基于SIFT算法的水电站出线场区域视频巡检

Video-based Patrol Inspection of the Outgoing Line Yard in Hydropower Stations based on SIFT Algorithm

针对出线场视频巡检缺陷识别准确率低的问题,提出了一种基于SIFT算法的水电站出线场区视频巡检方法。通过图像腐蚀算法对出线场巡检图片进行清洗,以消除噪声数据对图像数据识别造成的影响;然后采用Sobel算法对设备图像区域进行框定,并采用SIFT算法提取巡检图片特征。最后,采用卷积神经网络识别出线场设备缺陷。应用结果显示:该方法的缺陷识别准确率达到98.3%,验证了方法的有效性。

A video-based patrol inspection method based on SIFT algorithm is proposed for the outgoing line yard in hydropower stations to solve the low accuracy problem in defect recognition during the video inspection of the yard.Firstly,patrol inspection images of the outgoing line yard are cleaned with image erosion algorithm to eliminate the influence of noise data on image recognition.Then,Sobel algorithm is adopted to frame the device image area,and SIFT algorithm is used to extract the image features.At last,convolutional neural network is adopted to identify defects of the devices.The application results show that the defect recognition accuracy of the proposed method reaches 98.3%,which verifies the effectiveness of the method.