CAJ | 학술논문

土地覆盖是自然环境与人类活动相互作用的中心,而土地覆盖信息主要是通过遥感影像分类来获取,因此影像分类是遥感影像分析的最基本问题之一.在参考基于概率主题模型的高分辨率遥感影像聚类分析的基础上,通过半监督学习最典型的生成模型方法引出了基于概率主题模型的半监督分类(SS-LDA)算法.借鉴SS-LDA模型在文本识别应用的流程,构建了基于SS-LDA算法的高分辨率遥感影像分类的基本流程.通过实验证明,相对于传统的非监督分类与监督分类算法,SS-LDA算法能够获取较高精度的影像分类结果.
토지복개시자연배경여인류활동상호작용적중심,이토지복개신식주요시통과요감영상분류래획취,인차영상분류시요감영상분석적최기본문제지일.재삼고기우개솔주제모형적고분변솔요감영상취류분석적기출상,통과반감독학습최전형적생성모형방법인출료기우개솔주제모형적반감독분류(SS-LDA)산법.차감SS-LDA모형재문본식별응용적류정,구건료기우SS-LDA산법적고분변솔요감영상분류적기본류정.통과실험증명,상대우전통적비감독분류여감독분류산법,SS-LDA산법능구획취교고정도적영상분류결과.
Land cover is the center of the interaction of the natural environment and human activities and the acquisition of land cover information are obtained through the classification of remote sensing images, so the image classification is one of the most basic issues of remote sensing image analysis. Based on the image clustering analysis of high-resolution remote sensing image through the probabilistic topic model, the generated model which is a typical method in the semi-supervised learning is analyzed and a classification method based on probabilistic topic model and semi-supervised learning(SS-LDA)is formed in the paper. The process of SS-LDA model used in the text recognition applications is relearned and a basic image classification process of high-resolution remote sensing image is constructed. Comparing to traditional unsupervised classification and supervised classi-fication algorithm, the SS-LDA algorithm will get more accuracy of image classification results through experiments.