System and method for ranking and extracting salient contours for target recognition

• US 5,566,246 A
• Filed: 06/07/1995
• Issued: 10/15/1996
• Est. Priority Date: 03/28/1991
• Status: Expired due to Term

• Current Assignee: Texas Instruments, Inc.
• Original Assignee: Texas Instruments, Inc.
• Inventors: Rao, Kashi
• Primary Examiner(s): Mancuso, Joseph
• Application Number:
  US08/483,351
• Time in Patent Office: 496 Days
• Field of Search: 382/100, 382/154, 382/190, 382/195, 382/199, 382/282, 382/194, 356/16, 348/42, 348/46, 348/51, 348/47
• US Class Current: 382/154
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10032 Satellite or aerial image; Remote sensing
• CPC Class Codes: G06T 2207/10048 Infrared image
• CPC Class Codes: G06T 2207/20016 Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
• CPC Class Codes: G06T 2207/30212 Military
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/181 involving edge growing; involving edge linking
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
• CPC Class Codes: G06V 10/44 Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
• CPC Class Codes: G06V 10/443 by matching or filtering
• CPC Class Codes: G06V 10/462 Salient features, e.g. scale invariant feature transforms [SIFT]

Abstract

A system and method for obtaining salient contours from two-dimensional images acquired by a sensor is disclosed which process the two-dimensional images with an edge detector to produce edgels from each of the images, link the edgels into lists known as contours, compute a saliency value for each of the contours, rank the contours in decreasing order of saliency, and select certain ones of the ranked contours based on the requirements of a particular vision application.

Other devices, systems and methods are also disclosed.

Claims

1. A computer system for object matching in stereo comprising:

stereo structure for capturing and delivering a plurality of stereo image frames, each of said stereo image frames including a plurality of contours, each of said contours including edgels, said plurality of stereo image frames having associated therewith at least one three-dimensional parameter;

a vision system including a linker which links said edgels into said contours, said vision system connected to said stereo structure for receiving said plurality of stereo image frames from said stereo structure, and including a processing unit for determining a saliency for each of said contours and ranking said plurality of contours in order of saliency and extracting the most salient contours from said stereo image frames; and

a matcher connected to said vision system such that said vision system provides said most salient contours to said matcher whereby said matcher matches said most salient contours to estimate at least one three-dimensional parameter of said plurality of stereo image frames.

2. The computer system of claim 1 wherein said vision system comprises a plurality of contour extractors operating in parallel.

3. The computer system of claim 1, wherein said processing unit comprises a computing unit for establishing a saliency value for each of said contours and determining a plurality of saliency measures for each saliency value.

4. The computer system of claim 3, wherein said computing unit determines said saliency value, S, as follows:

##EQU5## where W₁ is the weight for the length saliency measure, W_s is the weight for the smoothness saliency measure, and W_c is the weight for the contrast saliency measure.

5. The computer system of claim 3, wherein at least one of said plurality of saliency measures determined by said computing unit includes a smoothness saliency measure, S_s, calculated as:

##EQU6## where Δ

k is the average change of curvature of the contour, Δ

k_min is the smallest average change of curvature for all contours, and Δ

k_max is the largest average change of curvature for all contours.

6. The computer system of claim 5 wherein said average change of curvature, Dk, is calculated by computing curvature at each edgel within said contour, determining any change of curvature, and computing an average of said any change of curvature.

7. The computer system of claim 6, wherein said determining any change of curvature is accomplished by convolving said computed curvature for each edgel using a difference operator.

8. The computer system of claim 3, wherein at least one of said plurality of saliency measures determined by said computing unit includes a contrast saliency measure, S_c, calculated as:

##EQU7## where C is the average contrast of the contour, C_min is the smallest average contrast for all contours, and C_max is the largest average contrast for all contours.

9. The computer system of claim 3, wherein at least one of said plurality of saliency measures determined by said computing unit includes a length saliency measure, S_l, calculated as:

##EQU8## where L, the length of the contour, is defined as the number of edgels in that contour, L_min is the length of the shortest contour, and L_max is the length of the longest contour.

10. The computer system of claim 1 where said matcher comprises an object matching system connected to receive preselected ones of said processed contours and match them to known contours of predetermined objects.

11. A vision system for object matching in stereo comprising:

stereo structure for capturing and delivering a plurality of stereo image frames, each of said stereo image frames including a plurality of contours, each of said contours including edgels, said plurality of stereo image frames having associated therewith at least one three-dimensional parameter;

a vision system including a linker which links said edgels into said contours, said vision system connected to said stereo structure for receiving said plurality of stereo image frames from said stereo structure, and including a processing unit for determining a saliency for each of said contours and ranking said plurality of contours in order of saliency and extracting the most salient contours from said stereo image frames; and

a matcher connected to said vision system such that said vision system provides said most salient contours to said matcher whereby said matcher matches said most salient contours to estimate at least one three-dimensional parameter of said plurality of stereo image frames.