摘要
针对传统图像融合提取细节和边缘信息的不足,提出了一种基于非采样轮廓波变换(NSCT)和脉冲耦合神经网络(PCNN)的改进梯度图像融合算法。首先源图像经过NSCT分解,得到低频和高频子带系数;其次将低频子带系数的空间频率,高频子带系数采用八方向Sobel算子检测的改进梯度,作为PCNN的脉冲输入激励,根据不同的点火映射图选择低频和高频的融合系数;最后通过NSCT反变换得到融合图像。实验结果表明,本文算法比小波方法、传统NSCT方法和NSCT-SF-PCNN方法图像融合效果好。融合图像的信息丰富,清晰度高,保留更多的边缘细节信息。
With the lack of the traditional image fusion in extracting details and edge information, an image fusion algorithm of improved gradient was proposed based on Nonsubsampled Contourlet Transform (NSCT) and Pulse Coupled Neural Networks (PCNN). Firstly, source images were decomposed through the NSCT, thus low frequency and high frequency subband coefficients were obtained; Secondly, spatial frequency of the low frequency and improved gradient of high frequency ,which was detected by eight direction Sobel operator, were input to PCNN as pulse. The fused coefficients of low frequency and high frequency were choosen according to different ignition map. Finally, the fused image was obtained tb_rough NSCT inverse transformation. The experimental results show that the proposed algorithm is better than wavelet method, the traditional method of NSCT and NSCT-SF-PCNN method. The fused image has rich information, high resolution and retains more edge detail information.
出处
《数字技术与应用》
2015年第11期124-127,共4页
Digital Technology & Application
关键词
图像融合
非采样轮廓波变换
脉冲耦合神经网络
改进梯度
image fusion
nonsubsampled contoudet transform(NSCT)
pulse coupled neural networks(PCNN)
improved gradient