计算机科学与探索 ›› 2014, Vol. 8 ›› Issue (3): 321-328.DOI: 10.3778/j.issn.1673-9418.1306019

• 人工智能与模式识别 • 上一篇    下一篇

SCDPT下的分层图像融合策略研究

宋福照+,杜军平,侯立莎   

  1. 北京邮电大学 计算机学院,北京 100876
  • 出版日期:2014-03-01 发布日期:2014-03-05

Research on Layered Image Fusion Rule with SCDPT

SONG Fuzhao+, DU Junping, HOU Lisha   

  1. School of Computer Science, Beijing University of Posts and Telecommunications, Beijing 100876, China
  • Online:2014-03-01 Published:2014-03-05

摘要: 针对空间对接背景下多源图像融合问题,提出一种基于SCDPT(shiftable complex directional pyramid transform)的分层图像融合策略。对于运动图像序列中运动目标由远到近过程中目标大小的变化,提出了一种分层融合的方法,将视觉特征明显的运动目标区域采用基于匹配度-区域内积能量的SCDPT融合算法,而非运动目标区域采用传统的融合算法。通过实验以及对实验结果的主观评价准则和客观评价准则,分析证明了图像的分层融合策略在标准图片和空间图片上都可以达到满意的融合效果。

关键词: 图像融合, 区域内积能量, 分层融合, 运动图像

Abstract: This paper proposes a layered image fusion strategy based on SCDPT (shiftable complex directional pyramid transform), which is under the background of the space docking. For a sequence of moving images, the size of moving target changes when it moves from far to near. This paper proposes a layered fusion method, in which the moving target region with obvious visual characteristics uses a fusion algorithm based on SCDPT with the image fusion rule based on matching degrees and region inner product energy, and the non-moving target region uses the traditional fusion algorithm. Finally, the experimental results of subjective evaluation criterion as well as objective evaluation criterion show that the layered image fusion strategy can achieve satisfactory fusion results both with standard pictures and space pictures.

Key words: image fusion, region inner product energy, layered fusion, moving image