Image processing system

• US 4,635,293 A
• Filed: 01/30/1985
• Issued: 01/06/1987
• Est. Priority Date: 02/24/1984
• Status: Expired due to Term

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Watanabe, Mutsumi
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): Mancuso, Joseph
• Application Number:
  US06/696,491
• Time in Patent Office: 706 Days
• Field of Search: 382/6, 382/34, 382/37, 382/44, 382/48, 358/101, 358/107, 364/414
• US Class Current: 382/130
• CPC Class Codes: G06T 5/80 Geometric correction

Abstract

According to the image self-alignment system of the present invention, desired partial image areas are extracted by a partial image area designator from a pair of to-be-aligned picture images used for, for example, medical purposes. Each partial image area is divided by an area divider into a plurality of image segments arranged in a matrix form. Each corresponding image segment pair is subjected to a first misalignment detection by using a first evaluation parameter called correlation coefficient data. Each image segment pair is then subjected to a second misalignment detection by using a second evaluation parameter including pixel density difference data for finer detection. The first and second picture images are automatically aligned based on highly reliable final misalignment data obtained by carrying out the first and second misalignment detections.

Claims

1. An image processing apparatus comprising:

first processor means for extracting a desired partial image area from each of a pair of first and second images to be subjected to position alignment;

second processor means for dividing each of said partial image areas into n x m image segments arranged in a matrix form, where n and m are positive integers;

third processor means for detecting, in a first mode, misalignment between each corresponding image segment pair of said first and second picture images by using a first evaluation parameter defined based upon correlation coefficient data, and for more accurately detecting, in a second mode, the misalignment between the same image segment pair of said first and second picture images by using a second evaluation parameter which is defined based upon pixel density difference data as well as the correlation coefficient data; and

fourth processor means for aligning said first and second picture images based on misalignment data obtained for all of said image segment pairs by said third processor means.

2. The apparatus according to claim 1, wherein said third processor means comprises:

first calculator means for sequentially calculating said correlation coefficient data, when said first detection mode is carried out, by shifting at least one of said corresponding image segment pair around an initial position on an image plane in such a manner that each shifting from said initial position makes the image segments before and after the shifting has some overlapping portions therebetween; and

first maximum detector means for determining maximum one piece of correlation data among plural pieces of said correlation data calculated for each pair of said corresponding image segments.

3. The apparatus according to claim 2, wherein said third processor means further comprises:

mode controller means, connected to said first maximum detector means, for electrically storing a threshold value of said correlation data and switching the detection mode from said first detection mode to said second detection mode when a variation in the correlation data between said image segment pairs, sequentially detected, decreases to fall below said threshold value.

4. The apparatus according to claim 3, wherein said third processor means further comprises:

second calculator means for summing the magnitudes of pixel density differences for each of said image segment pairs in said second detection mode to produce said pixel density difference data;

normalizer means, connected to said first calculator means and second calculator means, for normalizing said correlation data and pixel density difference data; and

third calculator means, connected to said normalizer means, for calculating said second evaluation parameter based on said normalized correlation data and pixel density difference data.

5. The apparatus according to claim 4, wherein said third processor means further comprises:

second maximum detector means for updating said second evaluation parameter to determine maximum one among a plurality of said second evaluation parameters calculated for each image segment pair;

mode termination means, coupled to said second maximum detector means, for electrically storing a threshold value of said second evaluation parameter and terminating said second detection mode performed for the associated image segment pair by said third processor means when said updated second evaluation parameter exceeds said threshold value; and

fourth calculator means for detecting center addresses of each of said image segment pairs put through said second detection mode and calculating a misalignment between said center addresses.

6. The apparatus according to claim 5, further comprising fifth processor means for sorting final correlation data and pixel density difference data, which are respectively obtained in said first and second detection modes for all of said image segment pairs of said partial image areas, for searching all of said image segment pairs for at least one image segment pair having both correlation data and pixel density difference data highly ranked and for calculating a misalignment vector between said at least one image segment pair.

7. The apparatus according to claim 6, wherein said fifth processor means comprises:

a first hierarchic memory for storing said plural pieces of final correlation data for all of said image segment pairs in order from the largest to the smallest;

a second hierarchic memory for storing said plural pieces of final pixel density difference data for all of said image segment pairs in order from the smallest to the largest; and

circuit means, coupled to said first and second hierarchic memories, for extracting at least one image segment pair with the correlation data and pixel density difference data both being in a specific high rank and for specifying misalignment vector data of said at least one image segment pair.

8. An image processing method comprising the steps of:

extracting a desired partial image area from a pair of first and second picture images to be aligned;

dividing each of said partial image areas into n x m image segments arranged in a matrix form, where n and m are positive integers;

sequentially calculating first evaluation function defined based on correlation data between a selected one of corresponding n x m image segment pairs while shifting one of the selected image segment pair on an image plane, thereby performing a first misalignment detection between said image segment pair;

terminating said first misalignment detection when a variation in said first evaluation function falls below a predetermined first threshold level, detecting a maximum of said first evaluation function and calculating a misalignment between the image segment pair at the time said maximum first evaluation function is detected;

sequentially calculating a second evaluation function including said correlation data and pixel density difference data between the selected image segment pair to thereby perform a second misalignment detection; and

calculating, when a maximum of the calculated second evaluation function exceeds a predetermined second threshold level, a misalignment quantity between the selected image segment pair at the time said maximum second evaluation function is detected and terminating said second misalignment detection, whereby final misalignment vector data for the selected image segment pair is obtained.

9. The method according to claim 8, further comprising a step of repeatedly performing said first and second misalignment detections for the image segment pairs other than said selected image segment pair, thus providing final misalignment vector data for all of said n x m image segment pairs included in said partial image areas of said first and second images.

10. The method according to claim 9, further comprising the steps of:

separately sorting plural pieces of final correlation data and pixel density difference data obtained for all of said image segment pairs;

searching said all of said image segment pairs for at least one image segment pair having those correlation data and pixel density difference data which are both in a predetermined high rank; and

specifying the misalignment quantity between said at least one image segment pair as highly reliable misalignment data and utilizing said specified misalignment quantity for aligning said first and second images.

11. The method according to claim 9, further comprising the steps of:

sorting said final correlation data in order from the largest to the smallest to form a hierarchic structure for said correlation data;

sorting said final pixel density difference data in order from the smallest to the largest to form a hierarchic structure for said pixel density difference data;

searching all of said image segment pairs for at least one image segment pair having those correlation data and pixel density difference data which are both in a predetermined high rank; and

specifying the misalignment quantity between said at least one image segment pair as highly reliable misalignment data and utilizing said specified misalignment quantity for aligning said first and second images.