LAUSR

Abstract The rapid detection and identification of specific subjects such as different types of pollutants, minerals or rare wood are of vital significance to industrial manufacture and scientific research. However, such works are quite challenging because of the extreme similarities in appearance or chemical composition of them. In this paper, a self-designed methodology is applied to detect and identify different remarkably similar objects rapidly and precisely. The methodology is based on laser-induced breakdown spectroscopy (LIBS), principal component analysis (PCA) and the newly designed double variance selecting method (DVSM). To specifically expound the theory and reliability of the methodology, four similar species of rosewood were taken as examples. In our study, LIBS was applied to detect the main elements in Dalbergia odorifera sample and heavy metal in polluted Pterocarpus erinaceus samples. Moreover, PCA was used to distinguish four species of rosewood (training set) and store the training data sets for further identification. In addition, the new self-designed method called DVSM was employed to select the characteristic lines efficiently to support PCA to complete the work. In the end, a new rosewood sample of the four species was regarded as the test set to test the methodology and a systematic method was designed to compare the training data sets with the testing data to complete the test of identification, from which we got the accurate results. All the results indicate that the self-designed methodology can detect and identify extremely similar objects like rosewood effectively.