论文部分内容阅读
基于扩展的高斯混合密度降解模型,本文提出一种利用遗传算法实现逐步寻优的特征挖掘模型。利用该模型可以从混合密度特征空间中挖掘出多个未知特征并以树状的层次方式逐步分离出来。为了更好地与实际情况相吻合,本文引入符号化的知识处理方法,提出了基于知识的空间特征逐步寻优挖掘模型,并将其应用到遥感影像的实例分析,取得了较好的效果。
Based on the extended Gaussian mixture densification model, this paper presents a feature mining model that uses genetic algorithm to achieve gradual optimization. Using this model, we can mine many unknown features from the mixed density feature space and gradually separate them in a tree-like hierarchical manner. In order to better match with the actual situation, this paper introduces a symbolic knowledge processing method, and proposes a mining model based on the knowledge of the spatial characteristics of the gradual optimization and its application to the case analysis of remote sensing images, and achieved good results.