| Summary: | 碩士 === 國立中央大學 === 資訊工程研究所 === 93 === At present, there are many commercially available safety equipments such as the safety belts, airbags, etc, on vehicles. They are used to diminish the degree of injury while car accidents happen. These passive safety equipments can't really prevent the occurring of car accidents in practice. Recently, many various safety systems have been proposed. These safety systems concentrate on the driver fatigue detection, lane detection, etc.
In this thesis, an attention detection system for vehicles is proposed to detect the driver’s driving behaviors and driving states while driving. Via the attention detection system warning signals will be issued in time when an accident is going to happen. In this way, car accidents can then be greatly reduced.
There are two functional blocks in the attention detection system. They include the driver attention detection block and the driving state analysis block. In the driver attention detection block, a web camera is used to acquire driver’s images while driving. Some image processing techniques are utilized to locate the positions of eyes and the center of the face. These three feature points consist of a triangle. The three edge vectors and the three included angles in the triangle can then be used to estimate the driver’s face orientations. In addition to the information about the face orientations, information about the eye-closure frequency and the information about the appearance of the face are integrated to determine whether the driver’s attention degree. The driving state analysis block involves in the use of an accelerometer circuit. The accelerometer can measure the degree of the acceleration of two different directions at the same time. The accelerometer circuit will record the driving states of the car. Then a finite state machine is used to analyze the driving states. Signals can be issued in time to warn the driver while improper driving behaviors are detected. In fact, we can also use the driving state analysis block to determine how often a bus or truck driver exhibits improper driving behaviors after he or she completes a journey. Experiments were conducted to test the performance of the proposed attention detection system.
|
|---|
