Power Line Scene Recognition Based on Convolutional Capsule Network with Image Enhancement

With the popularization of unmanned aerial vehicle (UAV) applications and continuous development power grid network, identifying line scenarios in advance is very important for safety low-altitude flight. The scene recognition (PLSR) under complex background environments particularly important. environment lines usually mixed by forests, rivers, mountains, buildings, so on. In these environments, detection slender difficult. this paper, a PLSR method backgrounds based on convolutional capsule network with image enhancement proposed. edge features scenes guided filter are fused framework. First, used to enhance order improve background. Second, extract depth hierarchical lines. Finally, output layer identifies non-power scenes, through decoding layer, reconstructed scene. Experimental results show that accuracy proposed obtains 97.43% public dataset. Robustness generalization test it has good application prospect. Furthermore, can be accurately extracted from module.