- 期刊
Classification of Multiangular, Multispectral Imagery Using Hidden Markov Model
植基於隱匿式馬可夫模式之多頻譜、多視角影像判釋
摘要
本研究以捷鳥2號衛星於公元2003年7月13日於美國加州Fresno機場近郊所攝取之四角度多頻譜影像進行判釋實驗，判釋方法係基於隱匿式馬可夫模式(HMM)以設計監督式及非監督式分類方式並探討其基於多視角及多頻譜資訊之判釋效果，本研究於監督式判釋實驗上，係採用HMM以模擬多頻譜、多視角影像間之關係，而於非監督式判釋實驗時，HMM則主要用於模擬空間之脈絡關係(Context)，且係採用新設計之HMM觀測密度調整法(Observation density adjustment)進行非監督判釋實驗，本研究發現多頻譜、多視角資訊確可增進影像之判釋精度，同時本研究發現基於HMM分類之精度要高於傳統方法如最大概似法及K-mean分類之判釋精度。
並列摘要
Four multispectral Images (MSI) were acquired upon various viewing angles by the Quickbird-2 satellite in passing over airport neighboring fields nearby Fresno, California on July 13, 2003. These appear to be the very original, multiangular, multispectral, high spatial resolution images acquired from space. The classification experiments are based on the implementation of hidden Markov models (HMM) to design both supervised and unsupervised classification schemes to test the classification performance relating to the angular and spectral information. For supervised classification, the HMM were applied to model the temporal dependencies between multiangular images, while for unsupervised classification, the HMM were adopted mainly to model the spatial dependencies (i.e. contextual information) using proposed observation-density adjustment methodology. The experimental results reveal that the addition of multiangular information to MSI imagery significantly improves the accuracy of scene discrimination. It is also concluded that HMM is a superior tool for dealing with multiangular, MSI data in comparing to the traditional supervised maximum likelihood classifier and k-means clustering method.