Rock classification via a mobile device

Mobile devices have been widely used to facilitate information exchanges and help our daily lives. In this paper, we present a rock classification system based on mobile environment. This system consists of a mobile phone and a remote server. First, the mobile phone is used to image rocks, due to the convenience in viewing and checking the captured images. However, it is difficult to automatically recognize the rocks in the phone because of insufficient computing capability. Therefore, the users can select and submit crucial images to a remote server in order to classify the type of rocks. The classification task is carried out by feature extraction, followed by a neural network-based classifier. The texture features consists of color, directionality, and granularity. With the extracted features, an ANFIS classifier is used to accomplish the recognition task. The experimental results show that our system can successfully classify rocks and achieve the ubiquitous rock classification task.