Comparison of Feature Extractors for Real-Time Object Detection on Android Smartphone

• Main Author: Saipullah, Khairul Muzzammil
• Format: Article
• Language: English
• Published: 2013
• Subjects: TA Engineering (General). Civil engineering (General)
• Online Access: http://eprints.utem.edu.my/id/eprint/8556/1/17Vol47No1_3.pdf; http://eprints.utem.edu.my/id/eprint/8556/
• Institution: Universiti Teknikal Malaysia Melaka

This paper presents the analysis of real-time object detection method for embedded system particularly the Android smartphone. As we all know, object detection algorithm is a complicated algorithm that consumes high performance hardware to execute the algorithm in real time. However due to the development of embedded hardware and object detection algorithm, current embedded device may be able to execute the object detection algorithm in real-time. In this study, we analyze the best object detection algorithm with respect to efficiency, quality and robustness of the algorithm. Several object detection algorithms have been compared such as Scale Invariant Feature Transform (SIFT), Speeded-Up Feature Transform (SuRF), Center Surrounded External (CenSurE), Good Features To Track (GFTT), Maximally-Stable External Region Extractor (MSER), Oriented Binary Robust Independent Elementary Features (ORB), and Features from Accelerated Segment Test (FAST) on the GalaxyS Android smartphone. The results show that FAST algorithm has the best combination of speed and object detection performance.