Selective salient feature based lane analysis

Lane analysis involves data-intensive processing of input video frames to extract lanes that form a small percentage of the entire input image data. In this paper, we propose lane analysis using selective regions (LASeR), that takes advantage of the saliency of the lane features to estimate and track lanes in a road scene captured by on-board camera. The proposed technique processes selected bands in the image instead of the entire region of interest to extract sufficient lane features for efficient lane estimation. A detailed performance evaluation of the proposed approach is presented, which shows that such selective processing is sufficient to perform lane analysis with a high degree of accuracy.