Method of directed pattern enhancement for flexible recognition

• US 8,014,590 B2
• Filed: 12/07/2005
• Issued: 09/06/2011
• Est. Priority Date: 12/07/2005
• Status: Active Grant

• Current Assignee: Leica Microsystems Gmbh (Danaher Corporation)
• Original Assignee: DRVision Technologies, LLC (Danaher Corporation)
• Inventors: Oh, Seho, Lee, Shih-Jong J.
• Primary Examiner(s): Mehta; Bhavesh
• Assistant Examiner(s): ENTEZARI, MICHELLE M
• Application Number:
  US11/301,292
• Publication Number:
  US 20070127834A1
• Time in Patent Office: 2,099 Days
• Field of Search: None
• US Class Current: 382/155
• CPC Class Codes: G06F 18/231 Hierarchical techniques, i.e. dividing or merging pattern sets so as to obtain a dendrogram
• CPC Class Codes: G06F 18/2415 based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
• CPC Class Codes: G06V 10/7625 Hierarchical techniques, i.e. dividing or merging patterns to obtain a tree-like representation; Dendograms
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects

Abstract

A directed pattern enhancement method receives a learning image and pattern enhancement directive. Pattern enhancement learning is performed using the learning image and the pattern enhancement directive to generate pattern enhancement recipe. An application image is received and a pattern enhancement application is performed using the application image and the pattern enhancement recipe to generate pattern enhanced image. A recognition thresholding is performed using the pattern enhanced image to generate recognition result.

The pattern enhancement directive consists of background directive, patterns to enhance directive, and patterns to suppress directive.

A partitioned modeling method receives an image region and performs feature extraction on the image region to generate characterization feature. A hierarchical partitioning is performed using the characterization feature to generate hierarchical partitions. A model generation is performed using the hierarchical partitions to generate partition model. The partitioned modeling further performs a partitioned matching step that matches an input point to the partition model to generate a matching score output.

A partition model update method receives a partition model and input data for model update. A partition model update is performed using the partition model and the data to generate an updated partition model.

Claims

1. A computer based partitioned modeling method for flexible recognition and extraction of patterns directed by human comprising the steps of:

a) Input a plurality of user directed specification of image regions from learning image to be processed including “

Patterns to enhance”

region, “

Patterns to suppress”

region and “

Background”

region;

b) Perform feature extraction on the image region having characterization feature output wherein the characterization feature output contains features derived from pre-processed grayscale intensity statistics of the image to be processed;

c) Perform a plurality of hierarchical clustering using the characterization feature for background pattern removal and remaining patterns for patterns to enhance and patterns to suppress having a plurality of hierarchical partitions outputs corresponding to background, patterns to enhance and patterns to suppress;

d) Perform model generation using the hierarchical partitions for each of the background, patterns to enhance, and patterns to suppress having their partition model outputs wherein the model generation further comprises a background learning step, a foreground pattern extraction step and a foreground pattern learning step and wherein the partition model outputs are used for partitioned matching for recognition and extraction of patterns.

2. The partitioned modeling method of claim 1 wherein the hierarchical partitioning method generates multiple points pre-clusters at the bottom of the hierarchy.

3. The partitioned modeling method of claim 1 wherein the model generation method estimates the model parameters for the partitions from the lowest level to the highest level using incremental update rule.

4. The partitioned modeling method of claim 1 further comprises a partitioned matching step that matches an input point to the partition model having a matching score output.

5. The partitioned matching step of claim 4 further comprises a filtering step to remove noise on matching score derived from processed image.

6. A computer based partition model update method for flexible recognition and extraction of patterns directed by human comprising the steps of:

a) Input a plurality of partition models consisting of background model, patterns to enhance model, and patterns to suppress model wherein the partition models are generated by user directed specification and features derived from grayscale intensity statistics of an image to be processed;

b) Input data for update of the background model, the patterns to enhance model, and the patterns to suppress model wherein the data is generated by features derived from grayscale intensity statistics of an image to be processed and wherein models are generated by a background learning step, a foreground pattern extraction step and a foreground pattern learning step;

c) Perform partition model update for each of the background model, the patterns to enhance model, and the patterns to suppress model using the partition model and the data having updated partition model outputs wherein the updated partition model outputs are used for partitioned matching for recognition and extraction of patterns.

7. The partition model update method of claim 6 wherein the partition model update performs progressive update comprising the steps of

a) Perform uniqueness detection using partition model and the data having a unique set and a similar set output;

b) Perform supplementary model generation using the unique set having supplementary model output;

c) Perform combined model generation using the supplementary model and the partition model having combined model output;

d) Perform progressive parameter update using the combined model and the a similar set having progressive updated partition model output.

8. The partition model update method of claim 6 wherein the partition model update performs regressive update comprising the steps of:

a) Perform regressive parameter update using partition model and the data derived, from processed image having regressive updated parameters output;

b) Perform regressive model update using the regressive updated parameters having regressive updated partition model output.