Vehicle Lane Position Estimation with Camera Vision using Bounded Polynomial Interpolated Lines

Applications of camera vision, such as lane departure warning systems, are limited by the quality of the frame image and the information contained within each frame. One common feature extraction technique in image processing is the use of the Hough transform, which can be used to extract lines from an image. The detected lane marking lines are used in the interpolation of a 2 nd order polynomial to estimate the shape of the lane marking’s curve in the image. However, blurry frames, additional road markings on the ground, and adverse weather conditions can ruin detection of these valid lane lines. To eliminate erroneous lines, a technique has been employed which bounds the previously detected 2 nd order polynomial with two other polynomials that are equidistant from the original polynomial. These bounding curves employ similar characteristics as the original curve; therefore, the valid lane marking should be detected within the bounded area given smooth transitions between each frame. The effects of erroneous lines within this bounded area can be reduced by employing a Kalman filter on the coefficients of the 2 nd order polynomial. The filter also allows for smooth transitions between curved and straight roads. The measurement of the position within the lane is carried out by determining the number of pixels from the center of the image and the estimated lane marking. This measurement value can then be converted to its real world equivalent and used to estimate the position of the vehicle within the lane. This technique is verified by comparing lateral distance measurements from RTK GPS measurements and the measurements from a camera. Results will show that this method performs well on straight roads but fails to perform well on curves.