EXPLORE PUBLICATIONS BY COUNTRIES

Image Registration, Normalized Cross Correlation, Steerable Pyramid, Contourlet Transform.

The main objective of image registration is to match two or more images captured at different times by different sensors or by different angles or from different viewpoints. Image registration has become a crucial step in most of the image processing tasks used in various areas. It is a key technology which is applied for computer vision, remote sensing, image processing, medical image analysis and other fields. Medical image registration is used to find a spatial transformation to match all the anatomical points and diagnostic points on the image. In general, the majority of registration methods consist of the following four steps: feature extraction, feature matching, transform modeling, and finally image resampling. The accuracy of a registration process is highly dependent to the feature extraction and matching methods. Cross Correlation is the basic statistical approach to image registration. It is used for template matching or pattern recognition. Template is considered as a sub-image from the reference image, and the image is considered as a sensed image. The objective is to establish the correspondence between the reference image and sensed image. It gives the measure of the degree of similarity between an image and template, which can be used for image registration. Normalized Cross Correlation (NCC) method is improved by using Feature Based Method. Image are effectively represented by any of its feature such as edges, points, curves etc. and these features are effectively used for image registration. Images are applied with the filters to extract edges. Post that NCC is done to find the sharp point on NCC plot. This method restricts us with only monomodal images. For multimodal images we have used Mutual Information as a measure of similarity. A widely used measure is Mutual Information (MI). This method requires estimating joint histogram of the two images. Experiments are presented that demonstrate the approach. The technique is intensity-based rather than feature-based. Mutual Information is effectively used as similarity measure between two images which can be monomodal or multimodal. Mutual information is consider as a measure of how well one image explains the other; it is maximized at the optimal alignment. To make this more effective Contourlet transform is used. Contourlet is a recent development of transform theory as an improvement of wavelets. It is a multiscale and multidirectional, two dimensional transform. It is a combination of Laplacian pyramid and directional filter bank. The discrete contourlet transform has a fast iterated filter bank algorithm that requires order N operations for N-pixel images. the contourlet transform effectively captures smooth contours that are the dominant feature in natural images. This leads us to more efficient image registration.