针对现有图像融合算法生成的融合结果质量不一、多数融合图像含有大量噪声的问题,提出一种融合图像增强方法。首先对源图像进行均值滤波并借助数字减影技术获得目标图像的显著区域,利用改进拉普拉斯算子对减影图像进行双尺度分解,得到对应的粗略聚焦区域和细化聚焦区域。然后根据像素级线性混合规则生成初始决策图,借助一致性检验算法对其进行细化以获得最终决策图。最后综合产生的结果重构新的融合图像。实验结果表明,所提方法不仅对现有融合算法生成的融合图像实现不同程度的增强,对噪声具有更强的鲁棒性,处理时间小于0.4 s;对待融合图像中小散焦或聚焦区域有更好的识别能力,识别的边缘信息更清晰光滑,并在客观指标上给出具体的验证结果。

With the aim of denoising the results of the existing image-fusion algorithms and making them more uniform with respect to quality, we propose a fusion image enhancement method. First, the source image is mean-filtered and salient area of the target image is obtained using the digital subtraction technology. The subtracted image is then decomposed in two-scale using an improved Laplacian operator to obtain the corresponding coarse and refined focus areas. Further, an initial decision graph is generated according to the pixel-level linear mixing rules, and the final decision graph is obtained by refining the initial decision graph using the consistency check algorithm. Finally, the results are synthesized to reconstruct a new fusion image. Experimental results show that the proposed method achieves different degrees of enhancements of the fusion image generated using the existing fusion algorithms, the image has improved robustness to noise, and processing time is less than 0.4 s. The small defocus or focus area in the fusion image is more. With good recognition ability, the edge information of recognition increases in clarity and smoothness, and specific verification results are given for objective indicators.