DEVICE AND METHOD FOR IDENTIFYING AND TRACKING OBJECTS IN A VIDEO SEQUENCE

摘要

A device for identifying and tracking objects in a video sequence composed of temporally successive images consisting of pixels, wherein the device comprises: a segmentation stage for segmenting at least a first image and a temporally succeeding second image; a correspondence determining stage for determining a plurality of correspondence pairs each comprising a pixel from the pixels of the first image and a pixel corresponding thereto from the pixels of the second image; an assignment stage for assigning the segmentation entity information and the set of items of probability information of at least one pixel of the first image on the basis of displacement vectors with respect to a relevant pixel of the second image; an object identifier assignment stage for assigning the pixels of one of the segmentation entities of the first image and the pixels of one of the segmentation entities of the second image to a common object identifier if the relevant segmentation entity of the first image and the segmentation entity of the second image image the same object; a fusion stage for fusing the set of items of probability information from the first image, said set being assigned respectively to a pixel of the second image, and the set of items of probability information of the relevant pixel of the second image; and an object category determining stage for determining the relevant object category of the segmentation entities.