| The cross correlation (CC), Gradient correlation and the mutual information (MI) are the most commonly used similarity measures for image registration. The CC is computationally very efficient and is highly robust in presence of noise but it results in an insufficiently sharp peak for precise object placement. On the contrast, MI results in much sharper peak but it is less robust in presence of noise and is computationally intensive. Hence, the main objective of this work is to make the CC peak sharper by using a judiciously chosen band pass filters that filter both the image and the template prior to correlating them. This proposed enhanced cross correlation (ECC) technique is particularly suited for the images having a wide range of frequencies in the frequency spectrum and for the cases in which the template and the image have high degree of similarity.;The work discusses the proposed ECC technique taking the above issues into concern and proves that the proposed technique is superior to 'Gradient Correlation' method for mugshot and iris images in the presence of Gaussian and/or salt and pepper (SP) noise. (Abstract shortened by UMI.). |
