Object recognition in a context-aware application

In a dynamic operational environment such as robotic or an autonomous navigation system, the interactions between humans and objects around them play an important role (context-awareness). The task of recognizing and tracking such objects introduces many challenges in the machine vision research field. In this paper, we propose a novel method that combines the information from modern depth sensors with conventional machine vision techniques such as Scale-invariant Feature Transform (SIFT) to produce a system that is capable of performing object recognition and tracking with a satisfactory level of accuracy in real-time. A prototype is implemented and tested to confirm that the proposed method does provide better performance comparing with currently used methods in image processing.