Design and Implementation of Object Recognition System Using SoC and FPGA

Abstract

Object tracking and recognition have a wide range of applications. Recent years, many projects have been done in this field. It is easy to see that its application plays an important role in medical and biomedical image processing, geographical information, industry image analysis, earth science, or satellite based military. Feature-based algorithms like Speeded Up Robust Features (SURF), and SIFT (Scale-invariant feature transform) are well suitable for such operations [3]. Among these, Oriented fast and Rotated BRIEF (ORB) has been proved to achieve optimal results. ORB is based on the well-known fast key point detector concept Binary Robust Independent Elementary Features (BRIEF) descriptor [2]. Nearest neighbour matching algorithms that are Fast Library for Approximate nearest Neighbours (FLANN) [1] which is known as one of the best algorithms for features matching. Simulation has been carried out on Window platform using Opencv-Python. This project also provides an idea to implement ORB algorithm on SoC (ARM–BCM2835) to increase the execution speed and reduce the complexity in system design. By applying Opencv-Python, which provides a rich library for image processing applications, time consuming and the complication in developing software can be reduce remarkably. Also, the proposed object recognition system is empowered with Altera FPGA CYCLONE III which is believed to handle all control signals to and from external devices.