Material Recognition Based on a Pulsed Time-of-Flight Camera

Abstract

This study presents a method for material recognition using a pulsed time-of-flight (ToF) camera. The method measures the material bidirectional reflectance distribution function (BRDF) as a feature for material recognition by a pulsed ToF camera. We use the measurements of incident light at different angles to form the BRDF feature vectors. The feature vectors are used to build a training and test set to train and validate a classifier to perform the recognition. We choose the radial basis function (RBF) neural network as a classifier based on the nonlinear characteristics of material BRDF. Finally, we construct a turntable-based measurement system and use the material BRDF as the feature for classifying a variety of materials including metals and plastics. The optimized RBF neural network can achieve a recognition accuracy of 94.6%.