Image registration quality assessment with similarity measures - A research study

Image registration has evolved independently, and occupied many research areas starting from underwater to earth to space. Selecting a most suitable similarity measure from the existing image similarity models is a challenging research activity in the digital image processing area. Even though there are many image registration models, similarity model is selected for this research study, because of its broad usage. Because of many similarity measures in the literature, there is a need to summarize few of them, in order to use three of them to evaluate the image registration quality through two performance metrics. For this purpose, a research study is carried out and summarized sixteen measures and three of them are computed; evaluated with PSNR and MSE performance metrics; Rotation and translation are two spatial transformations implemented using single modality cameraman image; and an attempt is made to assess the image registration quality with practical implementation. Finally, the complexity and research potential is emphasized in the challenging area of Image Registration.