论文部分内容阅读
随着遥感技术的发展，可获取的遥感信息越来越丰富，各种分类算法也层出不穷，然而在分类精度上却始终无法突破遥感数据本身的局限，而且不同的方法在判识的结果上往往有很大的不同。本文认为不同遥感信息、不同的判别分析方法，对不同的对象在不同的地表组合条件下其判别的能力是不同的，只有综合地运用这些不同的方法才可能在现有遥感资料的能力下提高分类精度。本文提出一种动态综合层次判别分析算法可将多种并行的分类算法综合在一起，取长补短，提高最后分类精度。通过在浙江省新昌县某地进行水土流失强度分析试验，取得了良好的效果。
With the development of remote sensing technology, more and more remote sensing information is available and various classification algorithms emerge one after another. However, the classification accuracy has not been able to break through the limitations of remote sensing data, and different methods often result in the identification there is a big difference. This paper argues that different remote sensing information and different discriminant analysis methods have different ability to discriminate different objects under different surface combination conditions. Only by using these different methods synthetically, it is possible to improve them under the existing capability of remote sensing data Classification accuracy. This paper presents a dynamic comprehensive hierarchical discriminant analysis algorithm can be a variety of parallel classification algorithms together, learn from each other to improve the final classification accuracy. Through the soil erosion intensity analysis test in Xinchang County, Zhejiang Province, good results have been achieved.