上海交通大学学报

上海交通大学学报（自然版）

• 自动化技术、计算机技术 • 上一篇    下一篇

基于视觉结构相似度的离焦图像融合方法

曹寅，蔡云泽，韩瑜
  

  1. （上海交通大学 电子信息与电气工程学院， 上海 200240）
  • 收稿日期:2010-03-26 修回日期:1900-01-01 出版日期:2010-11-30 发布日期:2010-11-30

A New Fusion Method for Lostfocus Images Based on Vision Structural Similarity

CAO Yin,CAI Yunze,HAN Yu
  

  1. (School of Electronic and Electric Engineering, Shanghai Jiaotong University, Shanghai 200240, China)
  • Received:2010-03-26 Revised:1900-01-01 Online:2010-11-30 Published:2010-11-30

摘要： 针对融合多张部分聚焦图像的问题，从离焦图像质量评价角度出发，提出了一种基于视觉结构相似度的图像质量评价标准，并以此标准将离焦图像融合转化为一个优化问题.通过特殊设计进化算法，从源图像出发，优化出一个质量评价最好的图像作为融合结果.结果表明，通过该方法能够得到较好的融合结果，并在继承性和可靠性方面优于流行的多尺度方法和特征组合方法.

关键词: 离焦图像融合, 图像质量评估, 视觉结构相似度, 进化计算

Abstract: When an image is captured by optical sensor, some parts of the scene will be blur due to lost focus. Image fusion is a method for integrating multifocus images into a composite image which is more suitable for work. In order to fuse multifocus images, a new blind image quality assessment metric, Vision Structural Similarity (VSS), was first proposed in this paper. Based on this metric, multifocus image fusion problem could be described as an optimization process. Through optimizing source images by evolutionary computation, the image with best VSS value would be chosen as fusion result. The experimental results of the proposed method show high performance in both consistency and reliability comparing with other fusion schemes.

中图分类号: