Accelerated Image Halftoning Technique Using Improved Genetic Algorithm(Special Section on Digital Signal Processing)

概要

論文の詳細を見る
This paper presents an accelerated image halftoning technique using an improved genetic algorithm with tiny populations. The algorithm is based on a new cooperative model for genetic operators in GA. Two kinds of operators are used in parallel to produce offspring : (i) SRM (Self-Reproduction with Mutation) to introduce diversity by means of Adaptive Dynamic-Block (ADB) mutation inducing the appearance of beneficial mutations. (ii) CM (Crossover and Mutation) to promote the increase of beneficial mutations in the population. SRM applies qualitative mutation only to the bits inside a mutation block and controls the required exploration-exploitation balance through its adaptive mechanism. An extinctive selection mechanism subject SRM′s and CM′s offspring to compete for survival. The simulation results show that our scheme impressively reduces computer memory and processing time required to obtain high quality halftone images. For example, compared to the conventional image halftoning technique with GA, the proposed algorithm using only a 2% population size required about 15% evaluations to generate high quality images. The results make our scheme appealing for practical implementationsof the image halftoning technique using GA.
社団法人電子情報通信学会の論文
2000-08-25