基于图像处理技术的险情识别智能巡检机器人的研究  

Research on Risk Recognition of Intelligent Inspection Robot based on Image Processing Technology

作　　者：李福鹏 梁国坚 杜鑫峰 陶熠昆 朱玲芬 

机构地区：[1]广东电网有限责任公司中山供电局,广东中山528400 [2]浙江国自机器人技术有限公司,浙江杭州310053

出　　处：《自动化与仪器仪表》2017年第6期10-12,共3页Automation & Instrumentation

摘　　要：随着智能电网的不断发展,变电站的无人值守和24小时监视不仅可以减少大量的人力和物力的投入,还可以降低人为因素而引起的事故。针对智能巡检机器人,提出了基于图像处理技术的智能巡检机器人对火灾险情进行识别与监控。在火焰的识别过程中通过对火焰的多特征进行了提取和分析,包括火焰的颜色特征、火焰的形状特征和火焰的变化特征。在火焰的多特征提取中也采用了边缘梯度掩膜、膨胀梯度掩膜和填充空洞等技术对图像进行预处理。实验结果显示基于图像处理的火焰识别技术可以准确地提取火焰的特征,极大地提高了电力系统中巡检机器人的火灾险情识别能力。With the developing of the smart grid, the unattended operation and 24 hours monitoring of substation not only can reduce the investment of much manpower and finance, but also the accident that causes by the human factor. For the intelligent inspection robot in the electrical power system, this work proposes the recognition and monitoring of fire risk based on image processing technology. The multiple feature detection and analysis of flame is used in the recognition of flame. The features include the color of flame, shame feature and change characteristic. The edge gradient mask, expansion gradient mask and cavity filling is also used in the multiple feature detection of flame. The experimental results indicate the flame recognition based on image processing can detect the flame accurately, and improve the recognition of fire risk based on the intelligent inspection robot in the electrical power system.

关 键 词：智能电网 巡检机器人 火焰识别 图像处理 多特征提取 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]