Method of detection the prevent coating on hyperspectral representation by Neumann-Pearson criterion

摘要

The objective of this paper is developing a method of detection of the prevent coating on hyperspectral representation according to Neumann-Pearson criterion. The mathematical models of pattern recognition by Bayes criterion were used as the primary method. For establishing the pattern recognition system an analysis of the spectral characteristics of the summer prevent coating (manufactures of Germany) was conducted. Also a calculation of density of the vegetation indices distribution was made. The statistical characteristics necessary for developing the mathematical model of pattern recognition system were obtained. The applicability of vegetation indices was shown upon detection of summer prevent coating against the background of foliage. A mathematical model of pattern recognition based on Neumann-Pearson criterion was presented. By the results of this paper was identified that the most informative index for identifying the prevent coating is the TCHVI (Three-Channel Vegetation Index), since it covers the three most significant spectral ranges: 550 nm, 650 nm and 900 nm. Based on the conducted experiments a possibility of using the pattern recognition method by Neumann-Pearson criterion to identify the prevent coating was proved.