基于CMOS的机器人视觉系统设计

Design and implementation of robot vision based on CMOS

摘要:

针对机器人系统任务的特点，设计了一种单目视觉系统。采用CMOS图像传感器，通过CPLD形成控制时序，搭建了30万像素分辨力的硬件系统。基于简化的HSI颜色模型，并采用抽样方法，实现了对象状态的判断；系统采用高性能DSP器件，在进行图像处理的同时实施机器人的运动控制。系统软件算法在Matlab下进行了仿真并移植于DSP。实验结果表明，该系统实现了特定对象的状态判别功能，并具有良好的实时性。

Abstract:

According to the characteristics of robot duty，this study designed a monocular vision system. With Complex Programmable Logic Device(CPLD) to form the control timing sequence，a Complementary Metal Oxide Semiconductor(CMOS) image sensor was adopted to build up a hardware platform with a resolution of 0.3 M pixel. Based on the simplified Hue Saturation Intensity(HSI) color model, the object status could be judged with a sampling method. The high-performance Digital Signal Processing(DSP) processor used could provide image processing and motion control functions simultaneously. The algorithm was verified with Matlab and then ported to DSP. Experimental result has shown that the system can implement the status judgment successfully.