| Summary: | 碩士 === 國立成功大學 === 資訊工程學系 === 102 === Lane departure warning is one of important parts in intelligent vehicle systems. This paper purpose a technique for lane markings detection and tracking. At first, lane boundaries are detected by line Hough transform. In edge detection, previous step of line Hough transform, we use Otsu’s method as an adaptive-thresholding method for Canny edge detection. To filter undesired detected line, so called noises, by line Hough transform, we employ two approaches, angle of lines and vanishing points, to come to the purpose. After lane markings are detected, we use lane tracking which is lower computation complexity and execution time to make our lane departure warning system real-time. Between detection and tracking, two judgment mechanisms are purposed. One is used to judge that it is going to detection or tracking, the other is used to judge whether lane departure. We collected several video and separate into two databases by cases to do experimenting.
|
|---|
