vector space as a moving vector of the background area;

an instruction means for causing a computer to shift the next image as the one moving vector; and

an instruction means for causing a computer to subtract the shifted image from the one image to detect the moving area.