关于清扫机器人自动异物检测研究

摘要

The HI-13 tandem accelerator within steel cylinder often shutdown due to falling fine impurities caused by high-voltage ignition,downtime by using the cleaning robot starting from the low end shielding box along the inner wall of the steel cylinder,through threshold segmentation method in image processing and clearing debris.Foreign object detection in a steel cylinder face illumination,pictures,misty,and target small easy issues such as missing,had a great impact on foreign object detection.In order to enable foreign object detection algorithm to better recognition rate,through the design of specialized collection process reduces uneven illumination,Retinex image enhancement removes haze,an improved Otsu threshold image segmentation by decreasing the unchecked,effective solution to the problems mentioned above.Experiments show that this algorithm in complex environments in a steel cylinder,you can maintain a high success rate of foreign object detection.%在清扫机器人对异物检测中,针对HI-13串列加速器钢筒内常因掉落细微杂质而引起高压打火造成停机,使用清扫机器人在停机后由低能端的屏蔽箱沿钢筒内壁进行,通过图像处理中的阈值分割法寻找并清理异物.由于在钢筒内进行异物检测面临光照不均、图片有薄雾以及目标微小容易漏检等问题,对异物检测造成了很大的影响.为了使异物检测算法能够保持较好的识别率,通过设计专门的采集流程减少光照不均,利用Retinex图像增强去除薄雾,利用改进的大津法选取阈值进行图像分割减少漏检,有效的解决了上述问题.实验表明,改进算法在钢筒内复杂环境中,可以保持较高的异物检测成功率.