摘要
为解决声纳图像中存在的高斯与脉冲噪声的同时去除问题,在简化的脉冲耦合神经网络(Pulse Coupled Neural Network,PCNN)模型的基础上,提出了一种结合了数学形态学与中值滤波的噪声抑制算法。首先利用PCNN输出二值图像确定噪声在图像中的位置,用数学形态学方法保持目标的完整性,然后利用中值滤波方法去除图像的脉冲噪声并对PCNN逐次迭代的输出结果进行调整去除高斯噪声。实验结果证明,方法在保持图像边缘信息的前提下,与其他方法相比获得了更好的去噪效果。
For reducing the Gaussian and pulse noise in sonar images simultaneously, a noval algorithm was proposed by incorporating mathematical morphology with median filter based on the simplified Pulse Coupled Neural Net-work (PCNN). At first, the algorithm used the output images of PCNN to confirm the positions of noise points, and used the mathematical morphology method to keep the integrity of target in the sonar images. Then the median filter was used for suppressing the pulse noise, and the iterative PCNN outputs were modified to reduce the Gaussian noise. Simulation results show that the proposed algorithm is more effective than general de -noising methods.
出处
《计算机仿真》
CSCD
2008年第10期219-222,共4页
Computer Simulation
关键词
噪声抑制
脉冲耦合神经网络
数学形态学
声纳图像
Noise suppression
Pulse coupled neural network (PCNN)
Mathematical morphology
Sonar images