A new image fusion performance measure using Riesz transforms

作者：Zhang Xiaoli; Li Xiongfei^*; Feng Yuncong

来源：Optik, 2014, 125(3): 1427-1433.

DOI：10.1016/j.ijleo.2013.09.020

• Image Fusion
• Performance Evaluation
• Visual Similarity
• Riesz Transforms
• Correlation Analysis

Image fusion is an alternative for enriching visual experience by combining images with complementary information. Assessment of image fusion performance would prove invaluable for parameter optimizing, fusion scheme selection and testing. This paper presents a new image fusion performance measure, which consists of two parts: the first one is a similarity measure for predicting the amount of information transferred form original images to fused image in Riesz domains: the other is a measure for characterizing the contrast of the fused image. Considering that different morphological components share different importance in Human Visual System, a gradient based image content partition algorithm is adopted to segment original images and fused images into three parts, and according the partition results, different weights are given to different pixels in the process of similarity measure calculating. Experimental results demonstrate the superiority of our measure compared with conventional measures in terms of computation complexity and accuracy.

• 出版日期2014
• 单位吉林大学