Pattern recognition system

• US 4,769,850 A
• Filed: 10/10/1986
• Issued: 09/06/1988
• Est. Priority Date: 11/26/1985
• Status: Expired due to Term

• Current Assignee: Cisco Technology, Inc. (Cisco Systems, Inc.)
• Original Assignee: International Business Machines Corporation
• Inventors: Itoh, Shoji, Suzuki, Hideo
• Primary Examiner(s): Groody, James J.
• Assistant Examiner(s): Parker, Michael D.
• Application Number:
  US06/917,583
• Time in Patent Office: 697 Days
• Field of Search: 382/25, 382/28, 382/16, 382/22, 382/41, 382/19, 364/725, 364/721, 382/6
• US Class Current: 382/132
• CPC Class Codes: G06V 10/443 by matching or filtering

Abstract

The present invention relates to apparatus for separately extracting circular and linear components in a complex image, the apparatus comprising: (a) a circular filter means for covering a portion of the complex image by filter blocks, wherein the filter blocks comprise a center block and at least one set of blocks angularly disposed to encircle the center block; (b) a first calculator element for computing average image concentration value for each of a plurality of directions with said circular filter means covering a subject portion of the complex image, wherein the concentration value for one directional component is a maximum; (c) a second calculator element for computing a first feature S in which the values of the directional components are evaluated with maximum directional component being enhanced; (d) a third calculator element for computing a second feature value S'"'"' in which the values of the directional components are evaluated with the maximum directional component being weakened; and (e) a mode selector for selectively applying (i) the first feature value S to detect linear image components and (ii) the second feature value S'"'"' to detect circular image components.

Claims

1. A pattern recognition system for extracting circular or linear components separately from a complex image including circular and linear components, said system comprising:

means for calculating feature values including (a) a directional multiple circular window filter which extracts a feature value for a given position of the complex image, (b) means for scanning various positions of the complex image with said filter and (c) means for enhancing at least one selected directional component of feature values extracted with said filter; and

means for detecting said circular or linear components based on said feature values calculated by said calculating means.

2. A pattern recognition system as in claim (1), wherein said filter comprises a triple circular window filter for said circular components and wherein said calculation means enhances the directional components other than the maximum directional component of said features extracted with said filter.

3. A pattern recognition system as in claim (2), wherein said detecting means searches the maximum value of said feature values and detects a circular component within a predetermined area centered at the position corresponding to said maximum feature value.

4. A pattern recognition system as in claim (3), wherein said circular components are shadows of candidates of lung cancers in a chest radiograph.

5. A pattern recognition system as in claim (1), wherein said calculating means uses a double circular window filter for said linear components and enhances the maximum directional component of said features extracted with said filter.

6. A pattern recognition system as in claim (5), wherein said detecting means binary-digitizes said feature values with a predetermined threshold.

7. A pattern recognition system as in claim (6), wherein said linear components are shadows of blood vessels in a chest radiograph.

8. A pattern recognition system as in claim (1), wherein said multiple circular window filter includes at least one pair of concentric circles, each annular region enclosed by a pair of concentric circles being equally divided into a plurality of segments;

wherein the average concentration value of the pixels located within each of said segments and the innermost circle represent a feature value.

9. Apparatus for separately extracting circular and linear components in a complex image, the apparatus comprising:

circular filter means for covering a portion of the complex image by filter blocks, wherein the filter blocks comprise a center block and at least one set of blocks angularly disposed to encircle the center block;

means for computing average image concentration value for each of a plurality of directions with said circular filter means covering a subject portion of the complex image, wherein the concentration value for one directional component is a maximum;

means for computing a first feature value S in which the values of the directional components are evaluated with the maximum directional component being enhanced;

means for computing a second feature value S'"'"' in which the values of the directional components are evaluated with the maximum directional component being weakened; and

means for selectively applying (a) the first feature value S to detect linear image components and (b) the second feature value S'"'"' to detect circular image components.

10. Apparatus as in claim 9, wherein said circular filter means includes a first filter having two concentric sets of blocks, each set encircling the center block at a different respective radius, and each set having a plurality of pairs of blocks symmetrically disposed relative to the center block with a corresponding directional component;

wherein each pair i of blocks in the inner set has an average image concentration value of W_Xi ; and

wherein each pair j of blocks in the outer set has an average image concentration value of W_Bi ; and

wherein said S value computing means comprises means for computing S as either;

1. If W_T ≦

Max W_X ≦

Max W_B ;

##EQU3##

12. Apparatus as in claim 10, wherein said filter means includes a second filter having a cental block and a set of blocks encircling the center block thereof, and a third filter having a center block and a set of blocks encircling the center block thereof, wherein the blocks of said second filter are all similar in size and smaller in size than the blocks of said third filter which are all similar in size;

wherein the encircling set of blocks for said second filter and said third filter, each has a plurality of pairs of blocks symmetric relative to the respective center block with a corresponding directional component;

wherein each ith pair of blocks of a respective set has an average image concentration value W_Xi and each respective set has a directional component with a maximum average image concentration value;

said S'"'"' value computing means comprising means for computing S'"'"' for said second filter and said third filter as;

##EQU4##

14. Apparatus as in claim 12, wherein said selective applying means includes:

means for selecting said S value computing means or said S'"'"' value computing means.

11. 2. Otherwise, S=0.

13. Apparatus as in claim 11, further comprising:

means for registering said filters to cover one portion of the complex image after another.

15. Apparatus as in claim 13, wherein each covered position and the feature value computed therefor, corresponds to a pixel and wherein the apparatus further comprises:

shadow detector means for selectively detecting circular components or linear components based selectively on either (a) S feature values for respective pixels in the complex image or (b) S'"'"' feature values for respective pixels in the complex image.