A data fusion method of electronic nose and hyperspectral to identify the origin of rice

Abstract In the process of rice quality supervision, it is common for rice to be sold with the label of high-quality origin. Therefore, it is important to provide an effective technique to identify the origin of rice. In this work, a data fusion method for electronic nose (e-nose) and hyperspectral is proposed in combination with support vector machine (SVM) to achieve the origin traceability of rice. Firstly, Stability competitive adaptive reweighted sampling (SCARS) is used to selection the key bands of the spectral information. Secondly, Max-Relevance and min-redundancy (MRMR) is applied to sort the features importance of e-nose. Finally, feature sets are generated based on the order of the feature importance of e-nose, and fuse with the key bands of hyperspectral, respectively. The results show that the data fusion method coupled with SVM achieve a good classification performance of 98.57%. It provides an effective technology to achieve the origin traceability of rice.