摘要： 提出了一种遗传k-均值算法，该算法通过改进标准遗传操作及采用可变变异率，使 其在矢量量化应用中表现出很好的性能。实验证明，该算法能够获得质量高于k-均值和模糊 k-均值算法的矢量量化码书，为设计全局最优码书提供了新思路。

关键词: 矢量量化, 遗传算法, 模糊, k-均值

Abstract: A genetic k-means algorithm for vector quantization is proposed in t his paper. By using k-means algorithm as a novel genetic crossover operator, thi s algorithm shows superior performance for image vector quantization in comparis on with k-means and fuzzy k-means algorithms. Also, a variable mutation rate is demonstrated to be helpful for improving the algorithm performance. This hybrid algorithm gives an idea to the combination of GA and conventional technologies f or solving more complex problem.

中图分类号: 

• TP18