International Journal of Applied Mathematics and Machine LearningAuthor's: Liu Shuaiqi, Zhao Jie, Geng Peng, Liu Xiuling and Sun Yuchao
Pages: [21] - [34]
Received Date: August 7, 2014
Submitted by: Jianqiang Gao.
Combined with the advantage of more low-frequency sub-bands coefficients in nonsubsampled direction filter bank-dual-tree complex wavelet transform (NSDFB-DTCWT), a medical image fusion algorithm based on NSDFB-DTCWT is proposed to overcome the complexity fusion rules in image fusion algorithm based on the multi-scale multi-resolution transform. First, NSDFB-DTCWT is utilized for decomposition of the source images, and then the maximum of the absolute value of all frequency coefficients of decomposition is selected to construct fusion image. Finally, the fusion image is gained after inverse NSDFB-DTCWT. The algorithm can both make effective use of the contour information of source images and preserve the details of source images. What’s more, its fusion strategy is very simple and has low computation cost. Experimental results demonstrate that the proposed method can not only obtain good visual effect, but also improve its objective evaluation.
fusion, medical image fusion, NSDFB-DTCWT, complex wavelet transform, nonsubsampled direction filter bank formation.
