Fuzzy algorithms for combined quantization and dithering

摘要: Color quantization reduces the number of colors in a color image, while subsequent dithering operation attempts to create illusion more with this reduced palette. In quantization, palette is designed minimize mean squared error (MSE). However, that follows enhances appearance at expense increasing MSE. We introduce three joint and algorithms overcome contradiction. The basic idea same two approaches: introducing quantizer training phase. fuzzy C-means (FCM) learning vector (FLVQ) are used develop combined mechanisms. third algorithm, we an objective function including inter-cluster separation (ICS) term obtain which suitable for dithering. goal enlarge convex hull after diffusion. contrasts images also enhanced proposed algorithm. test results these new using quality metrics model perception human visual system illustrate substantial improvements achieved