Dynamic image processing apparatus and method

• US 5,930,378 A
• Filed: 01/27/1997
• Issued: 07/27/1999
• Est. Priority Date: 01/31/1996
• Status: Expired due to Fees

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Kaneko, Toshimitsu, Kubota, Susumu
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): MEHTA, BHAVESH M
• Application Number:
  US08/791,493
• Time in Patent Office: 911 Days
• Field of Search: 382/205, 382/103, 382/281, 382/285, 382/107, 382/154, 382/255, 382/236, 348/285, 348/154-155, 348/699-700, 345/156, 345/435
• US Class Current: 382/107
• CPC Class Codes: G06T 7/20 Analysis of motion motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51

Abstract

In a dynamic image processing apparatus of the present invention, the focus of contraction of the optical flow of the background in the dynamic image is extracted. When a series of dynamic images are inputted, an optical flow extraction section extracts an optical flow representing motion vector of each block between two neighboring dynamic images. A Hough transform section votes each optical flow of the dynamic image to vote space by unit of block. The vote space comprises two planes and one circle positioned in parallel to each other. The dynamic image is included in one plane. The other plane is positioned symmetrically to the one plane with respect to a center position of the circle. The Hough transform section selects a block whose voted value is maximum in all blocks of the vote space. The selected block corresponds to the focus of contraction of the optical flow of background in the dynamic image.

Claims

1. A dynamic image processing apparatus, comprising:

image input means for inputting a series of dynamic images including pixels, wherein a focal length of said image input means is unknown;

optical flow extraction means for extracting an optical flow representing a motion vector for each pixel between two neighboring dynamic images;

memory means for storing a plurality of kernel patterns corresponding to a length of the optical flow, wherein the kernel patterns are fan-shaped patterns expanding toward a direction from a start point to an end point of the optical flow having a plurality of probability values stored along each concentric circle line in the fan-shaped patterns;

kernel pattern selecting means for selecting a kernel pattern corresponding to the length of the optical flow from said memory means;

adding means for adding the probability values of the selected kernel pattern for two planes and one circle to a block, wherein the two planes and one circle are positioned in parallel to each other, such that the dynamic image is included in the first of the two planes, and the second of the two planes is positioned symmetrically to the first plane with respect to a center position of the circle, and wherein the probability values added to the block correspond to a position of the optical flow in the two planes and one circle; and

block selecting means for selecting a block whose sum of the added probability values is a maximum of all blocks of the two planes and one circle after adding the probability values of all selected kernel patterns, wherein the selected block corresponds to a focus of contraction of the optical flow of background in the dynamic image.

2. The dynamic image processing apparatus according to claim 1,

wherein the kernel pattern is calculated by adding a likelihood of the optical flow of the background and a likelihood of the optical flow of a move object in the dynamic image, by inverting a logarithm of the addition result and by distributing the inverted logarithm toward a direction from the start point to the end point of the optical flow.

3. The dynamic image processing apparatus according to claim 1,

further including a camera parameter extraction means for projecting each optical flow of the dynamic image to a direction toward the focus of contraction where the focus of contraction is an infinite distant point, for calculating a histogram by unit of length of the projected optical flow, and for selecting one length of the histogram whose value is a maximum as a parallel moving parameter of a camera to input the dynamic image.

4. The dynamic image processing apparatus according to claim 3,

wherein said camera parameter extraction means includes means for dividing each optical flow of the dynamic image by a distance between the focus of contraction and the optical flow where the focus of contraction is included in the dynamic image, for projecting the divided optical flow to a direction toward the focus of contraction in the dynamic image, for calculating a histogram by unit of length of the projected optical flow, and for selecting one length of the histogram whose value is maximum as a zoom parameter of the camera.

5. The dynamic image processing apparatus according to claim 3, further including:

an optical flow correction means for predicting movement of each pixel of the dynamic image in accordance with the focus of contraction of the optical flow of the background;

means for comparing a predicted pixel with a corresponding pixel of a neighboring dynamic image; and

means for correcting the optical flow of the dynamic image if a value of the predicted pixel is different from a value of the corresponding pixel.

6. The dynamic image processing apparatus according to claim 5,

further including a move object separation means for calculating an average of the optical flows by unit of block in the dynamic image, and for deciding whether the block corresponds to the background or not by comparing the average with a threshold.

7. A method for processing a dynamic image, comprising the step of:

inputting a series of dynamic images including pixels, wherein a focal length of said image input means is unknown;

extracting an optical flow representing a motion vector for each pixel between two neighboring dynamic images;

storing a plurality of kernel patterns corresponding to a length of the optical flow, wherein the kernel patterns are fan-shaped patterns expanding toward a direction from a start point to an end point of the optical flow having a plurality of probability values stored along each concentric circle line in the fan-shaped patterns;

selecting a kernel pattern corresponding to the length of the optical flow from said memory means;

adding the probability values of the selected kernel pattern for two planes and one circle to a block, wherein the two planes and one circle are positioned in parallel to each other, such that the dynamic image is included in the first of the two planes, and the second of the two planes is positioned symmetrically to the first plane with respect to a center position of the circle, and wherein the probability values added to the block correspond to a position of the optical flow in the two planes and one circle; and

selecting a block whose sum of the added probability values is a maximum of all blocks of the two planes and one circle after adding the probability values of all selected kernel patterns, wherein the selected block corresponds to a focus of contraction of the optical flow of background in the dynamic image.

8. A computer readable memory containing computer-readable instructions to process a dynamic image, comprising:

inputting a series of dynamic images including pixels, wherein a focal length of said image input means is unknown;

instruction means for causing a computer to extract an optical flow representing a motion vector for each pixel between two neighboring dynamic images;

instruction means for causing a computer to store a plurality of kernel patterns corresponding to a length of the optical flow, wherein the kernel patterns are fan-shaped patterns expanding toward a direction from a start point to an end point of the optical flow having a plurality of probability values stored along each concentric circle line in the fan-shaped patterns;

instruction means for causing a computer to select a kernel pattern corresponding to the length of the optical flow from said memory means;

instruction means for causing a computer to add the probability values of the selected kernel pattern for two planes and one circle to a block, wherein the two planes and one circle are positioned in parallel to each other, such that the dynamic image is included in the first of the two planes, and the second of the two planes is positioned symmetrically to the first plane with respect to a center position of the circle, and wherein the probability values added to the block correspond to a position of the optical flow in the two planes and one circle; and

instruction means for causing a computer to select a block whose sum of the added probability values is a maximum of all blocks of the two planes and one circle after adding the probability values of all selected kernel patterns, wherein the selected block corresponds to a focus of contraction of the optical flow of background in the dynamic image.