Extraction of vehicle image from panoramic street-image

It is important to assess street-parking vehicles causing traffic problems in urban cities, however, it is performed manually and at a high cost. It is a top priority for reducing costs, to develop a detection system of those vehicles. We introduce a panoramic street-image, combined with view- and range- images. Panoramic street-image possibly provides useful information for our daily life. We propose a detection method, using a laser-range finder and a line-scan camera. Two kinds of cluster analysis are applied to range points: one is for clustering points at each scan, and the other for clustering points over several scans, each cluster of range points meaning a vehicle. As a result of verification experiments in real roads, a detection rate of 90 % is reached. Based on the results of clustering range data, all vehicle images are extracted from the corresponding panoramic street view-image. The error in extraction was within two scans of the laser-range finder.