基于降采样块匹配的数字视频3D降噪算法  

A Digital Video 3D Denoising Algorithm Based on Down Sampled Block Matching

作　　者：谷元保[1] 余康为 姚洪涛[1] 

机构地区：[1]长春理工大学计算机科学技术学院,长春130022

出　　处：《长春理工大学学报（自然科学版）》2016年第2期98-102,共5页Journal of Changchun University of Science and Technology(Natural Science Edition)

摘　　要：针对数字视频图像的高斯白噪声,提出了一种基于降采样块匹配的3D降噪算法。该算法先进行边缘检测并分块噪声估计,根据噪声强度设置阈值,然后利用图像的自相似特征,对搜索块内部像素降采样,再进行块匹配运算以降低算法复杂度。在自适应阈值筛选阶段,对选出的刚体运动块和非刚体运动块分别统一进行处理,对刚体运动区域进行时域加权滤波,对非刚体运动区域用改进的双边滤波器进行系数自适应滤波,减少了单独对每个块进行处理所产生的块效应,对细节处理更好。仿真结果表明,该算法能明显改善图像主观质量,提高信噪比。For the Gauss random noise of digital video. An improved 3D denoising algorithm based on the down sam-ple block matching is proposed. Firstly the new 3D denoising algorithm carries on the edge detection and the block noise estimation, according to the noise intensity setting threshold. Then, by using local similarity of image, the pixels in the search block are used to do crossing sampling block matching in order to reduce th-e complexity of the algo-rithm. In adaptive threshold selection stage, the selected rigid moti-on blocks and the selected non-rigid motion blocks are unified processed respectively with d-ifferent methods. The temporal weighting filtering method is applied to the rig-id motion area. Of non rigid motion area using the improved bilateral filter coefficients adaptive filter,reduces the indi-vidually for each piece of block effect produced by processing,bet-ter for details. The simulation results show that the proposed algorithm can significantly improve the subjective image quality,improve the signal-to-noise ratio.

关 键 词：边缘检测 噪声估计 降采样 运动检测 双边滤波 

分 类 号：TN911.73[电子电信—通信与信息系统]