On Highway Guardrail Segmentation Algorithms Based on Patrol Car Mobile Video in Complex Environments

In view of the fact that the amount and quality of manual statistics work of ordinary highway guardrail is huge and easy to be missed, the active contour algorithm based on the level set model, phase stretch transformation algorithm, Graph Cut algorithm and Lazy Snapping algorithm are adopted for the purpose of the extraction and segmentation for the guardrail areas in this paper. These four algorithms are respectively applied to three datasets collected in different environments including rural roads, urban roads and the national and provincial highways environments in order to test their effectiveness in handling environment variability. The experimental results show that the improved Lazy Snapping algorithm based on Graph Cut idea can partition the guardrail area well in complex environments. The results provide a basis for the subsequent research on the identification of highway accessory facilities based on mobile image intelligent analysis.