원문정보
초록
영어
According to the traffic light recognition problem of intelligent vehicle, this paper proposes a traffic light recognition algorithm based on compressive tracking. First, the candidate regions of traffic light are extracted. Second, after extracting the HOG feature of candidate regions, using the machine learning for classification and recognition, compressive tracking algorithm is used to track lights that have been identified a automatically. This algorithm combines feature recognition and tracking identification of traffic light and traffic light changes in scale and color can be identified normally. The algorithm proposed in this paper has been tested on actual road in intelligent vehicle; traffic light recognition effect is good.
목차
1. Introduction
2. Identification Based on Features
2.1. Color Segmentation
2.2. Morphological Operations
2.3. Shape Segmentation
2.4. Feature Extraction
2.5. Classifier Training and Recognition
3. Recognition Based On Compression Tracking
3.1. Compressive Sensing Principle
3.2. Compressive Tracking
3.3. Identification
4. Results and Discussion
5. Conclusion
Acknowledgements
References