A Survey of image registration algorithms for mosaicing

ABSTRACT

Photo-mosaicing is the technique to align and stitch images to produce today’s digital maps, satellite photos or panoramic mosaics. Algorithms for aligning images and stitching them into seamless photo-mosaics are among the oldest and most widely used in computer vision.

Most of the algorithms for image registration work by directly minimizing pixel-to-pixel dissimilarities. Calculating the dissimilarities of each pixel in the search areas introduce a great computation overhead. Thus minimizing the search area or reducing the number of pixels to be calculated is the intuitive way to lessen the computation complexity. The most common way is coarse-to-fine. A different category of algorithms works by extracting a sparse set of features and then matching these to each other. Featured-based approaches have the advantage of being more robust against scene movement and are potentially faster. Similarly, features extraction is the major reason leading to the high computation overhead. Comparison of the pixel-based and feature-based algorithms will be the conclusion.