Computer Science · Computer Vision and Pattern Recognition
Isolating Multiple Image Motions for Enhancement and 3D Analysis
Michal Irani, Benny Rousso, Shmuel Peleg
See on the mapAbstract
Motion computation in scenes having multiple moving objects is performed together with object segmentation by using a temporal integration approach. Using an accurate 2D motion estimate for image regions, they can be enhanced by fusing all successive frames covering the same region. Enhancement includes improvement of image resolution and filling-in occluded regions. It is also shown how an accurate 2D motion estimate for a single planar surface in a general static scene can help to compute the 3D motion performed by the camera. 1 Introduction A method for detecting and tracking multiple moving objects, using both a large spatial region and a large temporal region, is described. When the large spatial region of analysis has multiple moving objects, the motion parameters and the locations of the objects are computed for one object after another. The method has been applied successfully to 2D affine and projective motions in the image plane. Once an object has been tracked and segmente...