基于二通道不可分小波与深度学习的红外与可见光图像融合方法
Infrared and Visible Image Fusion Method Based on Two Channel Non-Separable Wavelet and Deep Learning

红外与可见光图像融合在武器检测、目标识别领域中扮演着重要角色，而融合的关键是通过适当方法从源图像中提取显著特征并将其组合生成融合图像，因此提出了基于不可分小波与深度学习的红外与可见光图像融合方法。首先，构造不可分小波滤波器组，通过二通道不可分小波滤波器将源图像分解为高频子图和低频子图；然后，利用深度学习网络提取高频子图的深度特征，采用多层融合的策略得到权值映射，利用权重图和高频细节得到融合后的高频子图；最后，对融合后的低频子图和高频子图进行重构，得到最终的融合图像。实验结果表明，与其他相关方法相比，本文方法在主观视觉和客观指标评价上都取得了更好的结果。
Infrared and visible image fusion is playing an important role in the field of weapon detection and target recognition, and the key of fusion is to extract significant features from the source images and generate fused image by appropriate methods, so the fusion method of infrared and visible image based on non-separable wavelet and deep learning is proposed. First, the non-separable wavelet filter banks are constructed; the source images are decomposed into high frequency sub-image and low frequency sub-image by two channel non-separable wavelet filter. Then, the deep learning network is used to extract the depth features of the high frequency sub-image, and the multi-layer fusion strategy is adopted to obtain the weight mapping, and the fused high frequency sub-image is obtained by using weight graph and high frequency detail. Finally, the low frequency sub-image and high frequency sub-image are reconstructed to obtain the final fused image. The experimental results show that compared with other related methods, the proposed method in this paper achieves better results in both subjective and objective evaluation.