Figure 20.5 The photosites in a filter-matrix camera are masked with color filters in a regular pattern. Finding the (R, G, B) triad for each pixel requires interpolating the two missing color channels from its neighbors, trading some loss of sharpness for the simplicity of one-shot color imaging.
the array is reduced by a factor of approximately 1.6.
To see how we obtain SR, SG, and SB values for every pixel, consider the pixels in an RGB filter matrix (see Figure 20.5). Each 2x2 unit cell contains one blue-, one red-, and two green-filtered pixels, and each of these is the center of a neighborhood comprised of eight pixels. Surrounding each blue pixel are four red pixels and four green pixels. Thus a complete (R, G, B) triad for a blue pixel at location (x, y) is:
Similar sets of equations apply for red and green pixels. In this way, a raw filter-matrix image is converted to yield a complete set of red-, green-, and blue-filtered images, SR, SG, and SB .
Once decoded, the color channels from a filter-matrix image can be pro cessed the same way other red/green/blue-filtered image sets are processed and color balanced—providing, of course, that the camera and software that take the image do not perform non-linear operations on the raw image data.
Was this article helpful?