摘要
针对NSCT移不变形、多尺度性和多方向性等特点,结合脉冲耦合神经网络,量融合规则,高频系数则采用PCNN融合规则。最终对融合后的系数经NSCT逆变换得到了融合图像。实验结果表明,该方法更好地保留了原图像中的有用信息,并提高了融合图像质量。
An image fusion algorithm is proposed based on the shift-invariant, multi-scale and multi-directional NSCT and the pulse coupled neural network PCNN. First, a multiscale decomposition of the source image is per- formed with NSCT, and then with low coefficient using improved regional energy integration rules and the high fre- quency coefficients using PCNN fusion rule, the final fused image is obtained through NSCT inverse transform of the fused coefficients. Experimental results show that this method better retains useful information in the original image, and improves the quality of the fused image.
出处
《电子科技》
2014年第4期30-33,共4页
Electronic Science and Technology
基金
湖南省教育厅基金资助项目(12C078)
关键词
图像融合
非下采样轮廓波变换
脉冲耦合神经网络
区域能量
image fusion
nonsubsampled contourlet transform
pulse coupled neural networks
regional energy
