遥感图像计算机分类的精度问题是阻碍计算机遥感信息处理系统实用化的一个关键问题。将分类后处理中的分类结果平滑过程模型化为约束优化问题,采用神经网络方法把分类结果平滑过程与遥感图像分类过程结合起来,提出了基于约束满足神经网络的遥感信息分类与后处理综合技术。实验表明该方法可明显提高森林类型划分、土地利用调查等遥感应用专题的分类精度。

The precision of computer classification is the main obstacle to the wide application of computer remote sensing information processing system. We model the smoothing classification results process as a constraint optimization problem, and integrate it with the classification process through neural network method, then present a remote sensing image classification and post_classification integrated technique based on constraint satisfaction neural network. This paper analyses The judging process of careerman, and discusses the theoretic model of neural network classification method in detail. The experimental results show that this method can improve the precision of classification saliently.