Hardware-based convolutional color correction in digital images

A system comprising:

• a main image processing engine, implemented at least in part in a graphics processing unit, wherein the main image processing engine is configured to:
• receive, from image capture components, an input image,
• provide a reduced-size version of the input image, and
• provide, to a display or a memory, an output image;
• and
• an image co-processing engine, implemented at least in part in an application-specific integrated circuit, wherein the image co-processing engine is configured to:
• receive, from the main image processing engine, the reduced-size version of the input image;
• and
• generate a two-dimensional chrominance histogram of the reduced-size version of the input image;
• convolve the two-dimensional chrominance histogram with a filter to create a two-dimensional heat map, wherein entries in the two-dimensional heat map represent respective estimates of how close respective tints corresponding to the respective entries are to a white point of the reduced-size version of the input image;
• select an entry in the two-dimensional heat map that represents a particular value that is within a threshold of a value in the heat map that represents a tint that is closest to the white point,
• apply the filter to the reduced-size version of the input image to determine output coefficients, wherein the filter is parameterized by the selected entry, and
• provide, to the main image processing engine, the output coefficients, wherein the output image is based on the output coefficients.