All-in-Focus Imaging using qshiftN DTCWT and LP in the Frequency Partition Domain

Abstract

Image fusion utilizes fusion rules to combine multifocus images of two or more images of the same scene, which produces a fully focused image. The fusion of images provides vital information about the objects, which is more useful for visual perception and medical diagnosis. Directional selectivity and shift invariance characteristics are found to be very important in the image fusion process. Due to poor shift invariance and directionality, traditional wavelet-based fusion methods create ringing distortions in the fused image. In this paper, a novel multifocus image fusion algorithm based on the quarter shift dual tree complex wavelet transform (qshiftN DTCWT) and laplacian pyramid (LP) in the frequency partition (FP) domain is proposed. Firstly, the source images are decomposed into row and column coefficients with a frequency factor in the FP domain. Secondly, due to its directional and shift invariant characteristics, row and column images are collectively fused with qshiftN DTCWT, which proved to be a very efficient multiresolution transform for image fusion. Finally, to increase the effectiveness of the qshiftN DTCWT in the FP-based method, the LP algorithm is used to produce an all-in-focus image. The result of the proposed strategy reveals the good quality of the fused image due to its better directionality and also reduces artifacts with its shift invariance property. The statistical results also show a better performance of the proposed methodology than that of traditional fusion methodologies.