Structure-guided automatic learning for image feature enhancement

• US 6,507,675 B1
• Filed: 03/23/2001
• Issued: 01/14/2003
• Est. Priority Date: 03/23/2001
• Status: Expired due to Term

• Current Assignee: Leica Microsystems Gmbh (Danaher Corporation)
• Original Assignee: Shih-Jong J. Lee
• Inventors: Oh, Seho, Huang, Chi-Chou, Lee, Shih-Jong J.
• Primary Examiner(s): Lee, Thomas D.
• Assistant Examiner(s): BRINICH, STEPHEN M
• Application Number:
  US09/815,466
• Time in Patent Office: 662 Days
• Field of Search: 382/266-269, 382/155-161
• US Class Current: 382/266
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 5/30 Erosion or dilatation, e.g. thinning
• CPC Class Codes: G06T 5/70 Denoising; Smoothing
• CPC Class Codes: G06V 10/44 Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
• CPC Class Codes: G06V 20/69 Microscopic objects, e.g. biological cells or cellular parts

Abstract

A structure-guided automatic learning system for image feature enhancement uses a learning image together with an application domain structure and detection target specification to produce a feature enhancement image processing recipe. An enhancement goodness measure is used to select between alternatives in the learning process. The feature enhancement recipe is used in an application module to process input images and produce a feature enhanced image output. Calipers are used for application domain structure and detection target specification. To unify the processing steps for all caliper specifications, a non-directional box caliper defined region such as a circle caliper or an arc caliper or other connected structures can be converted into a directional box caliper defined region so that a directional box caliper based feature enhancement method can be applied. The process can be inverted to convert a converted directional box caliper region back to the original format.

Claims

1. A caliper based structure-guided processing system comprising:

(a) Means to receive an image input;

(b) Means to receive a non-directional box caliper input;

(c) means to convert said non-directional box caliper input to a directional box caliper output;

(d) means to use the directional box caliper to perform structure-guided processing on the image input to produce a structure-guided processing output.

2. The caliper based structure-guided processing system of claim 1 wherein the non-directional box caliper is a circle caliper.

3. The caliper based structure-guided processing system of claim 1 wherein the non-directional box caliper is an arc caliper.

4. The caliper based structure-guided processing system of claim 2 further comprises a means to inverse convert the structure-guided processing output to the original image format defined by a circle caliper.

5. The caliper based structure-guided processing system of claim 3 further comprises a means to inverse convert the structure-guided processing output to the original image format defined by an arc caliper.

6. A structure-guided edge enhancement learning system comprising:

(a) a learning image input;

(b) an application domain structure and detection target specification input;

(c) edge enhancement processing parameter learning along the direction of edge structure having a first processing recipe output;

(d) edge enhancement processing parameter learning along the direction orthogonal to the edge structure having a second processing recipe output.

7. The structure-guided edge enhancement learning system of claim 6 further comprising the step of having an enhancement goodness measure output.

8. The system of claim 6 wherein the application domain structure and detection target are specified using a caliper method.

9. The system of claim 6 wherein the edge enhancement processing is an increasing idempotent processing sequence of morphological opening and closing operations.

10. The system of claim 9 wherein the morphological opening and closing operations include a directional elongated structuring element.

11. The system of claim 6 wherein the processing parameter is selected from the group consisting of the starting size of the structuring element, the increment size of the structuring element for each iteration, and the ending size of the structuring element.

12. The system of claim 7 wherein the processing parameter learning determines the structuring element size having the highest enhancement goodness measure.

13. The system of claim 7 wherein the enhancement goodness measure comprises

(a) a directional contrast measurement with a contrast image output;

(b) projection of contrast image in orthogonal direction within the caliper region to generate projection profiles;

(c) output statistics of the profile as the enhancement goodness measure.

14. The system of claim 12 wherein the output statistics is a maximum value.

15. The system of claim 6 further comprises an edge enhancement application module that enhances the edges of the image input using the processing recipes.

16. The system of claim 7 further comprises a general region enhancement learning process comprising:

(a) determine the bright enhancement goodness measure by bright edge enhancement learning;

(b) determine the dark enhancement goodness measure by dark edge enhancement learning;

(c) select bright edge enhancement process if the bright enhancement goodness measure is larger than or equal to the dark enhancement goodness measure;

(d) select dark edge enhancement process if the dark enhancement goodness measure is larger than the bright enhancement goodness measure.

17. A structure-guided automatic learning system for image feature enhancement comprising:

(a) means to receive a learning image input;

(b) means to receive an application domain structure and detection target specification input that are specified using a caliper method selected from the set consisting of a directional box caliper, a circle caliper, and an arc caliper;

(c) a structure-guided feature enhancement learning module processes the learning image using the structure and detection target specification having a feature enhancement processing recipe output.

18. A structure-guided contrast extraction system comprising:

(a) means to receive an input image;

(b) means to receive an application domain structure and detection target specification input that are specified using a caliper method;

(c) a structure-guided contrast extraction module processes the input image using the structure and target detection specification having a contrast extracted image output, wherein the structure-guided contrast extraction module further comprises;

i. means for closing of the input image by a structuring element determined by a caliper input;

ii. means for opening of the input image by a structuring element determined by a caliper input;

iii. means to output the difference between the closing result and the opening result.

19. A structure-guided line enhancement learning system comprising:

(a) a learning image input;

(b) an application domain structure and detection target specification input;

(c) line enhancement processing parameter learning along the direction determined by the application domain structure and detection target specification input having a processing recipe output, wherein the line enhancement process is an increasing idempotent processing sequence of morphological opening and closing operations.

20. The system of claim 19 wherein the morphological opening and closing operations include a directional elongated structuring element.

21. A structure-guided line enhancement learning system comprising:

(a) a learning image input;

(b) an application domain structure and detection target specification input;

(c) line enhancement processing parameter learning along the direction determined by the application domain structure and detection target specification input having a processing recipe output, wherein said processing parameter is selected from the group consisting of starting size of the structuring element, the increment size of the structuring element for each iteration, and the ending size of the structuring element.

22. A structure-guided line enhancement learning system comprising:

(a) a learning image input;

(b) an application domain structure and detection target specification input;

(c) line enhancement processing parameter learning along the direction determined by the application domain structure and detection target specification input having a processing recipe output and an enhancement goodness measure output, wherein the processing parameter learning determines the structuring element size having the highest enhancement goodness measure.

23. The system of claim 22 wherein the enhancement goodness measure comprises

(a) a directional contrast measurement with a contrast image output;

(b) projection of contrast image in orthogonal direction within the caliper region to generate projection profiles;

(c) output statistics for the profile as the enhancement goodness measure.

24. The system of claim 23 wherein the statistics is the maximum value of the enhancement goodness measure.