Recent Trends in Communication, Computing, and Electronics by Unknown

Author:Unknown
Language: eng
Format: epub
ISBN: 9789811326851
Publisher: Springer Singapore

Keywords

Image enhancementImage fusionDDWTPrincipal component analysis

1 Introduction

Fusion is a process of combining two or more distinct entities, similar is in the case of image fusion. Say, there are two images image A and image B containing some information, both images when analyzed individually possess no or less information but, after we fuse them together in resultant fused image; we find that resultant image gives some useful information. Image fusion is a technique which can be used in many fields such as military, security and surveillances, computer vision, astronomical images, medical imaging, navigation. Image fusion can be done from images with same modality or images from different modality as well; e.g., CT and MRI can also be fused together to obtain relevant and useful information [1]. If we try to understand in simple terms, addition of best pixels from two different images into third image could give relevant information and thus be called image fusion. This could be a simple technique of image fusion but, in order to get more relevant information from them, we have to work little harder, so we have to find better ways to fuse images to get more relevant information. Therefore, there are number of methods proposed in the field of image fusion in both the spatial and frequency or transformation domains, but the methods of image fusion should have the properties; that is, it transfers the advantageous information from input images to the integrated image, while making no loss of information during processing, and also should not add any artifacts in the resultant fused image. The spatial domain methods combine information in spatial space according to certain fusion rules like minimum-maximum, average. In contradiction, the frequency domain methods conduct multi-scale/multi-resolution decompositions of image before combining information from them and then the fusion rules are designed on basis of frequency context. Basically, it includes three stages, i.e., decomposition, coefficients fusion, and reconstruction [2]. The image fusion method of the spatial domain in the initial stages ordinarily acquires an image fusion scheme that is based on block, where each pair of blocks is combined with an intended activity level measurement such as summation of modified pixel values by applying Laplacian function [3, 4]. In the last few years, pixel-based image fusion methods have been proposed in [5, 6], deploying gradient information. However, spatial domain methods were not able to achieve all the properties stated above because spatial domain methods suffer from spatial distortion, and also the problem of blocking artifacts arises from the block-based spatial domain methods.

Therefore, to meet the requirement stated above for the image fusion, transformation domain methods were proposed in which images are transformed to frequency domain, i.e., wavelet transforms and pyramid methods [2, 7]. These methods firstly decompose the input images into approximation coefficients having variations in large scale and seriate detail coefficient having variations on small scale, after that fusion rules are applied on them for integrating them into a single fused resultant image. An image pyramid is a sequence of images in

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.