Scene-based automatic white balance

1. An image capture device comprising:

an image sensor configured to obtain an input image, wherein the input image is a raw image; and

a processor configured to:

obtain a raw image thumbnail and compute a histogram based on the raw image thumbnail;

obtain a first augmented image thumbnail and a second augmented image thumbnail, wherein the first augmented image thumbnail is used to determine a second histogram and the second augmented image thumbnail is used to determine a third histogram;

determine a scene classification for the raw image;

determine a first filter, a second filter, and a third filter, wherein the first filter is learned from one or more different instances of the raw image thumbnail and the determined scene classification, the second filter is learned from one or more different instances of the first augmented image thumbnail and the determined scene classification, and the third filter is learned from one or more different instances of the second augmented image thumbnail and the determined scene classification; and

store the first filter, the second filter, and the third filter.

2. The image capture device of claim 1 , wherein the first filter comprises stacked filters, and wherein one of the stacked filters corresponds to a scene classification.

3. The image capture device of claim 2 , wherein the scene classification includes a snow scene, an underwater scene, a beach scene, a vegetation scene, an urban scene, or an indoor scene.

4. The image capture device of claim 3 , wherein the processor is further configured to determine a probability of the scene classification.

5. The image capture device of claim 4 , wherein the processor is configured to apply the first filter, the second filter, and the third filter to one or more of the histogram, the second histogram, and the third histogram based on the determined probability of the scene classification to obtain scene-specific white balance weights and output a processed image.

6. The image capture device of claim 1 , wherein the processor if further configured to obtain metadata associated with the input image.

7. The image capture device of claim 6 , wherein the metadata associated with the input image includes an exposure time, an aperture setting, an ISO setting, algorithmic information, or semantics.

8. The image capture device of claim 7 , wherein the metadata includes semantics, and wherein the semantics include a detected face or a detected object.

9. The image capture device of claim 1 , further comprising:

a database configured to store the raw image thumbnail, the first augmented image thumbnail, and the second augmented image thumbnail.

10. The image capture device of claim 9 , wherein the processor is further configured to obtain the scene classification from an external processor, wherein the scene classification is based on the raw image or recently acquired raw images.

11. The image capture device of claim 9 , wherein the raw image thumbnail, the first augmented image thumbnail, and the second augmented image thumbnail are each associated with an exposure time, an aperture setting, an ISO setting, algorithmic information, or semantics.

12. An image capture device comprising:

an image sensor configured to obtain an input image; and

a processor configured to:

obtain a raw image thumbnail based on the input image;

compute a first histogram based on the raw image thumbnail;

obtain a first augmented image thumbnail and a second augmented image thumbnail, wherein the first augmented image thumbnail is based on the raw image thumbnail and is used to determine a second histogram, and the second augmented image thumbnail is based on the raw image thumbnail and is used to determine a third histogram;

determine a scene classification for the input image;

obtain a first filter, a second filter, and a third filter, wherein the first filter is learned from one or more different instances of the raw image thumbnail and the determined scene classification, the second filter is learned from one or more different instances of the first augmented image thumbnail and the determined scene classification, and the third filter is learned from one or more different instances of the second augmented image thumbnail and the determined scene classification;

combine the first filter, the second filter, and the third filter to obtain a combined filter;

perform a convolution of the combined filter to obtain an image from which the scene-specific white balance weights are retrieved;

apply the scene-specific white balance weights to the input image to obtain a processed image; and

output the processed image.

13. The image capture device of claim 12 , wherein the first filter comprises stacked filters, and wherein one of the stacked filters corresponds to a scene classification.

14. The image capture device of claim 13 , wherein the scene classification includes a snow scene, an underwater scene, a beach scene, a vegetation scene, an urban scene, or an indoor scene.

15. The image capture device of claim 14 , wherein the processor is further configured to determine a probability of the scene classification.

16. The image capture device of claim 12 , wherein the processor if further configured to obtain metadata associated with the input image.

17. The image capture device of claim 16 , wherein the metadata associated with the input image includes an exposure time, an aperture setting, an ISO setting, algorithmic information, or semantics.

18. A method comprising:

obtaining an input image;

obtaining a raw image thumbnail;

obtaining an augmented image thumbnail;

computing a histogram from an image thumbnail;

determining a scene classification result based on the input image;

determining a filter; and

applying the filter to the histogram to determine white balance correction coefficients and obtain a processed image.

19. The method of claim 18 , wherein the filter is a stacked filter, and wherein each filter of the stacked filter corresponds to a respective scene classification.

20. The method of claim 19 , wherein the respective scene classification includes a snow scene, an underwater scene, a beach scene, a vegetation scene, an urban scene, or an indoor scene.