Use The Algorithm of Adaptive Coherence Estimator for Hyperspectral Imaging Materials Identification and Classification

• Main Authors: Chia-Kai Lin, 林家楷
• Other Authors: Min-FanLee
• Format: Others
• Language: en_US
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/225y5w

碩士 === 國立臺灣科技大學 === 自動化及控制研究所 === 106 === Applications of the geometric-corrected remote-sensing images are used to supervise environment and identify crop. We take advantage of the spectrometer mounted on the aircraft to acquire hyperspectral imaging (HSI). Then use the algorithm of adaptive coherence estimator (ACE) to classify and identify the unknown materials. By performing the software, ENVI, to classify and identify materials in the geometric-corrected images. Based on the record of aircraft’s attitude, we corrected the distortion of the images. According to the flight height, latitude, longitude and orientation, we can record them as input geometry to correct the distortions. After correcting to north-up images, we compare them to Google map and measure the distance. Every pixel of a hyperspectral image has spectral reflectance, and there is a different value in different wavelength. The spectral curve is a continue curve that consists the reflectance of all wavelength. In the thesis, we use ACE to compare the spectral curve of VNIR and SWIR to referenced spectral library. Therefore, we can identify what is agriculture or mineral. We also can classify the relative materials and know the range of them. This study establishes applications and a procedure of hyperspectral imaging analyzing integration system which consists of acquiring the flight data, geometric correction, materials identification and classification.