摘要
障碍物检测识别是高压输电线路自主除冰机器人的关键技术之一。针对220 kV输电线路特殊的机器人工作环境,提出一种基于视觉的障碍物识别方法。首先对拍摄的障碍物图像进行中值滤波、膨胀腐蚀等预处理,经OTSU阈值优化计算后,用小波模极大值算法提取图像边缘。然后计算障碍物边缘图像的联合不变矩特征,再把矩特征输入小波神经网络进行障碍物图像的分类识别。并选取防震锤、悬垂线夹、耐张线夹三类障碍物做识别试验,还把小波神经网络与普通BP神经网络识别性能进行了比较,实验表明:以联合不变矩作为障碍物识别特征具有良好的可靠性和稳定性;小波神经网络识别分类的性能良好,比普通BP神经网络具有更快的收敛速度和更高的识别精度。
Obstacle recognition is one of the key techniques for the deicing robot on high voltage transmission line. According to the structure of 220 kV single split transmission line, an intelligent method of obstacle recognition based vision is put forward. At first, the obstacle images are preprocessed by middle filtering, expansion and erosion. Secondly, images are binaryzed by using OTSU algorithm and the edges of the images are detected by using wavelet modulus maximum algorithm. Thirdly,the united moment invariants of obstacle images are calculated, then wavelet neural network is proposed for recognizing the obstacles after moment features input to neural network. In experimental stage ,we select some obstacle images to be study group, such as counterweight, Strain clamp and suspension clamp. the experiments show that united moment invariants was used as feature vector is stable and reliable;the wavelet neural network has good recognition performance and it has higher precision and quicker convergence rate than normal BP neural network recognition method.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2011年第9期2049-2056,共8页
Chinese Journal of Scientific Instrument
基金
国家支撑计划项目(No.2008BAF36B01)
国家863项目(No.2008AA04Z214
No.2007AA04Z244)资助项目
