基于eCognition的高分辨率卫星影像的车辆检测实验设计

摘要

设计了利用eCognition进行高分辨率卫星影像的车辆检测实验的步骤,首先对目标影像选取关键参数进行多尺度分割;其次利用最小距离分类器进行分类,实验设计中将目标对象主要分为3类,即道路、汽车及其他类;最后以Worldview-1高分辨率卫星影像进行了车辆检测的实验.实验结果表明,所设计的实验过程能达到车辆检测的目的并且精度较高,实验过程的简单直观,可用于交通信息技术及应用专业本科实验教学课程.%A procedure for the experiment of the vehicle detection with high resolution satellite images by using eCognition is designed.Firstly,the key parameters of target images are segmented by multiple scales.Secondly,the minimum distance classifier is used to classify the target,which can be divided into the following three kinds: road,automobile and other category.Finally,the experiment of the vehicle detection is conducted with Worldview-1 high resolution satellite images.The experimental results show that the experimental process can achieve the purpose of vehicle detection and get high precision,and the experimental process is simple and intuitive.This experiment can be used in the undergraduate course of experimental teaching of transportation information technology and application majors.