System and method for edge detection of an image

• US 20040146201A1
• Filed: 01/27/2003
• Published: 07/29/2004
• Est. Priority Date: 01/27/2003
• Status: Active Grant

• Current Assignee: Boston Scientific Scimed (Boston Scientific Corporation)
• Original Assignee: Scimed Life Systems Incorporated (Boston Scientific Corporation)
• Inventors: Sathyanarayana, Shashidhar
• Granted Patent: US 7,502,525 B2
• Time in Patent Office: Days
• US Class Current: 382/199
• CPC Class Codes: G06T 2207/10072 Tomographic images
• CPC Class Codes: G06T 2207/10132 Ultrasound image
• CPC Class Codes: G06T 2207/30004 Biomedical image processing
• CPC Class Codes: G06T 7/13 Edge detection
• CPC Class Codes: G06T 7/194 involving foreground-background segmentation
• 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

Abstract

The improved system and method detects edges in an image defined by a data set. In one aspect, a plurality of arrays of pattern recognition classifiers are populated throughout an image. Each array may yield a value which indicates whether or not the array is on or near an edge to be detected. The edge may be interpolated between the arrays that are on or near an edge to be detected or extrapolated after an array on or near the edge. This edge detection method may result in more accurate edge detection.

Claims

1. A method for detecting an edge within an image defined by a data set, comprising:

identifying a plurality of distinct classes, wherein an interface is defined between the plurality of distinct classes, and further wherein the edge is defined by the interface;

populating the data set with a plurality of arrays of pattern recognition classifiers;

identifying each array that exhibits a predetermined response pattern indicating an increased likelihood that the array covers the edge; and

indicating the edge between the arrays that exhibit the predetermined response pattern.

2. The method of claim 1, wherein the step of indicating the edge includes interpolating between the arrays that exhibit the predetermined response pattern.

3. The method of claim 1, wherein the step of indicating the edge includes exterpolating from an array that exhibits the predetermined response pattern.

4. The method of claim 1, wherein each pattern recognition classifier yields a feature vector.

5. The method of claim 2, wherein interpolating the arrays comprises drawing a smooth curve through the centers of the arrays that exhibit the predetermined response pattern.

6. The method of claim 1, wherein populating the data set includes using a rectangular array of pattern recognition classifiers.

7. The method of claim 6, wherein populating the data set includes using a rectangular array of pattern recognition classifiers, each rectangular array comprising at least two pattern recognition classifiers.

8. The method of claim 6, wherein populating the data set includes using a rectangular array of pattern recognition classifiers, each rectangular array comprising at least two rows and two columns of pattern recognition classifiers.

9. The method of claim 1, wherein populating the data set includes using a circular array of pattern recognition classifiers.

10. The method of claim 1, wherein the identifying step includes identifying arrays whose predetermined response pattern indicate that the number of pattern recognition classifiers lying on one side of the edge is approximately equal to the number of pattern recognition classifiers lying on the other side of the edge.

11. The method of claim 1, wherein the identifying step includes identifying arrays whose predetermined response pattern indicate that the number of pattern recognition classifiers lying on or substantially close to the edge exceeds a threshold.