Statistical Colour Quantization for Minimum Distortion

摘要: Colour quantization for frame buffer displays, a basic computer graphics technique efficient use of colours, is statistically large scale clustering process and it poses difficult discrete optimization problem. Many heuristic color algorithms were proposed [7, 12, 22, 24] but they all suffer from various degrees non-adaptability to the colour statistics input images. In this article we will introduce an adaptive image algorithm that outperforms existing in subjective quality least square approximation sense. The new eliminates many current suboptimal treatments such as prequantization discarding few lower bits RGB values, restricted cuts orthogonal axes, partition criteria based on population or marginal distributions rather than variance minimization. A constrained global scheme incorporated into divide-and-conquer minimize distortion. time space complexities are O(N log K) O(N), where N number pixels K colours quantized image.