Target Registration of Electric Power Installation Based on Hybrid Network Model Under the Application of Intelligent Inspection Application

In the on-line monitoring of power equipment, it is necessary to identify the appearance image of the target, and the image fault recognition must first carry out the accurate matching of the image, so there is an increasingly urgent demand for the application technology of the visual positioning of the target detection equipment. In this paper, a hybrid network model based target seeking algorithm for power equipment is proposed, which combines two research routes: template matching of twin network and image registration based on feature point matching. In the aspect of network depth, the algorithm uses less feature size compression to reduce pooling. In order to ensure that the size of the feature map is the same as that of the original image, the detection resolution can be reduced. Methods the sub neighborhood partition of feature descriptor is improved to simplify the algorithm complexity. In order to avoid over fitting of the model, dropout strategy was used to ignore some neurons randomly during training. At the same time, the exponential descent learning rate curve is used to consider the convergence speed of the model in the early stage and the convergence in the later stage. The experimental results show that the algorithm can achieve the goal-seeking task of power equipment reliably and effectively in the complex and changeable substation detection environment.