为了解决用多尺度top-hat分解法融合双色中波红外图像时经常存在对比度提升有限、 边缘区域失真较重的问题, 提出了基于支持度变换和top-hat分解相结合的融合方法。 先用支持度变换法将双色中波图像分解为低频图像和支持度图像序列; 再从最后一层低频图像中用多尺度top-hat分解法提取各自的亮信息和暗信息; 用灰度值取大法分别融合亮信息和暗信息; 通过灰度值归一化和高斯滤波分别增强亮、 暗信息融合图像; 然后融合两低频图像和亮、 暗信息增强图像; 将融合图像作为新的低频图像和用灰度值取大法融合得到的支持度融合图像序列进行支持度逆变换, 得到最终融合图像。 该方法的实验结果同采用单一的支持度变换法融合和多尺度top-hat分解法融合相比, 融合图像的对比度提升了11.69%, 失真度降低了63.42%, 局部粗糙度提高了38.12%。 说明提出的从低频图像提取亮暗信息, 并经过分别融合、 增强, 再与低频图像进行融合, 能有效破解红外融合图像对比度提升和边缘区域失真度降低之间的矛盾, 为提高图像融合质量提供了新方法。

Fusion method of dual color mid-wave infrared images is presented in this paper in order to solve such frequently rising issues as limited contrast ratio improvement and serious marginal area distortion in the fusion of the above two images using multi-scale top-hat decomposition. The detailed procedure is shown as the following: A low-frequency component image and a sequence of support value images of the two subdivision band images of mid-wave infrared are obtained respectively with support value transform. Multi-scale bright and dim information are first extracted from the last layer of low-frequency image using the multi-scale top-hat decomposition method respectively. Then they are fused by selecting the maximum gray of each pixel in two subdivision band images of mid-wave infrared respectively. Following that, the two resulted images are enhanced using the gray-scale normalization and Gaussian filtering and fused with the two low-frequency images to get the low-frequency fusion image. After that, this fusion image is reversely transformed with the support sequence image fused by selecting the maximum gray. The final image is got at last. The result shows that compared with the simple support value transform fusion and the multi-scale top-hat decomposition fusion, the method suggested in this paper successfully increases the contrast ratio by 11.69%, decreases the distortion factor by 63.42%, and increases the local coarseness by 38.12%. All these show that the validity of fusion method proposed has been proved, which indicates that both bright and dim information from low-frequency images can effectively solve the contradiction between improving fused image’s contrast ratio and reducing its’ distortion after the both are fused and enhanced respectively, and then fused with the two low-frequency images, which provides a new useful method for improving the quality of fused inferred images.