遥感影像的分类与识别技术在土地资源调查中的应用

CLASSIFICATION OF RS-IMAGE AND RECOGNITION TECHNIQUE UTILIZE IN LAND AND SOURCES INVESTIGATING

遥感图像记录了地物在空间域、时间域、光谱域的变化信息。利用图像的分类技术 ,能够识别土地利用类型。计算机遥感分类识别原理 ,是利用地物的光谱能量特征差异性和结构特征差异性来识别地物信息。根据北京某地遥感图像实例资料 ,将土地利用类型分为监督分类、非监督分类、最大似然法分类、神经网络分类等。不同的分类方法有各自的特点且分类结果也有一定的差别 。

RS-images record geophysical changing information in space area?time area and spectrum area. So using classification technique of images we can recognize land-using-type. The principle of computer-RS classification recognizing is using the difference of feature of structure and spectrum energy to recognize information of surface features. Depending on example of RS-image in Beijing, land-using-type is divided supervised classification?unsupervised classification?maximum likelihood and neural network classification. Different method have its feature and different result. Neural net is approach to real.