Automated and scalable object and feature extraction from imagery

1. A method for extracting feature data from very high resolution (VHR) remotely-sensed multispectral Earth imagery, comprising:

storing VHR remotely-sensed multispectral Earth image data in a file system;

providing a plurality of feature extraction modules, wherein each respective feature extraction module is operable to analyze the image data to generate feature data regarding the image data;

accessing the VHR remotely-sensed multispectral Earth image data in the file system with at least a first of the plurality of feature extraction modules, wherein the plurality of feature extraction modules are operable to read data from and write data to the file system using a common interface format;

executing the first feature extraction modules to access the file system using the common interface format and generate first feature data corresponding to the VHR remotely-sensed multispectral Earth image data;

receiving the first feature data at the file system from the first feature extraction module, wherein the first feature data is received from the first feature extraction module using the common interface format;

executing a second feature extraction module to access at least the first feature data in the file system using the common interface format, wherein first feature data generated from the execution of the first feature extraction module is used during the executing of the second feature extraction module to generate second feature data; and

receiving the second feature data at the file system from the second feature extraction module, wherein the second feature data is received from the second feature extraction module using the common interface format.

2. The method of claim 1 , wherein the VHR remotely-sensed multispectral Earth image data comprises spectral band data corresponding to at least 8 multispectral bands.

3. The method of claim 2 , wherein the multispectral bands collectively range from at least about 300 nanometers in wavelength to at least about 1100 nanometers in wavelength.

4. The method of claim 2 , wherein the image data further comprises a plurality of short-wavelength infrared (SWIR) bands.

5. The method of claim 4 , wherein the plurality of SWIR bands range from at least about 1100 nanometers in wavelength to at least about 2400 nanometers in wavelength.

6. The method of claim 1 , wherein feature data received from the plurality of feature extraction modules is stored in corresponding relation to the portion of the VHR remotely-sensed multispectral Earth image data from which the feature data was generated.

7. The method of claim 1 , wherein first feature extraction module and the second feature extraction module comprise different programming languages.

8. The method of claim 1 , wherein the first feature extraction module and the second feature extraction module are independently developed and each communicate with the file system using the common interface format.

9. The method of claim 1 , wherein the common interface format comprises a feature identifier and an image identifier.

10. The method of claim 9 , wherein the feature identifier corresponds with particular feature data generated by a given feature extraction module.

11. The method of claim 10 , wherein the image identifier comprises an indication of a portion of the image for which feature data is to be generated.

12. The method of claim 11 , further comprising:

determining a subset of the plurality of feature extraction modules, at least including the first feature extraction module and the second feature extraction module, required to generate feature data.

13. The method of claim 12 , wherein the determining further comprises ordering the execution of the first and second feature extraction modules.

14. The method of claim 12 , further comprising:

distributing an initialization command to each of the first and second feature extraction modules, wherein the initialization command comprises a feature identifier and an image identifier according to the common interface format.

15. The method of claim 1 , wherein at least one of the feature extraction modules generates feature data prior to receiving a request for the feature data from an image processing module.

16. The method of claim 1 , wherein at least one of the feature extraction modules generates feature data on demand in response to a request for the feature data from an image processing module.

17. The method of claim 13 , wherein the determining is at least partially based on a feature of interest to be identified from the image data as requested from an imaging processing module.

18. The method of claim 13 , wherein at least one of the feature extraction modules comprises a total variational (TV) feature extraction module.

19. The method of claim 18 , wherein the TV feature extraction module comprises a multispectral filtering module using a plurality of spectral bands of an image to generate a filtered image corresponding to the image.

20. The method of claim 19 , wherein the number of spectral bands used in the filtering operation comprises at least 8 multispectral bands.

21. The method of claim 20 , wherein the filtering operation comprises minimizing a function including at least a first term representing the difference between the image and the filtered image and at least a second term representing the spatial homogeneity of the filtered image.

22. The method of claim 21 , wherein the first term and the second term include a multivalued vector for each pixel representative of the image or filtered image, respectively, wherein the multivalued vector values include a component for each row, column, and multispectral band in the image.

23. The method of claim 22 , wherein at least one of the first term or the second term include an L1 norm.

24. The method of claim 23 , wherein the second term is weighted by a smoothing factor.

25. The method of claim 24 , wherein the minimization of the function is solved using a Split Bregman technique.

26. The method of claim 25 , wherein the Split Bregman technique includes Gauss-Seidel updates.

27. The method of claim 26 , wherein the minimization is solved by execution on a graphics processing unit (GPU).

28. The method of claim 1 , wherein at least one of the feature extraction modules comprises a Gabor wavelet feature extraction module.

29. The method of claim 28 , wherein the Gabor wavelet feature extraction modules is operative to generate feature data by:

applying a plurality of Gabor wavelet functions to an image, wherein each of the plurality of Gabor wavelet function includes a different orientation; and

computing at least one feature based on at least two of the plurality of Gabor wavelet functions.

30. The method of claim 29 , wherein the at least one feature comprises at least one of a sum of the absolute values of a first Gabor wavelet function of the plurality of Gabor wavelet functions and a second Gabor wavelet function of the plurality of Gabor wavelet functions, a sum of the magnitude of values of the product of each perpendicularly oriented Gabor wavelet function of the plurality of Gabor wavelet functions, and the difference of the maximum and minimum values of magnitude of all of the plurality of Gabor wavelet functions for all image data.

31. The method of claim 30 , wherein for each orientation of the plurality of Gabor wavelet functions, a plurality of different Gabor wavelet functions with a different scale are calculated.

32. The method of claim 31 , wherein the different scales comprise varied values for at least one of a wavelength of the Gabor wavelet function or an envelope of the Gabor wavelet function.

33. The method of claim 32 , wherein each of the plurality of Gabor wavelet functions are applied to each pixel of each multispectral band of an image comprising a plurality of multispectral bands.

34. The method of claim 33 , wherein the at least one feature is determined independently for each of the plurality of bands.

35. The method of claim 34 , wherein the at least one feature is determined with respect to all of the plurality of bands.