http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
운전자 부주의 상태 감지를 위한 이종 센서 퓨전 방법에 관한 연구
김범규(Beomkyu Kim),김기만(Kiman Kim),김신욱(Shinwook Kim),라볼레 필립(Philippe Lavole),박하빛(Habit Park) 한국자동차공학회 2017 한국자동차공학회 부문종합 학술대회 Vol.2017 No.5
Driver status monitoring systems are being applied to vehicles to reduce traffic accidents caused by driver drowsiness and carelessness. A camera that recognizes a face (ex. Blinking of the eye), and a radar that uses biometrics (ex. Heart rate and Breath measurement) are examples. However, in the case of face recognition, the interior illuminance in the vehicle and the reflection of the glasses can influence the recognition result greatly. In the case of the biometric signal recognition, the recognition result is different depending on the characteristics of the individual. In addition, there is a disadvantage that the state of the driver judged by the camera may be different from the state of the driver judged by the heartbeat sensing unit. In this paper, we propose the method and the integrated judgement algorithm to recognize the carelessness of the driver more precisely through the sensor output and weight calculation. The output is obtained by fusing the sensed result from each sensor using the integrated sensor which can be used inside and outside of the vehicle. The types of sensors that can be applied can be added or subtracted as occasion demands (ex. the type of vehicle, the accuracy of the system, the cost, etc.) and the cost can be reduced by using the existing sensors (illumination sensor, temperature sensor, etc.).
뇌 신경모방 프로세서를 이용한 교통표지판 인식 방법 연구
박하빛(Habit Park),김기만(Kiman Kim),김신욱(Shinwook Kim),라볼레 필립(Philippe Lavole),김범규(Beomkyu Kim) 한국자동차공학회 2017 한국자동차공학회 부문종합 학술대회 Vol.2017 No.5
The traffic sign recognition (TSR) systems are generally composed of the following steps: extraction of the traffic sign region, segmentation of the traffic sign contents, and recognition of each character or contents. This task is challenging due to the long processing time of complex algorithms and the low recognition rate of non-uniform outdoor conditions such as diversity of illuminance conditions and the damaged traffic signs. To solve these problems, this paper proposes a TSR system based on brain-inspired processor. The main function of the system proposed by this paper is to track and recognize the signs by comparing the similarities with learned patterns of traffic signs in the neurons instead of using time-consuming complex algorithms. Furthermore, by learning images with various conditions, the proposed system is robust under diverse situations. The training dataset is acquired based on images taken while driving under various conditions such as a clear day and a rainy day. The experimental results show that the proposed approach performs well in the TSR systems.