Abstract
In this paper, we proposed a moving object extraction by level set method considering depth information obtained from a single camera image. Many conventional methods for moving object extraction are based on the inter-frame difference information, but accuracy degradation of extraction in uncovered background and occlusion region is easy to happen by the use of only frame difference information. In our proposal method, we improve the accuracy of moving object extraction by design of a new speed function of level set method using both edge information and depth information obtained from a single image as well as inter-frame difference information. The experimental results through computer simulations show the proposed method provides good extraction accuracy under the condition that there is significant difference of depth information between the object and background.
