摘要
提出一种新的图像融合方法,该方法将形态非抽样小波和引导滤波结合,实现一种快速有效并且对图像边缘保持明显的融合过程。该方法首先对源图像进行形态非抽样小波分解得到基础层图像,源图像减基础层图像得到细节图像。接着对细节图像进行显著比较得到权重图,然后对权重图进行引导滤波,源图像作为引导图得到优化后的权值图,最后将优化后的权值图分别与基础层图像和细节图像加权平均再相加得到最终的融合图像。相对于现有的图像融合算法,该方法能有效的保留源图像的边缘部分,有助于显著目标的识别。通过实验证明了该算法能得到具有高对比度的融合图像,融合效果良好,在实时图像融合中具有较高的应用价值。
A new image fusion method is proposed, which combined morphological un-decimated wavelets and guided filter to achieve a fast and effective and preserve images edges in the fusion process. Firstly, the method is based on morphological un-decimated wavelets decomposition of an image into base layer, source image subtraction base layer to get the detail image. Secondly, detail images image are compared to get weight map. Third, the guided image filtering is performed on each weight map with the corresponding source image serving as the guidance image to achieve the optimized weight map. Finally the optimized weight image respectivelywith weighted average of base image and detail image to get the final fusion result. Relative to the existing image fusion algorithm, the method can effectively preserve source image edges and contribute to target detection. Experiments show that the algorithm can get a fusion of images with high contrast and fusion works well and has a high value in the real-time image fusion.
出处
《深圳信息职业技术学院学报》
2015年第3期11-16,共6页
Journal of Shenzhen Institute of Information Technology
基金
国家自然科学基金项目
61271420
基于非抽样形态小波与视觉显著计算的图像融合的研究
关键词
图像融合
形态小波
引导滤波
非抽样
image fusion
morphological wavelets
guided filtering
un-decimated
