| Summary: | 碩士 === 國立臺灣科技大學 === 電子工程系 === 97 === In recent years, research and development of automated license plate recognition system in terms of car parking charges or traffic ban. this paper presents the implementation of a real-time FPGA license plate recognition system, and we achieve the detect license detection plate hardware base on SOPC, we got good performance of real-time detection .In this paper, the license plate detection we achieve by using discrete wavelet transform(DWT) and morphology. Furthermore the object Connect Component Labeling method to segment each independent block and identify the images license plate location coordinates. Before we get the characters of each registration coordinates, we should segment each region of the characters from the plate. The character image will be send to character recognition software (OCR). After that, the license plate characters identification will be finish.
Using above method, we can achieve this purpose on Terasic company’s DE2-70 FPGA development kit, which compose of Altera NIOS II soft core embedded processor and SOPC platform development environment. And it combined with camera whose resolution is 720X487.After the system finished the plate detection and cutting out characters, it shows the result from VGA interface to the Monitor.
Finally, we can successfully capture the license plate on the rate 96% and 85% on segment the characters, the constrained of our environment are the following: the vehicle 1 to 3 meters, the license plate can’t slope over 5 degrees, the region had no any obviously shelters or dust. If on the other environment, even if the original image slightly tilted or have different light intensity. According to the experiment, we can still reach 80.5% success rate of license plate detection.
|
|---|
