Real-Time Traffic Light Detection and Countdown Recognition
碩士 === 元智大學 === 電機工程學系 === 101 === This paper proposes a system for detecting traffic red light signal and recognizing the countdown numbers. Our research goal is to provide the driver a moving vehicle auxiliary warning system. We use the smart phone as a video capture device. This paper proposes fa...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Online Access: | http://ndltd.ncl.edu.tw/handle/40734880177258274859 |
| Summary: | 碩士 === 元智大學 === 電機工程學系 === 101 === This paper proposes a system for detecting traffic red light signal and recognizing the countdown numbers. Our research goal is to provide the driver a moving vehicle auxiliary warning system. We use the smart phone as a video capture device. This paper proposes fast image scanning method to enhance the scanning efficiency. We use the Region Growing method to localize the traffic light candidates. We use the Hough Transform method to confirm the location of traffic lights. We use the Super Resolution method to enhance the resolution of localized countdown numbers. Finally, we use the Adaboost method identifies number results. In the experiment, the detection rate can achieve above 70% and the recognition rate of countdown numbers is up to 95%. This shows that our proposed framework can be used in real world environment.
|
|---|