| Since the twentieth century,the number of cars has increased significantly,the traffic accidents have taken place more and more frequently,people are increasingly concerned about the traffic safety issues,the world’s Ministry of Communications is actively taking some effective measures to reduce traffic The accident happened.As a result,the driver’s line of sight tracking system emerges.But the existing system is generally limited to a simple scene,and must be done in the pre-calibration work carried out,and for the unconstrained person,no calibration and other issues still exist a great distance between accuracy,robustness and robustness and practical application.In order to solve these problems,this paper proposes a non-calibrated driver’s gaze zone estimation method based on BP neural network,which will focus on the three aspects:First of all,this article needs to first obtain the head direction angle and the sight gaze direction angle,in the process,for the driver’s body appears left and right shaking,or different drivers height,occurred relative to the camera left and right offset and up and down offset,A head attitude correction algorithm based on geometric relation is proposed.At the same time,this paper establishes the 3D eyeball model to estimate the pupil line of sight.Then,a non-calibrated driver’s gaze area estimation system based on BP neural network is constructed.The BP neural network model is used to train the driver’s head attitude and sight angle parameters.The regional classifier is constructed and the driver’s gaze area by the network model without calibration is estimated.Finally,the method is evaluated.The experimental results show that the method proposed in this paper can not only meet the requirements of academic research,but also realize the need of the driver’s gaze estimation in complex environment,which satisfies the requirements of real-time,accuracy and robustness of the experiment.And lay a good foundation for safe driving support system. |
PDF Full Text Request