DWT BASED IMAGE FUSION USING SOBEL OPERATOR AND SINGULAR VALUE DECOMPOSITION

Abstract

Image fusion is the procedure by which two or more images are combined into a single image retaining the significant features
from each of the original images. The image fusion is the combination of multi source image information which is retrieved from
the different sensors. These images are then registered to assure the corresponding pixels are aligned properly. Afterwards they are
fused using any of the fusion algorithms. It then generates fused image which is more accurate, all- around and reliable. In this paper a
new fusion technique has been proposed based on the discrete wavelet transform. For edge and contrast enhancement we use singular
value decomposition and sobel operator. The input image is decomposed into four frequency sub band images using DWT, and then
singular value decomposition is applied on low-low sub band image, were as sobel operator is applied on low-high, high-low, and highhigh frequency sub band images. The proposed technique has been compared with existing pixel based image fusion techniques such as
simple averaging, simple pixel maximum and pixel minimum, principal component analysis, and discrete wavelet transform. We
describe experimental results and comparative analysis of performance metrices that shows the superiority of the proposed method
over existing methods.