Abstract
This paper presents a new method of feature extraction and segmentation using near-infrared hyperspectral imaging camera. Hyperspectral imaging is a combination of digital imaging and spectroscopy. Hyperspectral imaging has been applied in the fields of remote sensing, agriculture, biological engineering and so on. The process of extracting effective information from the hyperspectral imaging data is required to automatic analysis method. In this paper, the hyperspectral imaging data segmentation method extracted the feature of the data based on principle component analysis (PCA) and K-means clustering are proposed. The proposed method is applied to the four evaluation experiment; (A) discrimination between fatty meat and lean meat, (B) discrimination between meat and flour, (C) classification of low oxygen environment and (D) classification of mouse organs. The results show the proposed method can select characteristic bands in the near-infrared range automatically. The proposed method is expected to be applied to other field.
