智能相机的设计研究

Design and implementation of smart cameras

提出了一种整机采用DSP+FPGA+AVR结构,基于低级算法和高级算法相结合的智能相机,设计制作了实验用样机。并用其对钢珠缺陷进行了检测,对该相机的性能进行了较为系统的测试和研究,提出了如何改进结构和算法以提高性能。实验证明该相机的正确识别率较高,相机的设计方案有广泛的适应性,对软件的结构和算法稍作修改即可应用于不同的工作场合。

Smart cameras execute autonomous vision tasks and report events and data to a remote control system. A new structure of smart camera comprised of DSP ,FPGA and AVR is provided in this article. In order to improving the processing speed, tow pingpong structures are used in hardware design, and low-level and high-level video processing algorithms are discussed in this platform. All the algorithms are tested with the defaults of steel balls. The experiment results prove the smart camera＇s high reorganization and high processing speed. The structure of the smart camera can be used is many places with the little change of the algorithms.