System and method for detection of specularity in an image
View Patent ↗In a first exemplary embodiment of the present invention, an automated, computerized method is provided for processing an image. According to a feature of the present invention, the method comprises the steps of providing an image file depicting an image, in a computer memory, assembling a feature vector for the image file, the feature vector containing information regarding a likelihood that a selected region of the image file is specular, providing a classifier derived from a computer learning technique, computing a classification score for the selected region of the image file, as a function of the feature vector and the classifier and classifying the region as being specular, as a function of the classification score.
1. An automated, computerized method for processing an image, comprising the steps of:
providing an image file depicting an image, in a computer memory;
assembling a feature vector for the image file, the feature vector containing information regarding a likelihood that a selected region of the image file is specular, the selected region of the image being a type-C token having a single robust color measurement among contiguous pixels thereof;
providing a classifier derived from a computer learning technique;
computing a classification score for the selected region of the image file, as a function of the feature vector and the classifier;
classifying the region as being specular, as a function of the classification score; and
processing the image to remove specular characteristics from the region classified as being specular,
wherein the processing the image to remove specular characteristics from the region classified as being specular includes modifying gradients of all the pixels within the region classified as being specular to remove specular characteristics.
2. The method of claim 1 wherein the feature vector comprises Gabor filter features.
3. The method of claim 2 wherein the Gabor filter features are computed in each one of a set of different color spaces.
4. The method of claim 3 wherein the set of different color spaces comprises an RGB space, an SUV space, an HSV space, a log RGB space, and a color opponent space.
5. The method of claim 2 wherein the Gabor filter features are computed in each of different orientations and wavelengths.
6. The method of claim 5 wherein the different orientations and wavelengths include orientations of 0, 45, 90 and 135 degrees and wavelengths of 6 pixels and 4 pixels, respectively.
7. The method of claim 1 wherein the feature vector comprises cues characteristic of specularity in an image.
8. The method of claim 1 comprising the further step of selecting the region of the image file for classification upon a determination that the region is saturated.
9. The method of claim 1 comprising the further step of selecting the region of the image file for classification upon a determination that the region satisfies specularity conditions in a source dependent color space.
10. The method of claim 1 comprising the further step of, when the selected region is classified as being specular, setting an indication in a specular mask.
11. The method of claim 1 wherein the selected region of the image comprises a token.
12. The method of claim 1 , wherein the processing the image to remove specular characteristics from the region classified as being specular includes reintegrating a gradient representation of the image, with the modified gradients, to regenerate the image without specularities.
13. The method of claim 12 further comprising restoring texture the region after the image is regenerated.
14. A computer system which comprises:
a CPU; and
a memory storing an image file containing an image;
the CPU arranged and configured to execute a routine to assemble a feature vector for the image file, the feature vector containing information regarding a likelihood that a selected region of the image file is specular, the selected region of the image being a type-C token having a single robust color measurement among contiguous pixels thereof, provide a classifier derived from a computer learning technique, compute a classification score for the selected region of the image file, as a function of the feature vector and the classifier, classify the region as being specular, as a function of the classification score, and process the image to remove specular characteristics from the region classified as being specular,
wherein the processing the image to remove specular characteristics from the region classified as being specular includes modifying gradients of all the pixels within the region classified as being specular to remove specular characteristics.
15. The computer system of claim 14 , wherein the processing the image to remove specular characteristics from the region classified as being specular includes reintegrating a gradient representation of the image, with the modified gradients, to regenerate the image without specularities.
16. The computer system of claim 15 further comprising restoring texture the region after the image is regenerated.
17. A computer program product, disposed on a non-transitory computer readable medium, the product including computer executable process steps operable to control a computer to: assemble a feature vector for an image file, the feature vector containing information regarding a likelihood that a selected region of the image file is specular, the selected region of the image being a type-C token having a single robust color measurement among contiguous pixels thereof, provide a classifier derived from a computer learning technique, compute a classification score for the selected region of the image file, as a function of the feature vector and the classifier, classify the region as being specular, as a function of the classification score, and process the image to remove specular characteristics from the region classified as being specular,
wherein the processing the image to remove specular characteristics from the region classified as being specular includes modifying gradients of all the pixels within the region classified as being specular to remove specular characteristics.
18. The computer program product of claim 17 , wherein the feature vector comprises Gabor filter features.
19. The computer program product of claim 18 wherein the Gabor filter features are computed in each one of a set of different color spaces.
20. The computer program product of claim 19 wherein the set of different color spaces comprises an RGB space, an SUV space, an HSV space, a log RGB space, and a color opponent space.
21. The computer program product of claim 18 wherein the Gabor filter features are computed in each of different orientations and wavelengths.
22. The computer program product of claim 21 wherein the different orientations and wavelengths include orientations of 0, 45, 90 and 135 degrees and wavelengths of 6 pixels and 4 pixels, respectively.
23. The computer program product of claim 17 wherein the feature vector comprises cues characteristic of specularity in an image.
24. The computer program product of claim 17 comprising the further step of selecting the region of the image file for classification upon a determination that the region is saturated.
25. The computer program product of claim 17 comprising the further step of selecting the region of the image file for classification upon a determination that the region satisfies specularity conditions in a source dependent color space.
26. The computer program product of claim 25 , wherein a bi-illuminant dichromatic reflection model is used to calculate an illumination source color.
27. The computer program product of claim 17 comprising the further step of, when the selected region is classified as being specular, setting an indication in a specular mask.
28. The computer program product of claim 17 wherein the selected region of the image comprises a token.
29. The computer program product of claim 17 , wherein the processing the image to remove specular characteristics from the region classified as being specular includes generating a material image expressed as a separate, multi-band representation, independent of an illumination image, the material image excluding the region classified as being specular.
30. The computer program product of claim 17 , wherein the image region has a single robust color measurement among contiguous pixels thereof if all of the pixels of a Type C token have colors within a noise level of a device used to record the image.
31. The computer program product of claim 17 , wherein the processing the image to remove specular characteristics from the region classified as being specular includes blurring the region classified as being specular.
32. The computer program product of claim 17 , wherein the processing the image to remove specular characteristics from the region classified as being specular includes reintegrating a gradient representation of the image, with the modified gradients, to regenerate the image without specularities.
33. The computer program product of claim 32 further comprising restoring texture the region after the image is regenerated.