In order to improve the denoising performance of denosing method of lifting wavelet package gradual changing threshold selection and quantization,a denosing method of lifting wavelet package gradual changing threshold based on ant colony optimization was put forward. The deepness first combination tree was used as lifting wavelet package decomposing tree of the same state signals. In order to optimize the characteristic separability, a gradual changing threshold optimization mathematic model was set up. All the leafage nodes’ thresholds of the deepness first combination tree were optimized synchronically. According to the characteristics of the model, an ant colony method based on moving searching window was designed to resolve the model. The searching window began with the worst feasible solution area. With the searching results of the ant colony, the searching window was moved to preferable area. The simulation demonstrates that the denoising performance of the denosing method of lifting wavelet package gradual changing threshold is improved by the denoising method of gradual changing threshold based on ant colony optimization.

为进一步改进提升小波包渐变式阈值选择与量化降噪方法的降噪性能,提出了基于蚁群优化的提升小波包渐变式阈值降噪方法。用深度优先合并树作为同一状态信号的提升小波包分解树,建立了以特征可分离性最优的渐变式阈值优化数学模型,模型对深度优先合并树的各叶子节点的阈值进行同步优化。针对模型特点,设计了求解模型的基于平移搜索窗的蚁群算法,搜索窗从最差可行解区域开始,根据蚁群搜索结果,逐步向解的优良区域平移。仿真实验表明,基于蚁群优化的渐变式阈值降噪方法,从最优化的角度进一步增强了提升小波包渐变式阈值降噪方法的降噪效果。

