Image recognition template generation

• US 4,658,428 A
• Filed: 07/17/1985
• Issued: 04/14/1987
• Est. Priority Date: 07/17/1985
• Status: Expired due to Fees

• Current Assignee: Honeywell Incorporated (Honeywell International Inc.)
• Original Assignee: Honeywell Incorporated (Honeywell International Inc.)
• Inventors: Schnell, Robert J., Bedros, Renee
• Primary Examiner(s): NOT, DEFINED
• Assistant Examiner(s): NOT, DEFINED
• Application Number:
  US06/755,874
• Time in Patent Office: 636 Days
• Field of Search: 382/30, 382/45, 382/46, 382/44, 382/25, 382/28, 382/34, 382/54
• US Class Current: 382/216
• CPC Class Codes: G06T 3/02 Affine transformations for image registration G06T3/147; for image mosaicing G06T3/4038
• CPC Class Codes: G06V 10/7515 Shifting the patterns to accommodate for positional errors

Abstract

A method and system for generating additional template images containing an object of interest in a plurality of defined orientations from a single stored digitized pixel matrix template image frame is disclosed by which the object of interest may be translated and rotated within the template frame and thereby presented in a variety of predetermined positions.

Claims

1. A system for generating additional template images containing an object of interest, having an angle associated with the major axis of the best-fit ellipse around said object, in a plurality of defined orientations from a single digitized template image frame containing said object of interest comprising:

means for generating a first digitized pixel template of a frame containing an object of interest in the form of object and background pixels, wherein said frame is divided into four quadrants by perpendicular reference axes;

storage means for storing said first digitized template;

means for locating the centroid object pixel which is the pixel located closest to the geometric centroid of the object of interest in said first digitized template;

means for determining the angle of the major object axis, wherein said major object axis is defined as the major axis of the best-fit ellipse around said object of interest in said first digitized template;

means for translating said centroid pixel to a desired position in a quadrant of the frame with respect to said perpendicular reference axes dividing said frame into quadrants;

means for rotating said angle of said major object axis to a desired second angle in a quadrant of said template frame;

means for generating a second digitized pixel template of said frame containing said object of interest disposed in a second position with respect to said perpendicular reference axes of said frame;

means for enhancing the clarity of said object of interest in said second digitized pixel template.

2. The system of claim 1 including means for generating additional digitized pixel templates containing said object of interest from said first digitized pixel template wherein said major object axis is disposed at additional defined angles.

3. The system of claim 1 wherein eight such digitized pixel templates are generated at increments of 111/4°

intervals in a said quadrant of said frame.

4. The system of claim 1 wherein said digitized pixel templates are generated on a pixel-by-pixel basis.

5. The system of claim 1 wherein said means for enhancing the clarity of said object of interest in said second image frame further comprises means for accounting for inter-pixel gaps by determining whether pixels neighboring object pixels are also object pixels and if so incrementing the pixel scan to the midpoint location between adjacent object pixels rather than incrementing it to the next pixel.

6. A method for generating additional template images containing an object of interest, having an angle associated with the major axis of the best-fit ellipse around said object, in a plurality of defined orientations from a single digitized pixel template image frame containing said object of interest comprising the steps of:

generating a first digitized pixel template of the frame containing the object of interest, wherein said frame is divided into four quadrants by perpendicular reference axes;

storing said first digitized image in the form of binary pixel values representing background and object pixels;

determining the centroid object pixel which is the pixel located closest to the geometric centroid of the object of interest in said first template;

determining the angle of the major axis of the best-fit ellipse around said object of interest in said first template;

translating said centroid pixel to a desired position in a quadrant of the frame with respect to said perpendicular reference axes dividing said frame into quadrants and rotating said angle of said major object axis to a second determined angle in said quadrant of said template in which said object of interest is disposed;

generating a second digitized pixel template of said frame containing said object of interest disposed in said second position with respect to said perpendicular reference axes of said frame;

enhancing the clarity of said object of interest in said second digitized pixel template.

7. The method of claim 6 including the step of generating additional digitized pixel templates from said first digitized pixel template by changing the major object axis of said object of interest within the quadrant.

8. The method of claim 6 wherein the step of enhancing the clarity of said object of interest in said second image frame comprises accounting for inter-pixel gaps by determining whether pixels neighboring object pixels are also object pixels and if so incrementing the pixel scan to the midpoint location between adjacent object pixels rather than incrementing to the next pixel.