Dam Intelligent Extraction Technology Based on UAV Remote Sensing Image

Abstract

China is rich in hydraulic resources. In order to make full use of water resources, many reservoirs have been built. Taking Guilin as an example, water conservancy facilities naturally become the focus and development object of Guilin. There are more than 130 large and small reservoirs in Guilin, which are used for irrigation, flood control, water supply, power generation and other purposes. Therefore, it is necessary to strengthen the safety warning of reservoir dams to ensure the safe operation. At present, it is an effective technical means to use UAV remote sensing data for intelligent extraction of reservoir dams and dam change monitoring. Firstly, this paper uses a variety of image preprocessing methods to reduce noise and external interference and improve the subsequent extraction rate and quality of the image. Secondly, LSC (Linear Spectral Clustering) superpixel segmentation combined with improved GrabCut algorithm is used to extract intelligent dams from reservoirs. The experimental results show that compared with SM algorithm and the original GrabCut algorithm, the proposed method for dam segmentation extraction effect not only improves the accuracy, but also greatly increases the running speed, especially for large resolution scenes.