Implementation of Features Based Image Mosaicing Algorithm

Abstract

The work presented in this memory, consists of features based image mosaicing and object tracking algorithms from multiple overlapped views. An image mosaic is a combination of a sequence of overlapped views, and it is a powerful mean of obtaining a larger view of a scene than the available within single views. Object tracking is the process of locating a moving object (or multiple objects) over time using a camera; these techniques are generally used on several images; like remote sensed images, biomedical images, satellite images or other digital images. Our developed algorithms are based mainly on key-points detection and matching in order to find geometric transformation to register overlapped images on each other. Correlation and descriptors based features matching techniques were implemented and compared in Matlab platform, then, we implemented our algorithms on LabVIEW platform, in which we have created an interface to facilitate the things for users. The efficiency of our implementation was verified from the quality of the obtained results using real images