中文摘要：

针对传统遥感影像分类方法效果不够理想,单一分类器各自存在不足等问题,该文提出了一种基于多分类器组合的遥感影像分类方法。采用级联和并联相结合的方式对多种子分类器进行组合;利用改进的基于先验知识的投票表决规则,实现遥感影像准确分类。以岳阳市TM遥感影像为例,采用多分类器组合方法进行分类处理,并将处理结果与单一分类器处理结果进行比较。通过误差矩阵对比可知,多分类器的Kappa系数精度高于单一分类器;对分类效果图进行对比分析,在细部效果方面多分类器分类效果优于单一分类器。研究结果表明：组合分类器的遥感影像分类效果明显优于单一分类器,且具有更好的扩展性。

英文摘要：

Aiming at the problem that traditional remote sensing image classification method is not satisfactory and the deficiency of single classifier, a remote sensing image classification method based on the combination of multi classifier was proposed. Through the combination of cascading and parallel to make a combined classifier which used the improved voting rules based on the prior knowledge meanwhile, accurate classification of remote sensing image was realized. Taking TM image of Yueyang city as an example, the proposed method and single classification method （maximum likelihood） were compared. The comparison result of error matrix showed that the Kappa coefficient of multiple classifier was higher than that of single classifier; the detail parts in classification results figure of multiple classifier was also better than that of the single classifier classification. Experimental results indicated that the effect of combined classifier for remote sensing image classification was superior to the single classifier and had deeper scalability.