Author Login Editor-in-Chief Peer Review Editor Work Office Work

Computer Engineering

Previous Articles     Next Articles

A Multi-source Image Fusion Framework Supporting Parameters Optimization

JIANG Shao-hua, TANG Wen-sheng   

  1. (Institute of Image Recognition and Computer Vision, Hunan Normal University, Changsha 410081, China)
  • Received:2012-07-16 Online:2013-10-15 Published:2013-10-14

一种支持参数调优的多源图像融合框架

蒋少华,唐文胜   

  1. (湖南师范大学图像识别与计算机视觉研究所,长沙 410081)
  • 作者简介:蒋少华(1976-),男,博士,主研方向:图像处理,模式识别;唐文胜(通讯作者),教授、博士
  • 基金资助:
    湖南师范大学青年优秀人才培养计划基金资助项目(ET12606, ET61008);湖南师范大学博士科研启动基金资助项目

Abstract: The multi-source image fusion is one of the hotspots in image processing in recent decades. There are many different image fusion algorithms and fusion rules, but all these fusion algorithms have their advantages and disadvantages. So, it indeed needs an image fusion framework which can automatically determine the suitable algorithm and fusion rules according to the purpose of application. Based on the status quo, a fusion framework is presented. This framework can adapt to various algorithms and various fusion strategies, and can automatically determine the fusion algorithm, the fusion parameters and fusion rules according to the practical purposes of application. The strategies and rules to eliminate the ambiguity of fusion results are put forward. The rigorous mathematical form definition of these rules is given also. A simple instance system is developed to demonstrate the feasibility of the study conclusion.

Key words: fusion framework, multi-source image, feature saliency, ambiguity, rule definition

摘要: 多源图像融合是图像处理领域重要的研究热点,但众多的图像融合算法和融合规则各有优缺点,有必要根据应用的目的,建立自动选择融合算法和融合规则的多源图像融合框架。为此,提出一种能适应各种融合算法和融合策略的框架。该框架可根据应用目的,自动选择融合算法,对融合参数和融合规则进行自动调优。用数学形式定义用于消除融合结果二义性的策略规则,并完成简单的验证系统,证明上述方案的可行性。

关键词: 融合框架, 多源图像, 特征显著度, 二义性, 规则定义

CLC Number: