Disclosed in the present application are a method and device for association of fuzzy data of multiple targets in a video. The method comprises: performing online target motion detection on a current video frame, and using detected possible moving objects as observed results; calculating occlusion degrees between the predicted results of different targets in the current video frame; respectively determining, according to the occlusion degrees, whether an occlusion occurs between each predicted result and other predicted results; if there is no occlusion between the predicted result and any other predicted results, performing first data association between the predicted result and the observed results; and if there is an occlusion between the predicted result and any other predicted results, performing second data association between the predicted result and the observed predicted results, the first data association being different from the second data association. In this way, the accuracy of multi-target tracking in a complex environment can be improved.
展开▼