遥感信息
遙感信息
요감신식
2015年
1期
26-32,50
,共8页
无限潜 Dirichlet 分配%非监督分类%Block-Gibbs%Dirichlet 过程
無限潛 Dirichlet 分配%非鑑督分類%Block-Gibbs%Dirichlet 過程
무한잠 Dirichlet 분배%비감독분류%Block-Gibbs%Dirichlet 과정
infinite Latent Dirichlet Allocation%unsupervised classification%Block-Gibbs%Dirichlet process
通过引入文本检索算法中的无限潜 Dirichlet 分配(infinite Latent Dirichlet Allocation,即 iLDA)模型,对遥感影像进行建模以获取地物的统计分布及其共生关系,从而实现遥感影像非监督分类。首先,将遥感影像有重叠地划分成一组大小相等的影像块(文集)。其次,以 iLDA 为基础,构建“像元”(视觉词)、“影像块”(文档)和“地物类”(主题)之间的条件概率关系,并采用 Block-Gibbs 抽样的方法来估计模型参数,从而构建基于 Block-Gibbs 抽样的 iLDA 遥感影像非监督分类模型(Block-Gibbs based iLDA,即 BG-iLDA)。最后,通过对 BG-iLDA 模型的逼近求解实现高分辨率遥感影像的非监督分类。实验结果表明,本文提出的基于 BG-iLDA 的面向对象非监督分类方法相对传统的 K-means 等算法精度更高,更能有效区分“同谱异物”的地物。
通過引入文本檢索算法中的無限潛 Dirichlet 分配(infinite Latent Dirichlet Allocation,即 iLDA)模型,對遙感影像進行建模以穫取地物的統計分佈及其共生關繫,從而實現遙感影像非鑑督分類。首先,將遙感影像有重疊地劃分成一組大小相等的影像塊(文集)。其次,以 iLDA 為基礎,構建“像元”(視覺詞)、“影像塊”(文檔)和“地物類”(主題)之間的條件概率關繫,併採用 Block-Gibbs 抽樣的方法來估計模型參數,從而構建基于 Block-Gibbs 抽樣的 iLDA 遙感影像非鑑督分類模型(Block-Gibbs based iLDA,即 BG-iLDA)。最後,通過對 BG-iLDA 模型的逼近求解實現高分辨率遙感影像的非鑑督分類。實驗結果錶明,本文提齣的基于 BG-iLDA 的麵嚮對象非鑑督分類方法相對傳統的 K-means 等算法精度更高,更能有效區分“同譜異物”的地物。
통과인입문본검색산법중적무한잠 Dirichlet 분배(infinite Latent Dirichlet Allocation,즉 iLDA)모형,대요감영상진행건모이획취지물적통계분포급기공생관계,종이실현요감영상비감독분류。수선,장요감영상유중첩지화분성일조대소상등적영상괴(문집)。기차,이 iLDA 위기출,구건“상원”(시각사)、“영상괴”(문당)화“지물류”(주제)지간적조건개솔관계,병채용 Block-Gibbs 추양적방법래고계모형삼수,종이구건기우 Block-Gibbs 추양적 iLDA 요감영상비감독분류모형(Block-Gibbs based iLDA,즉 BG-iLDA)。최후,통과대 BG-iLDA 모형적핍근구해실현고분변솔요감영상적비감독분류。실험결과표명,본문제출적기우 BG-iLDA 적면향대상비감독분류방법상대전통적 K-means 등산법정도경고,경능유효구분“동보이물”적지물。
In this paper,the infinite Latent Dirichlet Allocation (iLDA)model for unsupervised classification of images is introduced.An effective unsupervised classification method using the semantic information and the symbiotic relationship from iLDA is proposed,which is used for high-resolution panchromatic images.Firstly,the image corpus is structured by overlapped segmentation of the image into sub-images.Secondly,the relationship of conditional probability among pixels (visual-words), sub-images (documents)and land objects (topics)is built.By which,the proposed method using Block-Gibbs based iLDA (BG-iLDA)is modeled.And the model parameters are estimated using the Block-Gibbs sampling.Finally,the unsupervised classification of high-resolution panchromatic images is realized by approximate solution of the BG-iLDA.Experimental results show the classification precision of the proposed method is better than the K-means method,and the effect of the different object with the same spectral characteristics is appropriately displayed by the classification result.