基于改进Camshift的蒸汽发生器堵板螺栓视觉定位研究

为缩短强辐射环境下核电站蒸汽发生器堵板操作时间,研制了一种基于改进Camshift算法的核电站蒸汽发生器堵板螺栓视觉定位系统,协同操作机器人实现螺栓的自主定位、拆卸和安装工作。对经典Camshift算法进行了改进,考虑饱和度S分量上的特征信息,建立H-S二维直方图,提出通过颜色相似度的比较,将可分离性好的目标子区域结合机器人本体移动信息和机器人计算出的堵板螺栓信息作为更新Camshift算法的搜索窗口;并利用直方图补偿策略,提高螺栓定位精度。此外,在基于Cortex-M4的STM32高性能处理器平台上,验证了该算法的高效性;并将采集的视频信息及处理信息实时显示在7英寸LCD显示屏上。实验表明,该定位系统具有良好的鲁棒性及快速性,为嵌入式视觉定位工程开发提供了基础。

Study on bolt positioning of nuclear power plant steam generator based on a improved Camshift algorithm

In order to shorten the operation time of blocking plate in nuclear power plant steam generator under strong radiation environment, an improved Camshift algorithm was developed to locate the bolt of the steam generator and accomplish the automatic positioning, disassembly and installation of the bolt with cooperating robots. The classical Camshift algorithm was improved by considering the feature information of the saturation S component and establishing H-S two-dimensional histogram, combining good separability of the target sub-region by comparing the similarity of colors with robot body movement information and the blocking bolt information of the robot calculates , as a updating search window for the CamShift algorithm. And the histogram compensation strategy was used to improve the bolt positioning accuracy. In addition, the efficiency of the algorithm was verified on the Cortex-M4 based STM32 high-performance processor platform, and the collected video information and processing information real-time display on the 7-inch LCD screen. Experiments show that the positioning system has good robustness and fastness, which provided a basis for the development of portable visual positioning projects.