This paper proposes an algorithm capable of extracting precise lane information from images captured by a camera on structured roads with line-markers. To this end, the proposed algorithm uses four main techniques: extraction and grouping of line-boundary pixels, Gaussian-probability computation, weighting of region of interest (ROI), and particle filtering. Eventually, the algorithm can provide precise and stable lane information to driver assistant systems such as lane-keeping or lane-centering system. The results of experiments show that the algorithm is well adapted to various road conditions without localization errors in orientation and position.