RGB-NIR processing and calibration

A method for processing images acquired by a multi-spectral RGB-NIR (red/green/blue/near infra-red) sensor, comprising:

• receiving an RGB-NIR digital image from a multi-spectral RGB-NIR sensor, wherein the RGB-NIR digital image includes red (R) pixels, green (G) pixels, blue (B) pixels, and NIR pixels, wherein every other row in the RGB-NIR digital image includes NIR pixels that alternate with green pixels, and every other row in the RGB-NIR digital image includes green pixels that alternate with red and blue pixels;
• interpolating an NIR contribution to each R, G and B pixel value, wherein an NIR image is obtained;
• subtracting the NIR contribution from each R, G and B pixel value in the RGB-NIR digital image wherein a decontaminated RGB-NIR image is obtained;
• constructing a red, green and blue (RGB) Bayer image from the decontaminated RGB-NIR image;
• and
• processing the Bayer image wherein a full color image is obtained,
• wherein interpolating an NIR contribution to each R, G and B pixel value comprises:
• replacing R and B pixels in the RGB-NIR digital image with a G′ pixel whose value is a median of values of nearest neighbor G pixels to the R or B pixel being replaced, wherein an NIR-GGG′ image is obtained that includes NIR pixels, G pixels, and G′ pixels;
• and
• demosaicing the NIR-GGG′ image wherein a 3-channel image is obtained, where the 3-channel image includes a G channel image, a G′ channel image, and an NIR channel image
• wherein subtracting the NIR contribution from each R, G and B pixel value in the RGB-NIR digital image comprises:
• determining the NIR contributions αR, αG, αB are for each of R, G and B pixel values;
• detecting undershoot pixels in the RGB-NIR digital image, and replacing a value of each detected undershoot pixel with a value that is a minimum of K nearest neighbor pixels of a same channel as the detected undershoot pixel;
• and
• classifying pixels in the RGB-NIR digital image whose values are saturated by calculating a score of features of nearest neighbor pixels to the pixel being classified, wherein the features are relevant to a determination of whether the pixel being classified is saturated, and
• replacing the value of a pixel determined to be saturated with a convex combination of a fallback value and the pixel's decontamination value based on the calculated score.