基于DCNN的输电线路隐患区域图像识别方法

Image recognition method of hidden danger area of transmission line based on DCNN

针对线路隐患区域图像识别效果较差以及识别时间上升等问题,提出一种基于DCNN的输电线路隐患区域图像识别方法。通过改进的维纳滤波算法进行图像去噪,借助像质评价函数,建立图像标记工具,手工标记和类型划分数据集中没有训练过的数据,将分类结果输入到深度卷积神经(DCNN)中进行训练,联合改进的SVM分类器,建立输电线路图像分类模型,完成输电线路隐患区域图像分类与识别。实验测试证明,所提方法能够有效降低识别时间,获取更加满意的识别效果,为输电线路隐患识别提供保障。

Aiming at the problems of poor image recognition effect and rising recognition time of transmission line hidden danger area,a transmission line hidden danger area image recognition method based on DCNN is proposed.The improved Wiener filter algorithm is used for image denoising.With the help of image quality evaluation function,the image marking tool is established.The untrained data in the data set are manually marked and classified.The classification results are input into the Deep Convolution Neural Network(DCNN)for training.Combined with the improved SVM classifier,the transmission line image classification model is established,complete the image classification and recognition of transmission line hidden danger area.Experimental tests show that the proposed method can effectively reduce the identification time,obtain more satisfactory identification effect,and provide guarantee for transmission line hidden danger identification.