Image Mosaicing and Registration

Image registration is the process of overlaying images (two or more) of the same scene taken at different times, from different viewpoints, and/or by different sensors. The registration geometrically aligns two images (the reference and sensed images). Many applications require a panoramic image of the object but due to limitation of apparatus, only a single image of the object in large scale can be captured from the global views. Therefore, a series of images of the object are collected with overlapped areas and mosaiced to construct a panorama which can be stored in files or databases and be viewed quickly. This project has been undertaken keeping in view that all imaging systems require some form of registration. A few examples are aligning medical images for diagnosis, matching stereo images to recover shape, comparing facial images in a database to recognize people etc. Given the difficulty of registering images taken at different times, using different sensors, from different positions, registration algorithms come in many different shapes and sizes. Given the wide range of imaging applications, there are many different types of registration algorithms. In this project we are using an algorithm based on Correlation. For Automatic Image Registration Applications, we are detecting the features like edges by using Sobel Edge Detection Algorithm. For matching the features we are first Segmenting the image file in terms of different blocks and then applying the Hierarchical matching to create pyramid of block. Finally we are applying correlation based matching starting from the top level of pyramid. For Image Mosaicing Applications, we have to find out the control points. For finding control points we have to find out overlapping portions in both left profile and right profile of image. If both left and right profile have sufficient part common then 1 perform Segmentation to divide the image in terms of block of data and then for matching developing pyramid of blocks by using Hierarchical matching. Otherwise take a suitable pixel block size say of 32 x 32 pixels block from right profile image and search for the exact location of that 32 x 32 pixels block in the left profile image.