Small Unmanned Helicopter Visual Tracking And Recognition System Implementation

Small unmanned helicopter general concern in civil, industrial and military fields, and the airborne vision technology provides a necessary condition for the UAV to carry out the above tasks successfully. In this paper, visual tracking and identification system was designed for the small unmanned helicopter to enable it to search and secure landing ships at sea. This paper describes the basic principle of digital image processing and recognition algorithm for ships and landing mark. And also builds the system hardware test platform to laid the foundation for UAV to track and recognize the target.In terms of target recognition problem, firstly, this paper analyzed target tracking principle based on visual for UAV. And it briefly summarized varieties of methods which can process digital image and describe the target feature, including the image gray scale, binary and morphological image processing methods as well as the summarizes of color, shape and texture features. Secondly, this paper focuses on the contour edge detection and active contour model algorithm, and comprehensive application of two algorithms to extract the contour closest to the real ship. According to the target shape feature we extracted to completed the ship identification and tail positioning. Finally, in order to extract feature point coordinates landing mark on corner detection and matching algorithm research. At the same time, it is possible to solve the displacement and speed PTZ cameras need to be adjusted according to the position of flag carrier landing in the image, through leaving the target range in the center of the lens to complete the UAV Landing flag tracking process.Additionally, this paper discusses the hardware selection and construction of UVA vision system in detail and the algorithm processes of tracking and recognition, and do the simulation experiments on this experimental platform. According to the simulation results, the algorithm of this paper is correct and feasible to achieve a vision system for airborne target tracking and recognition process.