计算机工程与应用 ›› 2015, Vol. 51 ›› Issue (5): 193-197.

• 图形图像处理 • 上一篇    下一篇

基于改进拉普拉斯能量和的快速图像融合

孙晓龙1,王正勇1,符耀庆2,易  云1,何小海1   

  1. 1.四川大学 电子信息学院,成都 610064
    2.中海油能源发展股份有限公司北京分公司,北京 100027
  • 出版日期:2015-03-01 发布日期:2015-04-08

Fast image fusion based on sum of modified Laplacian

SUN Xiaolong1, WANG Zhengyong1, FU Yaoqing2, YI Yun1, HE Xiaohai1   

  1. 1.College of Electronics and Information, Sichuan University, Chengdu 610064, China
    2.CNOOC Energy Technology & Services Limited Beijing Branch, Beijing 100027, China
  • Online:2015-03-01 Published:2015-04-08

摘要: 为了得到优质的融合图像,提出了一种改进的拉普拉斯能量和(New Sum of Modified Laplacian,NSML)多聚焦图像融合算法。该算法在传统SML计算每个像素点的变步长拉普拉斯算子值仅有的水平和垂直方向的基础上,增加了斜对角线上的四个方向。同时通过分析NSML算法的计算过程,发现存在大量的重复计算,从而提出了基于积分图像的快速NSML图像融合方法。该方法通过简化NSML的计算过程,大大减少了图像融合处理过程消耗的时间,提高了图像融合的效率。实验结果表明,快速NSML方法在达到极佳融合图像质量的同时,提升了算法的实时性。

关键词: 图像融合, 改进拉普拉斯能量和, 空间域, 一致性校验, 积分图像

Abstract: A new algorithm is proposed based on Sum of Modified Laplacian(NSML), for the fusion problem of the multi-source images of the same scene. The traditional SML only calculates the Laplacian value of each pixel on horizontal and vertical direction, while NSML adds the four directions on the diagonal ramp. Fast NSML based on integral image is put forward in order to solve the problem that there are amounts of repeated calculations in the process of NSML. Fast NSML greatly reduces the time consumed by the fusion process and improves the efficiency of fusion by mean of simplifying the calculation process. The experimental results show that the method achieves excellent fusion of image quality, at the same time, enhances the real-time algorithm.

Key words: image fusion, sum of modified Laplacian, spatial domain, consistency check, integral image