Methods and apparatus for centroid based object segmentation in object recognition-type image processing system

• US 5,825,922 A
• Filed: 08/28/1996
• Issued: 10/20/1998
• Est. Priority Date: 11/12/1993
• Status: Expired due to Term

• Current Assignee: Laser Signals LLC (Intellectual Ventures LLC)
• Original Assignee: Martin Marietta Corporation
• Inventors: Pearson, William A., Benton, Richard W., Lofgren, John D.
• Primary Examiner(s): Couso, Jose L.
• Assistant Examiner(s): Bella, Matthew C.
• Application Number:
  US08/704,810
• Time in Patent Office: 783 Days
• Field of Search: 382/199, 382/197, 382/202, 382/205, 382/209, 382/217, 382/218
• US Class Current: 382/199
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10048 Infrared image
• CPC Class Codes: G06T 2207/30212 Military
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/66 of image moments or centre of gravity

Abstract

A digital signal processing system which uses an input image including a plurality of pixels to establish boundaries of objects in the input image. Each of the pixels has a gradient value and the object includes a centroid. Using a segmentation process, the digital signal processing system decides on a perimeter surrounding the object in the input image and then repeatedly searches for a pixel having the highest gradient value and marks that pixel having the highest gradient value as an edge pixel of the object.

Claims

1. A method of determining whether an image pixel of an input image having a plurality of image pixels including edge pixels corresponds to part of an object using a set of masks, each mask including a matrix of bits, wherein each image pixel of the plurality of image pixels has a gradient value and one of the image pixels of the plurality of image pixels corresponds to a centroid of the object, the method, performed in a signal processing system, comprising the steps of:

dividing the input image into four quadrants using the centroid, each of the quadrants having a corresponding set of masks;

identifying the edge pixels in the input image, each identifies edge pixel having a gradient value greater than a threshold gradient value; and

comparing each one of the sets of masks for each quadrant with a matrix of the image pixels within the same quadrant to

identify image pixels in the matrix that correspond to part of the object.

2. An image processing system for determining whether an image pixel of an input image having a plurality of image pixels corresponds to part of an object using a plurality of masks, each mask consisting of a matrix of bits, wherein each image pixel has a gradient value and one of the image pixels of the input image corresponds to a centroid of the object, the system comprising:

means for dividing the input image into four quadrants using the centroid, each of the quadrants having a corresponding set of masks;

means for identifying edge pixels among the image pixels of the input image, each edge pixel having a gradient value greater than a threshold gradient value; and

means for comparing each one of the masks for each quadrant with a matrix of image pixels within the same quadrant to

identify image pixels in the matrix that correspond to part of the object.

3. A method of establishing the boundaries of an object in an input image including a plurality of image pixels having gradient values of a predetermined scale, wherein the object includes a centroid, the method, performed by a signal processing system, comprising the steps of:

deciding on a perimeter surrounding the object in the input image;

repeatedly searching from an image pixel corresponding to the centroid of the object to the perimeter for an image pixel having a highest gradient value of the image pixels searched; and

marking, as an edge pixel, the image pixel having the highest gradient value for each of the repeated searches when said highest gradient value exceeds a predetermined threshold.

4. An image processing system for establishing the boundaries of an object in an input image including a plurality of image pixels having gradient values of a predetermined scale, wherein the object includes a centroid, the system comprising:

means for deciding on a perimeter surrounding the object in the input image;

means for repeatedly searching from an image pixel corresponding to the centroid of the object to the perimeter for an image pixel having a highest gradient value of the image pixels searched; and

means for marking, as an edge pixel, the image pixel having the highest gradient value for each of the repeated searches when said highest gradient value exceeds a predetermined threshold.