基于深度学习的无人机环境污染目标识别技术研究  

Research on Identification Technology of UAV's Environmental Pollution Target based on Deep Learning

作　　者：吴琼[1] 雷松泽[1] 王艳红 Wu Qiong, Lei Songze, Wang Yanhong(Xi＇an Technological University, Xi＇an 710021, Chin)

机构地区：[1]西安工业大学,陕西西安710021

出　　处：《环境科学与管理》2018年第7期91-94,共4页Environmental Science and Management

基　　金：陕西省教育厅专项科研计划项目(编号:17JK0364);陕西省工业领域重点项目(编号:2016KTZDGY4-09)

摘　　要：为了优化环境污染监测,需要对环境污染目标区域进行有效识别,提出基于深度学习和无人机技术的环境污染目标区域识别方法,采用无人机机载空间扫描方法进行图像采集,对采集的环境图像进行污染区域的边缘轮廓检测和图像分割处理,采用深度学习算法对环境污染区域的图像进行自适应分块标记和识别,实现对环境污染区域的图像检测和三维区域识别。采用该方法进行环境污染区域的目标识别准确性较好,对环境污染区域分块识别的精度较高,较强。In order to optimize the monitoring of environmental pollution,it is necessary to identify the target area of environmental pollution effectively. A method of identifying the target area of environmental pollution based on depth learning and UAV technology is proposed. The airborne space scanning method of UAV is used to collect images,and the environmental images are detected and segmented. The deep learning algorithm is used to self-adaptively label and recognize the image of the polluted area,and the image detection and 3 D recognition of the polluted area are realized. The simulation results show that the method has good accuracy and robustness in environmental pollution area recognition.

关 键 词：深度学习 无人机 环境污染 目标识别 

分 类 号：X22[环境科学与工程—环境科学]