【摘要】 针对220kV单分裂线路的结构特点,提出了一种基于小波矩的障碍物智能视觉识别方法.该方法采用Ostu算法二值化图像,采用小波模极大值算法提取图像边缘.通过提取障碍物边缘图像的小波矩,来得到一组局部最优的小波矩特征值,并在此基础上用小波神经网络进行障碍物的识别与分类.实验结果表明:所提出的方法能有效地识别高压输电线上的防震锤、悬垂线夹、耐张线夹等障碍物,并具有比普通3层BP神经网络方法更高的精度和更快的收敛速度.更多还原

【Abstract】 According to the structure of 220 kV single split transmission line,an intelligent visual method of obstacle recognition based on wavelet moments is put forward.It binarizes obstacle images by Ostu algorithm and detects the edges of the images by wavelet modulus maximum algorithm.Then,a set of locally optimum wavelet moment features are selected by calculating the wavelet moments of the edge images.Based on these features,a wavelet network is presented for obstacle image recognition and classification.The experiment results show that the obstacles such as counterweight,strain clamp and suspension on high voltage transmission line can be effectively recognized by the proposed method,and the proposed wavelet network method has higher precision and quicker convergence rate than the common three-layer BP(backpropagation) neural networks.更多还原