基于卡尔曼滤波的高速运动模糊图像自适应复原方法

打开文本图片集

关键词：卡尔曼滤波；高速运动；模糊图像；自适应复原；图像退化模型；加权系数；噪声向量；状态预测中图分类号：TN911.73-34；TP391 文献标识码：A 文章编号：1004-373X（2026）09-0038-04

High-speed motion blur image adaptive restoration method based on Kalman filtering

Yang Yan，Li Zilin，Li Zheng （TianjinChengjianUniversity，Tianjin3Oo384，China）

Abstract：Thenoiseinterferenceof high-speedmotion blurimage isdynamicandchangeable，whichbringssome chalenges toaccuraterestoration.TheadaptiveKalmanflteringcanestimateandcorrcttheparametersofthe Kalmanfiltering modelaccordingtothereal-tiechangeofthenoiseintheimage，optimizethefiteringdesignandreducethefilteringeroand ensure theacuracyoftheKalman filtering.Therefore，ahigh-speedmotionblurimageadaptiverestorationmethodbasedon Kalmanfilteringissudied.Aigh-speedmotionblurimagedegradationmodelisconstructed，ndinitialfilteringstoation processingisperformedonthedegradationmodelbybasicKalmanfiltering.Theweightedcoeficientisintegratedintothebasic Kalmanfltering toobtaintheadaptive Kalman filtering.According tothedynamicchangeofthenoisecharacteristicsofthehighspedmotionblurimage，thenoisevectoroftheimageiscorectedinrealtime，andthestatepredictionvalueoftheimageis updateddynamicalltoobtaintherestoredimagewithhighconsistencywiththerealhigh-speedmovingimage.Takingspeed skatingasanexample.Theproposedmethod isusedtorestorethehigh-speedmotionblurimageofspeedskating.The experimentsshowthatthefeaturesimilarityandedgepreservationefectoftherestoredspeedskatingimagearegood，theoise interferenceintheimageisfilteredouteffectively，andtheoverallqualityof theimageisimprovedsignificantly.

Keywords：Kalman filtering;high-speedmotion;blur image;adaptiverestoration;image degradationmodel；weighting coefficient;noise vector; state prediction

0 引言

高速运动模糊图像成为摄影摄像技术应用中一种较为常见的图像退化问题，其形成主要原因是在拍摄过程中，被拍摄人体或物体的快速运动以及相机本身存在抖动造成图像出现运动模糊现象[1-3]。（剩余5318字）