Multispectral Remote-sensing Image Vector Fusion Based On MultiWavelet Analysis And Its Simulation

Multispectral remote-sensing images offer high spectral resolution and much more information, provide potential superiorities in recognition and classification of the objects in earth's surface. Recently, the multispectral images fusion technology has attracted more and more attentions. In this paper, theories on wavelet and multiwavelet are both reviewed, and the main emphases are on the image vector fusion algorithm based on multiwavelet analysis and its application to the fusion of multispectral images.Firstly, multiwavelet analysis theories are systematically reviewed. In the real domain, the finitely supported, orthogonal, symmetric nontrivial scalar wavelet bases do not exist, while the multiwavelet offers the finite support, symmetry, orthogonality simultaneously and the wavelet theories are extended to vector field. Realizing and analyzing the existed homogeneous algorithms and considering the vector characteristics provided by the coefficients of the multiwavelet transformed image, pixel-based and region-based scalar fusion schemes are extended to vector case and a novel fusion algorithm is proposed as well. The new algorithm is based on vector fusion scheme in multiwavelet domain, which makes sufficient use of the correlation among the components of multiwavelet transform coefficient vectors to improve fusion quality. The original algorithm is carried out with emphases on the novelty of the fusion algorithm and the demonstration by using 64 bands real multispectral image. The experiment results demonstrate that the proposed multiwavelet vector fusion algorithm can obtain both better subjective vision characteristics and better objective evaluation indices and outperform the wavelet scalar fusion scheme. In addition, the multi-focus image fusion experiment results demonstrate the applicability of the proposed algorithm. In the end, a simulation software of multispectral remote-sensing images fusion based on multiwavelet analysis is achieved using VC++6.0.