Research On Spectral Feature Analysis Of Objects

The spectral feature analysis of objects is the precondition and base for classification and matching of hyperspectral images. The spectral features of objects are affected by temperature, weather, dispersion, and the instrument of surveying,the phenomena of different objects same image and different images same object appear frequently in the process of object measure, and the spectral features of objects provide a large amount of data and categories, which restrict the applications and developments of the hyperspectral images. We research specially on the spectral feature analysis of objects in this paper. The accomplishments include:1. Compared the methods that obtain the data of the spectral feature of objects, and introduced the practical process using the outdoor portable spectrometer or the indoor spectrometer. Owing to the characteristic of the a large amount of data and the abundant information in hyperspectral images, in this paper we introduce the data preprocessing and reflectivity transformation for the hyperspectral images in detail, which can be massive basis for the subsequent research.2. Analyzed the spectral features of vegetation, geo-technical objects, water bodies and cultural details, the spectral features with particularity and time dynamics can be recognized, yet with difficulty. The continuum-removed and differential coefficient are often used to enhance the spectral features of objects in analysis. A new arithmetic for extracting the continuum is improved; the result indicated that the effect of this arithmetic is better in practice.3. Programmed to realize the parameterization of the spectral features. In order to validate the parameterization, ameliorate spectral comparability measure.The result indicated that this program summarize the shape and numerical value indexes could acquire higher precision and better adaptability compared with traditional single measure, the precision and reliability of classified research using optimized distance measure have a big advantage.4. Established the analysis system of the spectral features. On the basis of spectral database, realized the plat of synchro analysis on spectral curve data obtained from hyperspectral images, spectral database and other correlative systems, realized the spectral feature visualization and analysis, and provided the interface for development.