基于内生长机制和卷积稀疏表示的红外与可见光图像融合

为了提升红外与可见光图像融合视觉效果,克服融合结果的伪影效应,提出一种基于内生长机制结合卷积稀疏表示的图像融合方法.首先,采用符合人类大脑推理的内生长机制对源图像进行分解,获取预测层和细节层;其次,对细节层采用卷积稀疏表示进行二次分解,获取二次细节层和基本层,并分别对其采用活动水平测度取大以及加权平均规则进行融合;再次,针对预测层定义ISR混合算子融合规则,并进行融合;最后,将融合后的预测层和细节层相加获取最终融合结果.实验中,采用3组具有代表性的红外与可见光图像进行算法测试,实验结果表明所提出的方法具有较好的主观视觉效果,并且客观评价指标更好,具有有效性.

Infrared and visible light image fusion based on internal generative mechanism and convolution sparse representation

In order to improve the visual effect of infrared and visible light image fusion and overcome the artifact effect of the fusion result, an image fusion method based on the internal generative mechanism and convolution sparse representation is proposed. Firstly, the source image is decomposed using the internal generative mechanism that conforms to the reasoning of the human brain to obtain the prediction layer and the detail layer. Then, the detail layer is decomposed using a convolution sparse representation to obtain the secondary detail layer and the basic layer, and the activity level measurement is made to be larger and the weighted average rule is fused separately. The ISR hybrid operator fusion rule is defined for the prediction layer. Finally, the fusion prediction layer and detail layer are added to obtain the final fusion result. In the experiment, three representative infrared and visible light images are used for algorithm testing. The experiment results show that the proposed method has good subjective visual effects, and the objective evaluation indicators are also better and effective.