The accurate classification of rice is important to effectively identify food fraud, especially in China, where rice is widely used as a staple. This study explored the application of near-infrared (NIR) and Raman spectroscopies and low-level and mid-level data fusion for the classification of four types of rice from similar origins. Classification models were constructed on separate data blocks after different preprocessing methods using partial least squares-discriminant analysis (PLS-DA) and support vector machine (SVM). Low-level data fusion spliced the two types of spectra and a classification technique was applied. Mid-level data fusion consisted of band selection and feature extraction from the spectra from each technique and the establishment of classification models. Based on fuzzy set theory, high-level data fusion was carried out for the results of PLS-DA models based on a single spectrum. The results show that mid-level data fusion is effective because its performance was superior to the use of one instrument. The highest accuracy was 87.5% for both NIR and Raman and 100% for the SVM model based on the mid-level data fusion. Mid-level data fusion of Raman and NIR spectroscopies is effective for the nondestructive identification of rice.

Almost all robust stereo visual odometry work uses the random sample consensus (RANSAC) algorithm for model estimation with the existence of noise and outliers. To date, there have been few comparative studies to evaluate the performance of RANSAC algorithms based on different hypothesis generators. In this work, we analyse and compare three popular and efficient RANSAC schemes. They mainly differ in using the two-dimensional (2-D) data points measured directly and the three-dimensional (3-D) data points inferred through triangulation. This comparison presents several quantitative experiments intended for comparing the accuracy, robustness and efficiency of each scheme under varying levels of noise and different percentages of outlier conditions. The results suggest that in the presence of noise and outliers, the perspective-three-point RANSAC provides more accurate and robust pose estimates. However, in the absence of noise, the iterative closest point RANSAC obtains better results regardless of the percentage of outliers. Efficiency, in terms of the number of RANSAC iterations, is found in that the relative speed of the perspective-three-point RANSAC becomes superior under low noise levels and low percentages of outlier conditions. Otherwise, the iterative closest-point RANSAC may be computationally more efficient.