Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
525305 | Transportation Research Part C: Emerging Technologies | 2008 | 20 Pages |
We developed a line-based skid mark segmentation and measurement system to solve issues related to randomness from pavement texture and measurement subjectivity at car accident scenes. The system was designed to operate along the longer straight lines that exist in boundaries between skid marks and pavement at the scene of an accident. The operational system consists of two processes: preprocessing and feature extraction. Preprocessing steps include skid mark positioning, slope angle detection, and segmentation, whereas feature extraction involves detecting light striations, striation segmentation, and calculating the widths of striations from images. Experimental validation and objective measurements illustrate that this system saves operation time and cost while performing with accuracy similar to that of manual methods.