Multifocus image fusion method using discrete fractional wavelet transform and improved fusion rules

Abstract

In order to solve the problem of image blur caused by different focal points of multifocus images, this paper provides a noval multi-focus image fusion method which uses discrete fractional wavelet transform (DFRWT) and improved fusion rules. First, the DFRWT is utilized to decompose the source multifocus images at multiple scales. Then considering the different characteristics of low-frequency subband as well as high-frequency subband, the local region gradient information criterion is selected as the low-frequency fusion rule and the neighbourhood variance weighted average criterion is selected as the high-frequency fusion rule. Lastly, inverse DFRWT processing is carried out to obtain the composite image. The validity and advantages of our proposed method are verified through simulation results in visual qualities and quantitative evaluation.