Research And Application Of Near Infrared Spectroscopy To The Discrimination Of Rare Woods

The prices of different species of rare woods vary greatly, and the enormous profits space leads to the result that fraudulent furniture sellers in the Chinese market can deceive consumers by selling substandard furniture as high-value items. Most consumers find it difficult or impossible to distinguish wood species. Studies at home and abroad indicated that wood identification could be realized based on near infrared spectroscopy(NIRS) technology with its rapid, non-destructive and accurate properties in recent years. However, there are not further researches about rapid identification of some ease-confusing rare-wood furniture species in the Chinese market using near infrared spectroscopy technology, which is difficult to adapt to the market demand.Spectral characteristics and spectra pretreatment method of six species of rare-wood logs and ten species of rare-wood furniture were analysed and researched, thus the optimal pretreatment method was found. Moreover, the spectral differences between rare-wood logs and furniture were analysed according to the optimal pretreatment method. Then, the preliminary discrimination models of rare-wood furniture by SIMCA, PLS-DA, ELM, LS-SVM and PNN can be built over a range of 1000~1650 nm. In comparison, the classification results by SIMCA, ELM and PNN are better. In further optimization of models, the effective wavelengths were selected by principal components analysis, continuum removal and MC-UVE-SPA with nearly 77%, 96% and 97% reduction of numbers of characteristic wavelengths, respectively. The overall results demonstrated that MC-UVE-SPA could exact the most effective wavelengths to build the models with the optimal recognition rate and rejection rate. Among those models, the optimal classification result was built by ELM model, and the correct rate is 99.77%.On the basis of this study, a method of combining NIRS technology and Internet of Things(IoT) was presented to establish near infrared rare woods discrimination system based on cloud computing. The design of the architecture, functions, and operation of this system were detailed herein and the developed system based on the structure of C/S(Client/Server) is implemented on a Windows platform. The proposed online system, which enables the rapid and accurate discrimination of rare woods, is verified through a test implementation.