基于近红外和可见光差分特征的图像融合方法

暗光环境拍摄的图像通常存在昏暗、噪声、偏色等退化问题。为此，提出一种基于差分特征的可见光与近红外融合方法。首先，观察到近红外与可见光图像成像不一致性，通过两者结构性差异构造差分特征；其次，利用差分特征映射为融合权重进行权重融合；最后，采用偏色校正方法对融合后易产生偏色区域进行亮度复原，使得融合后偏色区域颜色真实自然。提出的方法能切实有效地利用近红外信息，输出高质量清晰图像。实验结论表明，该方法在主观感受和客观评价上均优于现有算法。

Near-infrared and visible images fusion based on differential feature

Images captured in dim light often suffer from severe degradations, such as low visibility, intensive noises and color distortion. To tackle these issues, this paper proposed a near-infrared fusion method based on image differential feature. Firstly, it introduced the differential map which captured the inherent discrepancy of structures in the two different types of images. Then, it conducted feature mapping functions to form an optimal weight map and recoverd a visually pleasing result. To reduce the possible color cast in the image recovery process, it proposed jointly recovering the final image by using the visible image to reduce over brightness in the fused image. The proposed algorithm can always take full advantage of the near-infrared images and generate natural high quality outputs. Experimental results demonstrate the proposed method is superior to existing fusion methods in terms of visual quality and quantitative measurements.