| Summary: | 碩士 === 中興大學 === 電機工程學系所 === 102 === In this thesis, programmable object detection by using feature extraction of objects for vehicle and road sign is proposed. The presented programmable parameter for the feature and object extraction not only decrease the computation complexity but also increase the accuracy of object detection. Experimental results show the average accuracy rate of object detection can exceed about 90%. Moreover, we further proposed a simple recognition system for speed limit, triangle alarm and stop signs.
To reach the target of real-time processing, related hardware architecture is also presented for FULL HD@30fps by using algorithm design flow. Block based color character decision is used to minimize the memory usage. To reduce the computation time, fast symmetry map implementation is also adopted. Finally, the proposed hardware is simulated at Zynq-7000 platform and work at 11.8 MHz to reach the real time requirement.
|
|---|
