Color correction apparatus, method and computer program

A color correction method for images captured by a plurality of cameras ordered 1, 2, . . . , n using a correction order, the correction order defining a plurality of camera pairs and an order of the camera pairs, each camera pair defining a reference camera and a target camera, comprising:

• for each camera pair in accordance with the order of the camera pairs:
• a first step of generating a corrected target image by correcting a pixel value of a target image captured by the target camera based on a reference image captured by the reference camera, and
• a second step of replacing the target image with the corrected target image,
• wherein each camera is treated as a target camera at least one time, when camera n is corrected, the correction continues reversely, and the reference camera in a certain camera pair is treated as the target camera in a previous camera pair except for a first camera pair of the correction order, and
• the second step includes a correspondence output step of determining pixel pairs, each pixel pair having a pixel in the target image, and a pixel in the reference image corresponding to the pixel in the target image;
• and
• a color correction step of generating a lookup table indicating pixel values before and after color correction based on the pixel pairs, and for correcting pixels in the target image based on the lookup table,
• wherein the color correction step includes a 2-dimensional Gaussian distribution generating step of generating a 2-dimensional Gaussian distribution for each pixel pair, the 2-dimensional Gaussian distribution having a mean equal to pixel values of the pixel pair, and a predetermined variance;
• a total distribution generating step of generating a total distribution by accumulating the 2-dimensional Gaussian distribution or a product of the 2-dimensional Gaussian distribution and a weight over all pixel pairs, wherein the weight=(N−|Preference−Ptarget|)/N, where Preference is a pixel value of the reference image pixel in the pair, Ptarget is a pixel value of the target image pixel in the pair, and N is the possible maximum pixel value in an image;
• and
• a lookup table generating step of generating the lookup table by setting a distribution value for each coordinate on an x-y plane based on the total distribution, and by selecting one coordinate for each x value,
• wherein an x value of the selected coordinate indicates a pixel value before correction, and a y value of the selected coordinate indicates a pixel value after correction.