A Mobile Phone Based Rock Classification System

Abstract

Ubiquitous computing has become an area of study in computer science and electrical engineering. Mobile devices, such as mobile phones, are widely used in daily life for various purposes. Mobile devices can facilitate human interaction and access to information resources anytime and anywhere. In this paper, we propose a mobile phone based rock classification system. For this system, the mobile phone played the roles of image capture and rock information display. Rock images were classified via the remote server. Three texture features, i.e. directionality, color and granularity, were exploited to classify rock textures. A TGA based approach was developed to generate the optimal color histogram. The granularity was quantified by the fractal dimension. To alleviate the degradation resulting from texture inhomogeneity, we propose the RW-k-NN classifier to integrate the color and granularity information. The experimental results prove our system can accurately classify rocks. This implies that our system lends itself to a ubiquitous rock classification system.