本文分析了确定性模拟退火技术、竞争学习算法在图像编码中的压缩机理 ,提出了一种新的随机竞争学习矢量量化算法 .该算法将竞争过程与代价函数最小化结合起来 ,在学习过程中引入模拟退火 ,并针对矢量量化图像编码的特点 ,提出了新的参数选取策略 ,具有对初始码书依赖性小 ,不会局部最小 ,收敛速度快 ,码书性能好等优点 .文中还通过计算机实践对该方法进行了性能分析 ,验证了算法的有效性和鲁棒性 . This paper analyzes the compression mechanism of deterministic simulated annealing and competitive learning algorithm in image coding, and proposes a new random vector quantization algorithm for competition learning, which combines the competition process with the cost function minimization. During the learning process The introduction of simulated annealing and vector quantization based on the characteristics of coding, a new parameter selection strategy, with the initial codebook with small dependencies, no local minimum, convergence speed, codebook performance is good.This paper also by computer The performance of this method is analyzed in practice, which verifies the effectiveness and robustness of the algorithm.