Diseño e implementación de un sistema autónomo de seguimiento de características

Abstract

In computer vision, one of the most common problems is analyze the motion of an objective in an image sequence. To solve the tracking of a target the present project has combined detection and tracking in a static visual field of a camera with an electro-mechanical system. For comply this goal has been used computer algorithms of artificial visión like Canny, Shi-Tomasi, Lucas Kanade Algorithm; which ones has been applied and has been used to find the target information. In addition to pointing the camera on the target by controlling required angles of Euler to generate a two-dimensional movement, the controller is designed with the purpose of maintaining the centroid of the features of the objetive focused on the center of the camera visual region. The feedback is performed by acquiring new images, which are subjected to preprocessing before calculating the necessary error for generating a control action. All developed algorithms running on an embedded computer system called UDOO, which one allows to have a mobile and lightweight module therefore this project could be used and installed as a payload of a UAV.