Video calibration with illumination invariant image

1. A method in a vehicle, comprising:

acquiring a color image having a portion that includes a color calibration pattern;

generating, based on the color calibration pattern, a color-corrected color image that includes a plurality of pixels;

determining an illumination invariant image based on the color-corrected color image and transform parameters α v and α r ; and

actuating navigation of the vehicle based on the illumination invariant image, wherein determining the illumination invariant image includes classifying pixels of the plurality of pixels of the color-corrected image into one of a plurality of categories, and selecting the transform parameter α v or the transform parameter α r for each classified pixel of the color-corrected color image based on the plurality of categories.

2. The method of claim 1 , wherein generating the color-corrected color image includes using values in the color calibration pattern in a series of linear equations that are solved to form transform parameters to color-correct the color image.

3. The method of claim 2 , wherein determining the illumination invariant image based on transform parameters α v and α r is performed depending upon a segmentation classification.

4. The method of claim 3 , wherein the series of linear equations are solved to obtain red-green-blue values for the color image that are the same as red-green-blue values of a color image acquired when the sun is at its highest point.

5. The method of claim 1 , wherein the color calibration pattern includes a plurality of colors at a plurality of intensities.

6. The method of claim 5 , further comprising using a near infrared (NIR) video camera that images near infrared frequencies of light in addition to a color video camera to increase a range of colors available for calibration.

7. The method of claim 6 , wherein the range of colors include red, green and blue.

8. The method of claim 1 , further comprising determining whether to transform an RGB pixel of the color image with transform parameter α v or transform parameter α r depending upon whether the RGB pixel is classified as “vegetation” or “non-vegetation”.

9. The method of claim 8 , wherein the RGB pixel is classified as “vegetation” or “non-vegetation” based on segmenting the color image by a trained convolutional neural network.

10. A system, comprising:

a processor; and

a memory, the memory including instructions to be executed by the processor to:

acquire a color image having a portion that includes a color calibration pattern;

generate, based on the color calibration pattern, a color-corrected color image that includes a plurality of pixels;

determine an illumination invariant image based on the color-corrected color image and transform parameters α v and α r ; and

actuate navigation of the vehicle based on the illumination invariant image, wherein determining the illumination invariant image includes classifying pixels of the plurality of pixels of the color-corrected image into one of a plurality of categories, and selecting the transform parameter α v or the transform parameter α r for each classified pixel of the color-corrected color image based on the plurality of categories.

11. The system of claim 10 , wherein generating the color-corrected color image includes using values in the color calibration pattern in a series of linear equations that are solved to form transform parameters to color-correct the color image.

12. The system of claim 11 , wherein determining an illumination invariant image based on transform parameters α v and α r is performed depending upon a segmentation classification.

13. The system of claim 12 , wherein the series of linear equations are solved to obtain red-green-blue values for the color image that are the same as red-green-blue values of a color image acquired when the sun is at its highest point.

14. The system of claim 10 , wherein the color calibration pattern includes a plurality of colors at a plurality of intensities.

15. The system of claim 14 , wherein the plurality of colors are obtained using a near infrared (NIR) video camera that images near infrared frequencies of light in addition to a color video camera to increase a range of colors available for calibration.

16. The system of claim 10 , further programmed to determine whether to transform an RGB pixel of the color image with transform parameter α v or transform parameter α r depending upon whether the RGB pixel is classified as “vegetation” or “non-vegetation”.

17. The system of claim 16 , wherein the RGB pixel is classified as “vegetation” or “non-vegetation” based on segmenting the color image by a trained convolutional neural network.

18. A system, comprising:

a sensor operative to acquire a color image;

a processor; and

a memory, the memory including instructions to be executed by the processor to:

acquire a color image having a portion that includes a color calibration pattern;

generate, based on the color calibration pattern, a color-corrected color image that includes a plurality of pixels;

determine an illumination invariant image based on the color-corrected color image and transform parameters α v and α r ; and

actuate navigation of the vehicle based on the illumination invariant image, wherein determining the illumination invariant image includes classifying pixels of the plurality of pixels of the color-corrected image into one of a plurality of categories, and selecting the transform parameter α v or the transform parameter α r for each classified pixel of the color-corrected color image based on the plurality of categories; and

a controller to control motion of the vehicle according to the navigation by actuating vehicle components based on commands received from the processor.

19. The system of claim 18 , wherein the sensor is a color video camera.

20. The system of claim 18 , wherein the vehicle components include braking, steering and powertrain.