Digital Image Separation Based on Joint PDF of Mixed Images

Abstract

In a real environment, Image separation is very difficult task from observed fused and merged image some fingerprint application and cloud detection and measurement .It is frequently arising problem in image processing field .Image separation is more typical case of image de-noising where more than one image are to be reconstruct from a single observation. The whole problem resembles, the task a human can solve fused image separation problem where using two images (two finger print). We can separate the overlapped finger print with different technique. In this thesis we examine the image separation problem basis that two fused image are independent to each other. The technique of image separation aims to estimate the original image and mixing matrix using only the fused image .we are using two technique for estimate the mixing matrix (1) Estimate the mixing matrix ,given an estimate fused image .(2) Estimate the original image given an estimate mixing matrix. In this thesis we have some prior information about the image on the basis of information of image we can estimate the mixing matrix with help of different graphical scatter plot of two fused image we are trying to estimate mixing coefficient Image separation is based blind source separation . BSS has been applied image separation .biomedical image processing, remote sensing, communication field, data mining, neural network, exploration seismology. Source separation problem can be formulated using ICA technique