PCNN与灰度形态学相结合的图像去噪方法.pdfVIP

2008年4月 北 京 邮 电 大 学 学 报 ADr．2008 第31卷 第2期 Journal of Beijing University of Posts and Telecommunications Vo1．31 NO．2 文章编号：1007．5321(2008)02—0108—05 PCNN与灰度形态学相结合的图像去噪方法 马义德， 张红娟 (兰州大学信息科学与工程学院，兰州730000) 摘要：通过总结脉冲噪声的特性，将脉冲耦合神经网络(PCNN)与数学形态学相结合，提出了一种基于PCNN与数 学形态学的脉冲噪声去除方法．首先利用PCNN的同步脉冲发放特性定位脉冲噪声点的位置；然后利用数学形态 学开．闭滤波对其进行去噪处理，并将结果与中值滤波、形态学开．闭滤波及PCNN与中值滤波相结合的方法进行了 比较．结果表明，本文方法取得了良好的效果，但是以比中值滤波和形态学滤波更多的运算时间为代价的．在和 PCNN与中值滤波相结合的方法结果相当的情况下，本文方法的运算时间较少． 关 键 词：脉冲耦合神经网络；中值滤波；形态学滤波；脉冲噪声；峰值信噪比 中图分类号：TN911．73 文献标识码：A A New Image Denoising Algorithm Combined PCNN with Gray-Scale Morphology MA Yi—de． ZHANG Hong一]uan (School of Information Science＆ Engineering，Lanzhou University，Lanzhou 730000，China) Abstract：Under summarizing the characteristics of pulse noise，the pulse coupled neura1 network (PCNN)is combined with gray—scale morphology，and 3 new pulse noise removing algorithm based on PCNN and gray—scale morphology is proposed．Firstly，the noisy points are located using the syn— chronous pulse burst property of PCNN，then，are locally processed using the gray—scale open—close fil— ter．Finally the algorithm is compared with median filtering，gray—scale open—close filtering and with the algorithm combined PCNN with median filtering．Experimental results show that the proposed al— gorithm performs wel1．However，it is at a cost of more calculating time than median filtering and gray— scale open-close filtering．But，with similar results，it spends less time than that of the algorithm com— bined PCNN with median filtering． Key words：pulse coupled neural network；median filtering；morphological filtering；pulse noise；peak signal—．to—．noise ratio 由于受到图像采集设备、图像传输过程和存储