Hyperspectral remote sensing is becoming more and more important amongst remote sensing techniques. In this paper, we present a hyperspectral database (HyperDB) designed to cooperate with an embedded hyperspectral image processing system developed by the authors. Hyperspectral data are recognized and categorized by their land coverage class and band information, and can be imported from various sources such as airborne and spaceborne sensors carried by airplanes or satellites, as well as handhold instruments based on in situ ground observations. Spectral library files can be easily stored, indexed, viewed, and exported. Since HyperDB follows standard design principles—independence, data safety, and compatibility—it satisfies the practical demand for managing categorized hyperspectral data, and can be readily expanded to other peripheral applications.