基于连续小波阈值去噪算法的目标检测研究  

Research on Target Detection Based on Continuous Wavelet Threshold Denoising Algorithm

作　　者：刘冰[1,2] 刘雪梅 

机构地区：[1]达州职业技术学院,达州635001 [2]西南大学计算机与信息科学学院,重庆400715 [3]四川省达州市统计局,达州635000

出　　处：《现代计算机（中旬刊）》2017年第5期64-68,共5页Modern Computer

基　　金：四川省教育厅重点科技计划项目(No.14ZA0330);四川省达州市2014年科技计划项目(No.2014-8220)

摘　　要：对含噪图像的去噪处理是目标检测任务中的一个经常性工作,由于传统软/硬阈值去噪算法其估计小波系数在所设定的阈值处存在不连续和不可导的缺陷,常常导致重构图像的质量达不到目标检测的要求。对传统软/硬阈值去噪算法存在的问题进行详细的分析,并据此重新构造一个在所设定的阈值处具有连续可导的阈值函数,同时还给出相应的算法步骤。仿真实验表明,运用所提出的算法能够很好去除含噪图像中的噪声,较之传统软/硬阈值算法效果更加明显,其处理后的图像能很好地满足更进一步的目标检测任务的要求。The denoising of the noisy image is a recurring work in the target detection task, because of the traditional soft and hard threshold de- noising algorithm, the estimated wavelet coefficients have discontinuous and non-derivable defects at the set threshold, often leading to the quality of reconstructed images cannot meet the requirements of target detection. Analyzes the problems of traditional soft and hard threshold denoising algorithms in detail, and reconstructs a threshold function with continuous derivative at the set threshold. At the same time, gives the corresponding algorithm steps. The simulation results show that the algorithm proposed can remove the noise in the noisy image, which is more obvious than the traditional soft and hard threshold algorithm. The processed image can well meet the requirements of further target detection tasks.

关 键 词：图像去噪 软/硬阈值去噪 连续小波阈值 均方误差 峰值信噪比 

分 类 号：TN911.7[电子电信—通信与信息系统]