Object Tracking in Video Sequences

Praćenje objekata u videozapisima jedna je od važnih primjena računalnog vida. U ovom završnom radu analizirano je praćenje objekata, a odabrani algoritmom je camshift algoritam koji koristi informaciju o boji objekta, koji modelira histogramom. Taj histogram se iskorištava za izračun funkcije koja opisuje vjerojatnost položaja objekta u svakoj pojedinoj slici u videozapisu. Određivanjem ekstrema te funkcije pronalazi se točna lokacija objekta.Object tracking is one of the most important fields of computer vision. In this final thesis we analyzed object tracking and chosen algorithm for this purpose was camshift. Camshift algorithm uses information about color of the object which is modeled by histogram. This histogram is used for calculating function which describes probability of object appearing at each pixel of the video frame. By determining the extreme of these functions, the exact location of the object can be found