文档介绍：2007 ? 8 ? Au g ust 2007 — 147 —???? puter Engineering ? 33 ? 1 6 ? V o l ?. 3 3 N o .16 · ?????????· ??щ?? 1 000 — 34 28( 200 7) 16 — 0147 — 03 ?????? A ?????? TN91 1. 73 基于相关匹配及自适应模板更新的目标跟踪新方法黄飞,李德华,姚迅( 图像信息处理与智能控制国家教育部重点实验室,华中科技大学图像识别与人工智能研究所,武汉 430074) 摘要: 传统的相关匹配算法在背景比较简单的情况下可以较好地跟踪到目标, 但实际获得的图像存在变形、噪声、遮挡等问题, 并且也很难获得比较好的跟踪效果。针对上述问题,提出了一种修正的 MCD 相关匹配算法和多帧累积的模板更新策略, 并对实际图像进行了仿真。实验结果表明,算法在图像存在变形、噪声、遮挡时也可以达到比较理想的跟踪效果。关键词: 相关匹配; MCD ;多帧累积;自适应模板修正 New T a rget T r acking Method Based on Corr elation Matching and Adaptive Modal Modification HUANG Fei, LI De-hua, Y A O X un (S tate m i ssion Key Laboratory for Im age Processing and Intelligence Contro l, Institute for Pattern Recognition and Artificial Intelligence, Huaz hong University of Science and T echnology , W uhan 430074) 【 Abstract 】 Sy stem s based on cor r e lation- tr acking can tr ack objects based on 2- dim e nsional shape, how ever tr aditional cor r e lation- tr acking m e thods have dif f i c ulty in detecting an object when the object is tr ansfor m e d or par tially o b scur ed by clutter . T h is paper desc r i bes one of the m a in p r oblem s of im ag e- based cor r elatio n tr acking sy stem s, and pr oposes a new solution of using m odifi ed M C D corr elation m a tching and adaptive m odal m odification based on m u ltipl e fra m e cu m u l a tion. The syste m is te st