The Application of the Spectral Similarity Scale Algorithm and Expectation-Maximization for Unsupervised Change Detection using Hyperspectral Image

하이퍼스펙트럴 영상의 무감독 변화탐지를 위한 SSS 알고리즘과 기대최대화 기법의 적용

Recording data in hundreds of narrow contiguous spectral intervals, hyperspectral images have provided the opportunity to detect small differences in material composition. But a limitation of a hyperspectral image is the signal to noise ratio (SNR) lower than that of a multispectral image. This paper presents the efficiency of Spectral Similarity Scale (SSS) in change detection of hyperspectral image and the experiment was performed with Hyperion data. SSS is an algorithm that objectively quantifies differences between reflectance spectra in both magnitude and direction dimensions. The thresholds for detecting the change area were determined through Expectation-Maximization (EM) algorithm. The experimental result shows that the SSS algorithm and EM algorithm are efficient enough to be applied to the unsupervised change detection of hyperspectral images.