Abstract

Multi-focus imaging fusion is a technique that puts together a fully focused object from the partly focused regions of several objects from the same scene. For producing a high quality fused image, directional selectivity and invariance characteristics are important. The ringed artifacts, however, were inserted into a fused image because of a lack of invariance and misdirection. A multi-focus image fusion algorithm is proposed to resolve these issues, in conjunction with qshiftN dual-tree complex wavelet transform and modified principal component analysis. First, the source images are translated into the FP domain. It helps in the obtaining of the row frequency components and column frequency components. Then the row-frequency elements and column-frequency elements are combined with a dual tree-complex wavelet qshiftN to transform the origin frames. Dual tree complex wavelet transforms with qshiftN has demonstrated that it provides an effective transformation for multi-resolution imaging fusion with its directional and shift-invariant characteristics. To enlarge the effectiveness of the qshiftN dual-tree complex wavelet transform in frequency partition-based method, the modified principal component analysis (MPCA) algorithm is used. The proposed fusion approach has been tested on a numeral of multi-focus images and compared to various popular methods of imaging fusion. The experimental results indicate that in subjective performance and objective assessment, the proposed fusion approach could deliver better fusion results.