Entropy optimized image fusion: Using particle swarm technology and discrete wavelet transform

This paper proposes a method of image fusion whose input consists of multiple images with selective information which can be combined to produce a single substantive, complete image. Through this paper, we discuss enhanced image fusion using a Particle Swarm Optimization (PSO) based framework in the DWT domain. The Dual Tree Discrete Wavelet Transform (DTDWT) method used for decomposition of the source images in this paper is advantageous due to its increased shift invariance and an orientation selectivity property in comparison to the DWT method. In this paper, an optimized weighted averaging rule is applied for fusing the wavelet coefficients. An objective function, which is optimized based on entropy, forms the basis for the discovery of ideal weights for image fusion in the Particle Swarm Optimization method used. The weights thus calculated form the basis for the image fusion and the resultant is obtained by applying an inverse DTDWT. The fused image obtained as the end result is weighed up against other existing techniques like DWT, averaging, simple minimum and PCA using entropy as the basis.