| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To solve a problem that undesired pedestrians, electric poles, street trees, or the like are sometimes included as obstacles in an image when imaging a street by an in-vehicle camera or in a video image captured by a handheld camera, wherein there are few practical systems to remove such obstacles captured by a moving camera from the image by combining existing technologies, and in an actual system, it is difficult to remove the obstacles with high accuracy and stability. <P>SOLUTION: A technique to efficiently remove only the foreground objects using spatio-temporal image analysis is proposed. In particular, a technique to automatically remove the foreground objects using plane-based SFM is proposed by utilizing that the town areas like buildings, houses, etc. as imaging objects are mainly constituted by planar structures. By using the spacio-temporal analysis, the complex obstacles like pedestrians, street trees, electric poles, etc. in an outdoor environment, which has been difficult to remove in the past techniques, can be efficiently removed. <P>COPYRIGHT: (C)2011,JPO&INPIT</p> |
