Pattern recognition

• US 7,079,688 B1
• Filed: 05/23/2000
• Issued: 07/18/2006
• Est. Priority Date: 05/26/1999
• Status: Expired due to Fees

• Current Assignee: Siemens AG
• Original Assignee: Seimens AG
• Inventors: Deco, Gustavo, Schuermann, Bernd
• Primary Examiner(s): MARIAM, DANIEL G
• Application Number:
  US09/979,657
• Time in Patent Office: 2,247 Days
• Field of Search: 382/153, 382/159, 382/181, 382/190, 382/195, 382/209, 382/218, 382/224, 382/305, 707 1- 10
• US Class Current: 382/190
• CPC Class Codes: G06T 7/00 Image analysis
• CPC Class Codes: G06V 10/20 Image preprocessing

Abstract

In a pattern recognition system, a pattern which is to be recognized later is prescribed in a learning phase. This pattern is detected sequentially, that is to say the informative areas of the pattern are detected and, moreover, the spatial relationship between the areas is also stored. In the recognition phase, a hypothesis which indicates a presumed pattern and, furthermore, indicates where such further prominent areas should be located in the pattern to be recognized if the presumption is correct, is generated on the basis of the acquired data of a first area of a pattern to be recognized, and on the basis of the stored data. Thus, patterns are learned through their location information, on the one hand, and through their spatial relationship to each other, on the other hand, stored and then re-recognized. The application options of the present invention reside, for example, in robotics, text analysis and image analysis, but also in the field of medical technology (for example in automatic tumor detection in mammographies, etc).

Claims

1. A pattern recognition system for determining a match between a recognition area of a pattern to be recognized and a reference area of a reference pattern, comprising:

storage means for storing prescribed reference areas of a prescribed reference pattern, and prescribed relative spatial reference relationships, each prescribed relative spatial reference relationship describing relative spatial positions between two of the prescribed reference areas;

selection means for selecting a first recognition area from a pattern to be recognized;

comparing means for comparing the first recognition area with the prescribed reference areas to select a first reference area from the prescribed reference areas of the prescribed reference pattern;

first selection means for selecting a second reference area from the prescribed reference areas by using the first reference area and one of the prescribed relative spatial reference relationships which describes a relative position of the first reference area with respect to the second reference area, wherein the second reference area is formulated as a hypothesis for a presumed second recognition area;

second selection means for selecting a second recognition area from the pattern to be recognized by using the first recognition area and the one prescribed relative spatial reference relationship which describes the relative position of the first reference area with respect to the second reference area, and

determination means for determining a match between the second recognition area and the second selected reference area, wherein said determining of the match determines whether the hypothesis is verified.

2. The pattern recognition system as claimed in claim 1, wherein the pattern to be recognized is recognized as the prescribed reference pattern if the match exceeds a prescribable threshold, and otherwise the pattern to be recognized is not recognized as the prescribed reference pattern.

3. A method for determining a match between a reference area from a reference pattern and a recognition area from a pattern to be recognized, the reference pattern having prescribed reference areas and relative spatial reference relationships which in each case describe relative spatial positions between the reference areas, comprising:

selecting a first recognition area from the pattern to be recognized;

determining a first reference area from the prescribed reference areas by comparing the first recognition area with the prescribed reference areas;

determining a second reference area from the prescribed reference areas by using the first reference area and one of the relative spatial reference relationships which describes a relative position of the first reference area with respect to the second reference area, wherein the second reference area is formulated as a hypothesis for a presumed second recognition area;

determining a second recognition area from the pattern to be recognized by using the first recognition area and the one relative spatial reference relationship which describes the relative position of the first reference area with respect to the second reference area; and

determining a match between the second recognition area and the second selected reference area, wherein said determining of the match determines whether the hypothesis is verified.

4. The method as claimed in claim 3, wherein the pattern to be recognized is recognized as the reference pattern when a first prescribed measure of matching is achieved, and otherwise the pattern to be recognized is not recognized as the reference pattern.

5. The method as claimed in claim 4, further comprising, in case of non-recognition of the pattern to be recognized, repeating said determining of the second reference area, the second recognition area and the match using a different one of the relative spatial reference relationships.

6. The method as claimed in claim 5, wherein said repeating is carried out iteratively until the pattern to be recognized is recognized as the reference pattern, or a second prescribed measure of matching is reached.

7. The method as claimed in claim 3, wherein the hypothesis is generated by using an artificial neural network.

8. The method as claimed in claim 3, wherein the pattern to be recognized describes an object to be manipulated by a robot

9. The method as claimed in claim 3, wherein the pattern to be recognized is an image, generated by an imaging system, of a clinical picture.