Vision-Based Daytime Brake Light Detection Using Front-Mounted Car Camcorder
|關鍵字:||自動車;駕駛安全輔助系統;車輛偵測;車尾燈偵測;煞車燈偵測;Automobile;driver assistance system;vehicle detection;taillight detection;brake light detection|
Due to the rapid expansion of car ownership worldwide, driving traffic safety becomes a recently rising issue in the automobile industry. Increasing driver assistance systems have been developed to detect potential problems in advance by continuously monitoring the vehicle surroundings and the driving behaviors. Detecting brake lights during the daytime is a challenging work due to varied illumination conditions and is of vital importance for preventing drivers from mortal and costly rear-end collisions. In this thesis, we propose a vision-based daytime brake light detection system using a front-mounted car camcorder, which tends to be widespread deployed. First, front vehicle candidates are detected by a cascade of Gentle Adaboost classifiers utilizing Histogram of Oriented Gradients descriptors. To improve the detection accuracy, symmetry of taillights is further employed to filter out the false detected candidates. Once a vehicle is detected, the regions of its taillights are utilized to differentiate the brake lights from nonbrake lights by investigating both radial symmetry and luminance features. Experiments conducted on real videos captured by front-mounted car camcorders demonstrate that satisfactory results can be obtained by the proposed daytime brake light detection system.
|Appears in Collections:||Thesis|