为克服单一分类器在遥感影像分类精度和效率方面的限制,有必要构建多分类器系统,集不同分类器的优点,获得比单一分类器更高的精度。针对遥感影像的特点和分类的需求,在遥感影像多分类器集成系统需求分析和系统设计的基础上,运用IDL语言在ENVI遥感影像处理平台下实现系统开发。遥感影像多分类器集成系统的主要功能包括遥感影像文件处理、特征选择与提取、分类预处理、分类、多种模式的多分类器集成(固定组合模式、用户自定义模式、向导模式和推荐模式)等。通过分类实例对系统应用进行介绍,表明本系统能够有效地提高遥感影像分类精度。

In order to overcome the limitation of the accuracy and efficiency of a single classifier, Remote Sensing Multiple Classifier System (RSMCS) is proposed. The system is able to combine the advantages of diffident classifiers and in the meantime acquires higher accuracy than that for a single classifier. According to the characteristics of remote sensing images and the acquirement of classifications, based on the requirement analysis and design for the system, the system is developed by using IDL language on the ENVI remote sensing information processing platform. The major functions of the system include remote sensing image file processing module, image feature selection and extraction module, remote sensing image classification pre-processing module, and remote sensing multiple classifier ensemble module (fixation combination style, user-defined style, wizard style, and recommend style). Furthermore, two experiments are presented to illustrate that the system could effectively improve the accuracy of remote sensing image classification.