| Summary: | 碩士 === 國立高雄應用科技大學 === 資訊工程系 === 100 === In Taiwan, the traffic monitor on the road is more and more attention. And more and more monitors to monitor the traffic on the road. The footage that the monitor shoot may be a complex environment. The goal that the car in the image which the monitor photographed is want to monitor the car is legitimate driving. The license plate detection is the most practical way. So we search more and more related literature to find the fitness license plate detection method and implement it.
However, a lot of factors must to be considered in a complex environment. We chose the AdaBoost algorithm to implement the license plate detection work. As this method has been train the training data separately from the license plate and non-license plate. Learn what is license plate and what is non-license plate, and so that the influence by the environment factors are significantly reduced, and that may achieve nice detection result. Implement process is applied the integral image to accelerate the computation of feature values, and use the Haar-like features to compute the feature values. Finally, the experimental results are achieve nice results.
|
|---|
