| 摘要 |
PURPOSE: To quickly and accurately identify an unknown spectrum through the minimum comparison by converting the spectrum into the spectrum corresponding to known samples displayed at every point. CONSTITUTION: A spectrum identifying method uses a microcomputer 10 associated with a spectrum image sensor 14, a sensor interface 12, a classification library 18, and a RAM for interacting with identification software 20 as hardware. Firstly, a spectrum is converted into the spectrum corresponding to known samples displayed at every point and a mathematical angle is decided by deciding the similarity among all vectors. Then a first vector which is the closet to an average vector and represents the center of the vector group and a second vector which is the most dissimilar to the first vector and represents the local center of the group are selected and, at the same time, all vectors are divided into the selected two vector groups. Then another vector representing the local center of each divided vector group is selected. |
