Publication

Abstract : Hyperspectral imaging, an emerging technology is currently used for the detection and identification of vegetation, minerals, different objects and background. The imaging technique perceives information from across the electromagnetic spectrum, which aims at obtaining the spectra of each pixel for identification. The advantage of using hyperspectral images over multispectral images is that the former contains over 200 contiguous bands which makes it more sensitive to the subtle variations in the captured reflected energy. The higher spectral resolution of the hyperspectral image provides more spectral detail for each pixel. With the spectral information provided by multispectral images, it is possible to identify the background in the image that is, to differentiate between land cover, water, vegetation and so on. But for a hyperspectral image, it is possible to identify vegetation cover and to classify the species present in the area with the help of the continuous spectrum obtained for each pixel in an image. The obtained spectra are classified based on the labelled spectra available in the spectral library. Hence it is widely used in applications such as agriculture research, mineral exploration, drought assessment, water body analysis and so on. The application dealt in this paper is mapping a vegetation species. Vegetation mapping gives us an important set of information regarding its abundance and distribution in a particular area. This paper deals with the mapping of Mormon tea in and around the Mexican region using the EO- 1 Hyperion data. The L1R data is pre-processed to make the sensor and atmospheric corrections followed by feature extraction using MNF (Minimum noise fraction) transform and map the exact location of Mormon tea species using SAM (Spectral Angle Mapper) classifier in the ENVI software. Performing MNF overcomes the limitation of data redundancy in the hyperspectral image.