基于全局人工鱼群算法的非下采样Contourlet变换阈值去噪研究

Nonsubsampled Contourlet Transform Threshold De-noising Based on Global Artificial Fish Swarm Algorithm

变换域阈值去噪是一种广泛使用的简单有效的图像去噪方法.采用具有平移不变性的非下采样Contourlet变换来实现图像的时域到Contourlet域的转换,将全局人工鱼群算法应用到各层系数的阈值选取过程中,提出了基于全局人工鱼群算法的非下采样Contourlet变换阈值去噪,通过对该算法与遗传算法、基本人工鱼群算法的收敛速度、去噪阈值及去噪效果的对比仿真表明,该算法能准确地找到全局最优阈值,且收敛速度更快,去噪效果更好.

The transform domain threshold de -noising is an easy and effective image de -noising method which is used widely. This paper adapted nonsubsampled Contourlet transform to convert the image from time - domain to Contourlet - domain, using global artificial fish swarm algorithm to select the thresholds of each level coefficient. Nonsubsampled Contourlet transform thresh- old de - noising based on global artificial fish swarm algorithm was proposed. Comparing simulation results by convergence rate, threshold de - noising and de - noising effect of the proposed algorithm, genetic algorithm and basic artificial fish swarm algorithm showed that the proposed algorithm could find out the best global threshold accurately with faster convergence rate and better de - noising effect.