Method and system for interactive segmentation using texture and intensity cues

1. A method for representing image data as a graph, comprising:

receiving image data;

receiving one or more seed points within the image data;

computing image features based on the received image data and the seed points;

representing the image data as a graph wherein each pixel of the image data is represented as a node and edges connect nodes representative of proximate pixels of the image data; and

establishing edge weights for the edges of the graph using a classifier that takes as input, one or more of the computed image features, wherein each of the above steps is performed by a computer system.

2. The method of claim 1 , wherein the image features are computed using Law masks, dyadic Gabor filter banks, wavelet transforms, quadrature mirror filters, discrete cosine transforms, or eigenfilters.

3. The method of claim 1 , wherein computed image features are preprocessed by rectification and smoothing prior to the representing of the image data as a graph.

4. The method of claim 1 , wherein one or more of the computed image features are selected for use based on separability.

5. The method of claim 1 , wherein the classifier is a supervised classifier that is trained using Support Vector Machine (SVM).

6. The method of claim 1 , wherein the classifier is a discriminative classifier or a classifier that uses membership or fit to a generative model.

7. The method of claim 1 , wherein the classifier is trained using membership or fit to a Gaussian Model or a Gaussian Mixture Model.

8. The method of claim 1 , additionally comprising performing graph-based segmentation on the graph with established edge weights representing the image data using the received seed points.

9. The method of claim 1 , additionally comprising performing random walker segmentation on the graph with established edge weights representing the image data using the received seed points.

10. The method of claim 1 , additionally comprising generating easily separable image data based on the graph with established edge weights representing the image data.

11. A method for representing image data as a graph, comprising:

receiving image data

computing one or more image intensity features and one or more image texture features based on the received image data;

representing the image data as a graph wherein each pixel of the image data is represented as a node and edges connect nodes representative of proximate pixels of the image data; and

establishing edge weights for the edges of the graph based on at least one of the image intensity features and at least one of the image texture features,

wherein each of the above steps is performed by a computer system.

12. The method of claim 11 , wherein establishing edge weights for the edges of the graph based on at least one of the image intensity features and at least one of the image texture features includes training and using a classifier that takes as input, one or more of the computed image intensity features and one or more of the computed image texture features and outputs data that is used as edge weights.

13. The method of claim 11 , wherein the image texture features are computed using Law masks, dyadic Gabor filter banks, wavelet transforms, quadrature mirror filters, discrete cosine transforms, or eigenfilters.

14. The method of claim 11 , wherein computed image intensity and texture features are preprocessed by rectification and smoothing prior to the representing of the image data as a graph.

15. The method of claim 11 , wherein one or more of the computed image intensity and texture features are selected for use based on separability.

16. The method of claim 12 , wherein the classifier is a supervised classifier that is trained using Support Vector Machine (SVM).

17. The method of claim 12 , wherein the classifier is a discriminative classifier or a classifier that uses membership or fit to a generative model.

18. The method of claim 12 , wherein the classifier is trained using membership or fit to a Gaussian Model or a Gaussian Mixture Model.

19. The method of claim 11 , additionally comprising performing graph-based segmentation on the graph with established edge weights representing the image data using the received seed points.

20. The method of claim 11 , additionally comprising performing random walker segmentation on the graph with established edge weights representing the image data using the received seed points.

21. The method of claim 11 , additionally comprising generating easily separable image data based on the graph with established edge weights representing the image data.

22. A method for segmenting an image, comprising:

receiving an image;

receiving one or more seed points within the image;

computing one or more image intensity features and one or more image texture features based on the received image data;

representing the image as a graph wherein each pixel of the image is represented as a node and edges connect nodes representative of proximate pixels of the image;

establishing edge weights for the edges of the graph based on at least one of the image intensity features and at least one of the image texture features; and

segmenting the graph representation of the image including established edge weights using graph-based segmentation and the received seed points, wherein each of the above steps is performed by a computer system.

23. The method of claim 22 , wherein establishing edge weights for the edges of the graph based on at least one of the image intensity features and at least one of the image texture features includes training and using a classifier that takes as input, one or more of the computed image intensity features and one or more of the image texture features and outputs data that is used as edge weights.