摘要
针对目前电力机房巡检机器人仪表识别精度不高的问题,提出了一种新的仪表检测识别方法。该方法分为两个部分,第一部分为仪表检测过程:提出了一种基于相似性度量的目标检测方法。首先,采用Faster RCNN网络生成一系列候选集,然后由高置信度的目标区域建立特征模板,再根据特征相似性对低置信度的目标区域进行判别,最后将筛选结果和特征模板作为检测结果,从而实现对仪表类型的识别与定位。第二部分为指针式仪表读数识别过程:采用自适应Canny算法进行边缘检测,然后采用八领域轮廓跟踪算法串联边缘点获取轮廓,并通过分析轮廓形态,提取指针和刻度线段。最后通过改进的刻度修复算法解决光照下刻度缺省问题,从而建立完备的刻度坐标系,再根据坐标系中的指针与刻度之间的相对位置得到仪表读数。实验结果表明,本算法对于光照条件、拍摄角度、电力机房环境干扰具有较好的鲁棒性,提高了仪表识别的准确性和实时性。
Aiming at the problem that the instrument recognition accuracy of the patrol robots in electric power room is not high,a new method of instrument detection and recognition is proposed.Th method is divided into two parts.The first part is the process of instrumentation detection. A target detection method based on based on similarity is proposed. Fristly Faster RCNN is used to build a series of candidate proposals.Then the anchors create feature templates by target areas with high confidence,and make a further selection in the low-confidence proposals according to the feature similarity.Finally,the detection results are composed of the reserved proposals and the templates to achieve recognition and positioning of the meter type. The second part is the process of pointer instrument recognition. An adaptive Canny algorithm is proposed for edge detection.Then the eight-field contour tracking algorithm is used to connect the edge points to obtain the contour,and by analyzing the contours shape,the pointer and the scale line segment are extracted.Finally,the calibration default problem under illumination is solved by the improved calibration repair algorithm,thus a complete scale coordinate system is established.The instrument reading is recognized according to the relative position between the pointer and the scale.The experimental results show that the proposed algorithm has good robustness to illumination conditions,shooting angles and environmental disturbances in the power room,which improves the accuracy and real-time performance greatly.
作者
司朋伟
樊绍胜
Si Pengwei;Fan Shaosheng(School of Electrical & Information Engineering,Changsha University of Science & Technology,Changsha 410114,China)
出处
《信息技术与网络安全》
2019年第4期50-55,共6页
Information Technology and Network Security
