摘要
为了提高模拟退火算法的最终解的质量,文中对控制算法进程的冷却进度表进行了优化选取,尤其在控制马尔可夫链长方面,给出了依据算法搜索过程的反馈信息来控制马尔可夫链长的方法。将该算法与LBG算法相结合,应用于矢量量化图像编码,既保持了模拟退火对初始码书依赖性小、不容易陷入局部极值的优点,又具备LBG算法的易于实现和计算量小的特点。仿真实验表明,该算法提高了码书的编码性能。
The cooling schedule which controls the process of Simulated Annealing is optimized. In particalar, according to the feedback information of search course of Simulated Annealing, a method for controlling the length of Markov Chain is determined. The proposed method is applied to the vector quantization image coding, which is not very dependence on initial codebook, and not easy to fall into local minimization, with features of LBG algorithm in easy implementation, simple calculation. Simulation results show that this method can effectively improve the performance of the codebook.
出处
《通信技术》
2008年第2期81-82,88,共3页
Communications Technology
基金
国家自然科学基金项目(60601016)
陕西省自然科学基金项目(2006F40)
关键词
模拟退火
冷却进度表
马尔可夫链
矢量量化
图像压缩
simulated annealing
cooling schedule
markov chain
vector quantization
image compression
