Target Detection Method of Hyperspectral Image by Indexing and Querying based on Pyramid-technique

摘要

The present invention relates to a method for detecting material of a hyperspectral image by indexing and querying based on a pyramid-technique. The method for detecting material of a hyperspectral image according to the present invention comprises the steps of: preprocessing spectral data in a spectral library by using differential spectroscopy; constructing an index for the preprocessed spectral data by using a pyramid-technique; generating a key value from material data extracted from a hyperspectral image; moving to a position in which a key value of spectral data similar to the key value of the material data exists within the index; extracting candidate materials from the spectral library by querying according to a specific query algorithm within a predetermined extraction range based on the position after the movement; filtering other materials among the candidate materials extracted by the querying; and outputting a final candidate remaining through the filtering. [Reference numerals] (AA) Start; (BB) End; (S11) Pre-process spectral library data using differential spectroscopy; (S12) Construct an index for the preprocessed data by using a pyramid-technique; (S21) Generate a key value from material data extracted from a hyperspectral image; (S22) Move to a position in which a value similar to the key value exists within the constructed index; (S23) Process a query using an K-nearest query algorithm based on an area query; (S24) Filter other materials among candidate materials extracted by the query; (S25) Output the filtered results