Name
Abstract | ||
|---|---|---|
A method of automatically and accurately detecting regions of sidewalks is useful for assisting pedestrians. There are several types of sidewalk borders: steps, white lines, guardrails, walls, plants, and ditches, for example. However we can assume that the structure of an environment along some direction is uniform for several meters regardless of the components of the environment. Therefore, we present a method to detect the direction and extract borders of sidewalks by using this assumption. First, an inputted image is transformed to a rectified horizontal map by using affine transformation, and the map is divided into small regions. Each region represents about 5 × 5-10 meters. Then, the direction of each region is determined by calculating the sharpness in the change of the average intensity of each column. Moreover, candidates of a borderline are detected by using the sharpness graph. These candidates are refined by using the dynamic contour model method. Finally, the two most interior lines are detected as borders of the sidewalk. We applied this method to about 10,000 images, which are obtained by a test run in outdoors, and detected a border in about 80% of the images at this time. |
Year | Venue | Keywords |
|---|---|---|
2007 | Artificial Intelligence and Applications | low-speed buggy,sharpness graph,interior line,sidewalk border,small region,affine transformation,horizontal map,dynamic contour model method,inputted image,white line,average intensity,intelligent transportation system,pattern recognition |
Field | DocType | Citations |
Affine transformation,Computer vision,Graph,Computer science,Artificial intelligence,Intelligent transportation system | Conference | 0 |
PageRank | References | Authors |
0.34 | 2 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Kentaro Kayama | 1 | 22 | 5.57 |
| Ikuko Eguchi Yairi | 2 | 67 | 14.16 |
| Seiji Igi | 3 | 83 | 14.01 |