面向对象的 ADS80遥感影像分类在地理国情监测中的应用研究

摘要

ADS80 image has multi -band image information such as red, green, blue and near -infrared.In this paper, we takeADS80 image as data source, using object-oriented method for remote sensing images classification research.Research shows that the shape, texture, structure and other geometric characteristics of the image can be fully exploited, and the segmentation accura-cy of the image significantly improved with the method that combine ADS80 image and object-oriented classification.The image after classification has more abundant and accurate information, which has better application in the survey and monitoring of national geo-graphic conditions.%ADS80影像具有红、绿、蓝、近红外等多波段影像信息。本文以ADS80影像为数据源，采用面向对象的分类方法对遥感影像进行自动提取分类研究。研究表明，采用ADS80影像与面向对象分类相结合的方法，可充分发掘影像中的形状、纹理、结构等几何特征信息，显著提高影像的分割精度，使分类后的影像具有更丰富而准确的信息，在地理国情普查及监测中具有较高的应用价值。