Object identification using mobile devices

To detect object from complex background, illumination variations and texture by machine is very difficult but important for adaptive information service. In this research, we present a preliminary design and experimental results of object recognition from a mobile device that utilizes the texture and the color features by image pre-processing with a simple vector distance matching classifier to train and extract the characteristics. The result shows that the proposed method can adopt the few characteristic values and the accuracy can reach up to 100% of object identification rate when making a querying in a mobile phone. The Euclidean distance is also used to represent the object similarity. The similarity can reach 87.5%, 62.5%, 75% and 87.5% respectively.