Abstract: A system and method for processing data representative of a color image is disclosed. Image data represents an intensity of photoexposure of an imaging array at specific locations in the imaging array and in distinct spectral regions corresponding to color channels. A process identifies “white” regions in the color image by comparing the intensities of photoexposure of groups of associated pixels which are responsive to photon energy in different spectral regions. If the intensities of photoexposure of the pixels in the group of associated pixels are proportionally equivalent, these pixels are determined to be in a white region of the image. White balancing gain coefficients are then based upon the pixel intensity values at pixel locations in the white regions of the image.

Abstract: Described herein are a system and method for reconstructing a color image from sparse image data collected at an imaging sensor which provides intensity data for exactly one color for each pixel location. A pattern for color selection preferably defines an assignment of color information to each pixel location from neighboring pixels. Thus, the color selection pattern provides an assignment of multicolor image data for each pixel location including color information collected at the pixel location in the imaging sensor and color information selected from neighboring pixels according to the color selection pattern. The color selection pattern may be formulated to minimize the effect of color artifacts in the resulting reconstructed image.