Detection Of Power Poles Based On Geographical Location Information

In recent years, autonomous inspection of overhead power lines using the rotor flying robot has been developed rapidly. The robot through airborne cameras and navigation equipment has access to massive overhead power line inspection video/image, as well as their geographic location and power towers. Only be power towers quickly and accurately determined in the mass inspection images with contribute to the position of the robot’s autonomous and efficient inspection as well as the next point of fault detection. For the problem, this paper makes full use of relative position information of both robots and power towers and presents the power tower detection method based on the fusion of the geographic information.To solve this problem, this paper addresses two key questions:First, getting the spatial relationship between the flying robot and the power tower, the area of the power tower can be pre-estimated in the image, which is called as the GI area; Second, in the GI area, we use the spatial relationship to restrict the detection region of DPM, thus the power tower can be detected accurately.For the area estimation of the power tower, this paper focuses on the spatial relationship between the flying robot and the power tower, and navigation equipment positioning errors. To solve these problems, we use a common geometric transformation method to get their relationship. Because navigation equipment positioning errors generally meet the Gaussian distribution, based on the relative position, this paper focuses on the Gaussian error which impacts the GI area.For the power tower detection, this paper further studies of deformable part model (DPM) method in the power tower detection application. The DPM method can effectively solve the light, perspective, background conditions and different shapes such as electricity tower electricity tower detection. In order to further improve power tower detection speed and precision, the paper takes full use of the GI area to achieve a rapid and accurate power tower positioning. Experimental results show that the flying robot capturing video images do the test to verify the integration of geographical information-deformable part model approach (GI-DPM), while power tower remains almost unchanged detection accuracy, the detection speed is nearly two-fold than DPM’s.For further study about advantages and disadvantages of the position information and the deformable part model, the proposed method called fusion of geographic information of the power tower detection, which completes fast, accurate positioning in the image for the next point of failure. The method also ensures promotion and application of autonomous polling overhead power line.