Novel fast color reduction algorithm for time-constrained applications

In this paper, we propose a new adaptive approach of color quantization. It can significantly reduce the time consumption during the process compared with available methods but still maintains a good quality (greater than 30 dB of PSNR). It is implemented as a part of the media stream compression algorithms for a True Color Signboard System. We adapt and create some techniques to speed up the process. We start with a sampling technique on an RGB color space before constructing the 3D histogram of color distribution, and then we use the dynamic programming based on Wu's algorithm to construct the cumulative moment distribution. Then, we put the cutting plane through the centroid of that box. This plane is perpendicular to the axis, on which the sum of the squared Euclidean distances between the centroid of both of sub-boxes and the centroid of the box is the greatest. The sub-box, which contains the greatest value representing variance, is repeatedly sub-divided into the smaller sub-boxes until reaching the desired number of the representative colors. From our whole process, we gain approximately up to 50% less time consumption than Wu's quantizer [ACM Trans. Graph. 11 (1992) 348] and it is significantly faster than existing algorithms as shown in the result.