论文部分内容阅读
针对传统脉冲耦合神经网络(PCNN)图像融合算法中最优融合结果无法自适应确定及神经元参数取固定常数所造成的同步脉冲周期无法随图像特征改变的不足,提出了一种基于人工鱼群寻优的自适应双通道PCNN图像融合算法。利用合成空间雷达(SAR)图像的辐射分辨率和可见光图像的清晰度分别作为双通道PCNN对应神经元的链接强度值,PCNN的信号衰减常数、阈值放大系数和水平调节因子3个参数采用人工鱼群寻优,目标函数由互信息(MI)和结构相似度(SSIM)两种图像质量评价指标构建,最终获得近似最优的融合图像。实验结果表明,本文算法图像融合结果优于传统拉普拉斯变换、离散小波变换和参数取固定值的PCNN图像融合算法及其一些改进算法。
In order to overcome the shortcomings that the synchronization pulse period can not be changed with the image characteristics due to the inability to adaptively determine the optimal fusion result and the neuron parameters taking the fixed constants in traditional pulse coupled neural network (PCNN) image fusion algorithm, an artificial fish Optimized Adaptive Dual Channel PCNN Image Fusion Algorithm. Using the resolution of SAR images and the sharpness of visible images as the link strength values of PCNN, signal attenuation constant, threshold amplification factor and horizontal adjustment factor of PCNN, artificial fish The objective function is constructed by two image quality evaluation indexes: mutual information (MI) and structure similarity (SSIM), and finally the fused image with approximate optimality is obtained. The experimental results show that the proposed algorithm is superior to PCNN image fusion algorithm and some improved algorithms of traditional Laplace transform, discrete wavelet transform and parameter fixed value selection.