Fusion methods based on dynamic-segmented morphological wavelet or cut and paste for multifocus images

Authors:
Phen-Lan Lin;Po-Ying Huang
Affiliations:
Department of Computer Science and Information Engineering, Providence University, Shalu, Taichung, Taiwan;Department of Computer Science and Engineering, National Chung Hsing University, Taichung, Taiwan
Venue:
Signal Processing
Year:
2008

Citing 4
Cited 2

Multisensor image fusion using the wavelet transform

Graphical Models and Image Processing
Image Processing - Principles and Applications

Image Processing - Principles and Applications
A simple and efficient algorithm for multifocus image fusion using morphological wavelets

Signal Processing
Information combination operators for data fusion: a comparative review with classification

IEEE Transactions on Systems, Man, and Cybernetics, Part A: Systems and Humans

Multi-focus image fusion based on salient edge information within adaptive focus-measuring windows

SMC'09 Proceedings of the 2009 IEEE international conference on Systems, Man and Cybernetics
The direct use of curvelets in multifocus fusion

ICIP'09 Proceedings of the 16th IEEE international conference on Image processing

Quantified Score

Hi-index	0.08

Visualization

Abstract

This paper presents a dynamic-segmented morphological wavelet fusion method (DSMWF) and a dynamic-segmented cut and paste fusion method (DSCP). Non-focus regions tend to spread around within multifocus images. The proposed methods firstly divide each multifocus image into segments and select each sharpest segment at the same location within all images as the ''focus segment'', based on DCT spectrum concentration on high-frequency sub-band. Each focus segment is further divided into smaller blocks having uniform visual complexity d based on Laplacian edge density. Finally, method DSMWF applies a single-level variable size morphological wavelet fusion method to each block of 2^dx2^d and method DSCP applies direct cut and paste of the sharpest block to each block of 2^dx2^d, respectively, to obtain a fused image. The experimental results demonstrate that (a) the PSNR of the fused image using DSMWF is 2-3dB better than that of MMWF in an average, (b) the occurrence of reconstructing both pixels with position error and underflow value is greatly reduced with DSMWF, (c) the performance of DSCP is much superior to that of both MMWF and DSMWF, and (d) block sharpness assessment based on DCT spectrum concentration on high frequency sub-band performs better than DWT and Laplacian edge for this application.