The Research On The Recognition Method Of Vehicle Running State In The Lane

The Driver Assistance System based on Machine Vision and Pattern Recognition hasbecome an important research field of Intelligent Vehicle, it is also one of the key difficultiesof research in the field of intelligent transportation. Identify the vehicle’s running state in thelane as a vital function in driver assistant system, which attract more and more attention.In this paper, focus on the recognition method of vehicle running state in the lane,consist of two aspects:For lane recognition, Firstly, finish the gray processing and median filtering. Consideringthe drawbacks of traditional Otsu threshold segmentation algorithm, using the improved Otsuthreshold segmentation algorithm can acquire a large progress in the segmentation effect andprocessing time. According to the actual structure of the road, dividing the lane image intotwo blocks, then use the improved Hough transform to detect lane and use Kalman filtering totrace lane.For the recognition method of vehicle’s running state, which consist of two methods.The recognition method using the calibrated camera need camera’s parameters and accordingto the lane parameters, can estimate the vehicle’s position in the lane, then finish therecognition. The recognition method using the uncalibrated camera, mainly based on thechange of yaw angle. Compare the transformation curve of yaw angle with the lane image,which certify this algorithm can effectively identify the vehicle’s running state.