| 摘要 |
The present invention relates to an apparatus for detecting specific areas from a hyperspectral image and a method thereof. According to the present invention, an apparatus for detecting specific areas from a hyperspectral image includes: an input unit which receives a hyperspectral image captured by a hyperspectral camera; a feature band selecting unit which extracts spectral information from the hyperspectral image, and selects a feature band from the extracted spectral information; a feature information extracting unit which extracts feature information from the selected feature band by applying a band ratio technique to the selected feature band; a learning unit which obtains a detection rate by applying a learning machine to the extracted spectral information, and learns the extracted feature information; a feature information database which stores the learned feature information; and a specific area detecting unit which detects specific areas from a target image inputted on the basis of the feature information database. According to the present invention, a feature band can be selected by a detection rate-based simulated annealing algorithm to reduce the capacity of data to be processed and data processing time, as compared to using the full wavelength range, thereby improving the performance of the apparatus and method for detecting specific areas. In addition, the relative values of bands can be drawn by the band ratio technique, thereby effectively detecting constant specific areas despite changes in the lighting. |
