摘要
提出了一种基于小波对比度和神经网络的多聚焦图像融合算法。首先对各源图像进行小波变换,根据变换后系数计算出图像的小波对比度,选取源图像部分区域小波对比度作为前馈神经网络的训练样本,调整神经网络权重;然后用训练好的神经网络组合融合图像的小波系数,对组合后的系数进行一致性校验;最后对该系数进行小波逆变换,得到融合图像。实验结果表明,该算法能够较好地解决多聚焦图像融合问题,生成的融合图像效果优于传统图像融合方法。
A novel multi-focus image fusion method using wavelet contrast and neural network was proposed. Firstly, source images were decomposed with Wavelet Transform （WT） and wavelet contrasts were obtained, Some of them were selected as samples for training feed-forward neural network. Then, the wavelet coemcients of fused image though neural network were output and verified by consistency, Finally, the fused image was obtained with inverse WT. Experiment results show that the proposed method outperforms traditional WT method for multi-focus image fusion.
出处
《计算机应用》
CSCD
北大核心
2006年第7期1590-1591,1601,共3页
journal of Computer Applications
基金
国家自然科学基金资助项目(60473015)
关键词
图像融合
小波变换
对比度
神经网络
image fusion
wavelet transform
contrast
neural network
