| Summary: | 碩士 === 南臺科技大學 === 電機工程系 === 104 === The thesis proposes an IoT plate recognition system for anti-theft purpose. It joins efforts between the public and the police to overcome vehicle theft problems. By using the building abilities of camcorder and communication in the devices, it provides obsolete mobile phones a new role to play. The mobiles device continuously captures and analyzes every vehicle plate it sees and communicates with server cloud to get the information of whether the plate number connecting any crime. Therefore, the system shall benefit many aspects including increasing theft detection rate and deterring criminal events.
The thesis focuses on the essential part of the system, which proposes an approach of plate recognition techniques. In the system, it modifies the method of vehicle license plate location under complex scenes, adopting the algorithm for vehicle license plate tilt correction based online fitting method, and then employing template matching method for character recognition and achieving license plate recognition system.
In order to verify the license plate recognition system, the algorithm is tested with experiments of two megapixel samples. The results of benchmark show that the success rate of license plate location and character segmentation reaches 89.2% and 90.1%respectively. Different platforms are also tested for efficient comparison. For instances, the TI Beaglebone gets processing times of 0.7s under 640x480 pixel samples, while an android device gets 3s~8s under 2 megapixel samples.
|
|---|
