A critical review on Image Mosaicing

Abstract

Image Mosaicing is a method of combining two or more images into a single set of large image. Mosaic image is created from a partial views to obtain large view of a scene and it has many applications. This paper discusses a review on distinct algorithms of feature detectors and descriptors. Typically five phases are included in an Image Mosaicing. They are: Feature Extraction, Image Registration, Homography Computation, Image Warping and Blending. The Feature Extraction can be done by using multiple corner detector algorithms. Many feature Descriptors are discussed with necessary equations. Homography can be estimated by different methods, like RANSAC-Random Sample Consensus, LMS-Least median of squares and Hough Transform. Image Warping used for creative purposes and corrects the distort images. Image Blending is the technique to minimize artifacts of the mosaic imageby modifying the image gray levels to appear richvicinity of an output image