White balance and color correction for interior vehicle camera

1. A method performed by one or more computing devices comprising processing hardware and storage hardware, the method performed by the processing hardware executing instructions from the storage hardware, the method comprising:

receiving an image captured by a camera, the image comprising a representation of a cabin area of a vehicle;

demosaicing the image; and

optimizing white balancing of the image based on previously determined data associated with the cabin area of the vehicle, the previously determined data comprising color information for a material of the cabin area of the vehicle, the optimizing comprising:

classifying pixels in the image;

solving global illumination for the image;

based on the classification of the pixels and the global illumination, solving a bidirectional reflectance distribution function (BRDF) for the color information about the material in the cabin area of the vehicle; and

based on the BRDF, generating a white balance matrix for the image;

applying the white balance matrix to the image; and

outputting the white-balanced image.

2. The method according to claim 1 , further comprising:

based on the BRDF, generating a color correction matrix for the image; and

applying the color correction matrix to the image.

3. The method according to claim 1 , further comprising applying a noise reduction filter to the image prior to generating the white balance matrix.

4. The method according to claim 1 , further comprising:

determining that an uncertainty level of the white balance matrix exceeds a threshold;

based on the determining that the uncertainty level exceeds the threshold, activating lighting of a light in the cabin area of the vehicle;

capturing a second image from the camera during the activated lighting of the light; and

based on the second image, solving a second BRDF for the color information about the material in the cabin area of the vehicle.

5. The method according to claim 4 , further comprising recomputing the white balance matrix based on the second BRDF.

6. The method according to claim 1 , wherein the previously determined data further comprise information about a chrome material in the cabin area.

7. The method according to claim 6 , wherein pixels in the image that correspond to the chrome material are used as color space information for solving the BRDF.

8. The method according to claim 1 , wherein the material comprises a seat, a headrest, a seatbelt, or a material covering a structural element of the cabin area.

9. Computer-readable storage hardware storing information configured to, when executed by a computing device, cause the computing device to perform a process, the process comprising:

receiving an image from a camera incorporated in a cabin of a vehicle; and

optimizing color correction and white balancing of the image via prior knowledge of the cabin of the vehicle, the prior knowledge comprising color information, the color information comprising information about color and/or reflectivity of one or more surfaces of the cabin area, the optimizing comprising:

applying a segmentation algorithm to the image to either classify pixels in the image or to compute depths of pixels in the image;

based on the classification or depth of the pixels, solving a bidirectional reflectance distribution function (BRDF) for the color information; and

based on the BRDF, computing a white balance matrix and a color correction matrix for the image;

applying the white balance matrix and the color correction matrix to the image; and

after the step of applying the white balance matrix and the color correction matrix to the image, storing or displaying the image.

10. The computer-readable storage hardware according to claim 9 , wherein the optimizing further comprises solving global illumination for the image, and wherein the solving the BRDF is further based on the global illumination.

11. The computer-readable storage hardware according to claim 10 , wherein the solving the BRDF is further based on information about position of the camera and position of the sun relative to the camera.

12. The computer-readable storage hardware according to claim 9 , wherein the white balance matrix and the color correction matrix are further based on pixels in the image determined to correspond to a reflective element of the cabin.

13. The computer-readable storage hardware according to claim 9 , wherein the image is captured by the camera while controlling lighting of a cabin light in the cabin, and wherein the white balance matrix and the color correction matrix are further based on the lighting of the cabin light.

14. The computer-readable storage hardware according to claim 13 , wherein the lighting is controlled responsive to a determination about noise corresponding to the white balance matrix or the color correction matrix.

15. A method performed by a computing device, the method comprising:

receiving an image from a camera built into a cabin of a vehicle;

demosaicing the image and applying a noise reduction step to the image;

applying a segmentation algorithm to the image;

solving global illumination for the image;

based on the segmentation of the image and the global illumination, solving a bidirectional reflectance distribution function (BRDF) for at least one of color or reflectance information of material in the cabin of the vehicle;

based on the BRDF, computing a white balance matrix and a color correction matrix for the image;

applying the white balance matrix and the color correction matrix to the image; and

storing or displaying the white-balanced and color-corrected image.

16. The method according to claim 15 , wherein the segmentation algorithm comprises a convolutional neural network.

17. The method according to claim 15 , wherein the color correction matrix and the white balance matrix are computed based on information about lighting external to the cabin.

18. The method according to claim 15 , further comprising determining a noise level according to the white balance matrix or the color correction matrix.

19. The method according to claim 18 , further comprising activating a cabin light of the cabin based on determining the noise level.

20. The method according to claim 15 , wherein the white balance matrix and color correction matrix are computed from the image by first applying a noise filter to the image.