PENJEJAK OBJEK DENGAN BEBERAPA KAMERA MENGGUNAKAN ALGORITMA THREE STEP SEARCH

Abstract

Abstract In this paper we address the development of an object tracking system with multiple fixed cameras using motion estimation. Three step search algorithm is used to estimate object movement that across camera. This algorithm will process two frame f(t-1) and f(t), where f(t-1) represents reference image, and f(t) represents current image at one camera to produce motion vectors value. Sequence of images that arrive from cameras will be analyzed, then motion vectors and number of moving blocks are obtained. The motion vectors imply that object is moving to right, left, up, or down. Multiple cameras are placed as nodes in a map area having cardinal coordinate. View direction of each camera according cardinal coordinate is whether south, east, north, or west. Based on motion vectors value and number of blocks that are moving from each camera, our algorithm will respond by moving the view from one camera to another, tracking objects that cross the area covered by cameras. Key words : object tracking, three step algorithm