摘要
脉冲噪声是导致图像质量退化的重要因素.针对嵌入式系统应用的实时处理要求,提出基于多重决策论的降噪算法.针对传统决策论算法在降噪性能和算法稳定性中存在的问题,结合硬件设计的特点,在决策过程中引入了多重判定来选取邻域最优插值像素进行脉冲噪声点的像素替换,以提高降噪效果;并采用流水线结构的电路设计来提高其硬件实现特性.逻辑综合和基于FPGA的设计验证结果表明,该算法对于各类任意灰度值、不同浓度的双极型脉冲噪声都具有优异的脉冲降噪性能,且算法的计算复杂度低、所占用的逻辑资源和存储资源少,可以很好地满足高清视频图像处理和实时应用的要求.
In real-time video applications, impulse noise is a major factor affecting image quality. In this paper, a new multi-decision based de-noising algorithm is proposed for a better reduction of bipolar impulse noise. The new algorithm uses multi-decision method to obtain the optimal value in the filter window to replace the polluted pixel. Meanwhile, pipeline structure is utilized to improve the characteristics of hardware implementation. The experimental results show that, together with an improvement in robustness, the proposed algorithm has a magnificent de-noising performance for highly polluted images. With low computational complexity and minimized usage of resources in hardware implementation, the proposed algorithm has a full capability to be used in any video application that requires real-time processing.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2013年第6期832-842,851,共12页
Journal of Computer-Aided Design & Computer Graphics
基金
上海市科委"科技创新行动计划"(08706200100)
关键词
多重决策论
脉冲降噪
硬件实现
流水线架构
实时处理
multi-decision
impulse de-noising
hardware implementation
pipeline
real-timeprocessing
