Research On The Key Technology Of Two-DOF Pan-Tilt-Zoom System Based On Image

It is difficult for the traditional pan-tilt-zoom(PTZ) system driven by electromagnetic motor to meet the growing demand for security monitoring, robot, aerospace and other fields. The key bottleneck problem is high positioning accuracy?high dynamic performance and miniaturization of PTZ system. This paper presented a PTZ system driven by a single obelisk-shaped ultrasonic motor(USM), and made research on the closed-loop feedback control algorithm based on image technology.This paper applied a single multi degree-of-freedom(DOF) ultrasonic motor to drive directly the PTZ system, while traditional scheme needs several motors to realize multiple DOF rotation. The advantages of ultrasonic motor were able to solve the current drawbacks, such as heavy and complex structure. The USM mentioned above took advantages of local bending vibration as working mode and it was driven by the friction force between the stator and spherical base. After finite element modal analysis of the vibrator and driving characteristics experiment of the prototype, this paper established the whole electromechanical coupling model and made simulation of steady-state output characteristics of the PTZ system. For the sake of solving the complex nonlinear factors, such as time-varying parameters, dead zone and so on, the fuzzy PID control algorithm was used to improve the control performance, which was proved by the experimental results.Besides, the traditional PTZ is always an open-loop control system, the image technology was used to serve as closed-loop feedback. On the one hand, this paper only used the camera calibration experiment and Speed-up robust features point matching technology to mensurate the actual angle and speed output of the PTZ, which left out the additional position sensors. On the other hand, the image matching process was too slow to serve as feedback, so image-based visual tracking control method was proposed to meet the demands for real-time. This paper improved the performance of existed high-speed tracking with kernelized correlation filters algorithm based on feature fusion, adaptive scale and so on. The improved algorithm achieved balance between real time and robustness, whose practical effectiveness was verified by experimental results.The research shows that the design scheme is feasible which use ultrasonic motor as driving unit and image technology as feedback to build a light weight, small size and high precision intelligent PTZ system. The work in this paper provides theoretical basis and technical support for the follow-up study. At the same time, structure design, machining technology and related control method should be further optimized.