欢迎来到天天文库
浏览记录
ID:37377935
大小:367.19 KB
页数:6页
时间:2019-05-22
《基于改进光流法电子稳像算法》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、第40卷第3期煤炭学报Vol.40No.32015年3月JOURNALOFCHINACOALSOCIETYMar.2015程德强,郭政,刘洁,等.一种基于改进光流法的电子稳像算法[J].煤炭学报,2015,40(3):707-712.doi:10.13225/j.cnki.jccs.2014.0641ChengDeqiang,GuoZheng,LiuJie,etal.Anelectronicimagestabilizationalgorithmbasedonimprovedopticalflowmethod[J].JournalofChinaCoalSociety,
2、2015,40(3):707-712.doi:10.13225/j.cnki.jccs.2014.0641一种基于改进光流法的电子稳像算法程德强,郭政,刘洁,钱建生,张媛媛,张佳琳(中国矿业大学信息与电气工程学院,江苏徐州221008)摘要:针对煤矿井下机载抖动视频序列低照度、低清晰度以及易受噪声影响的特点,提出了一种基于改进光流法的电子稳像算法。算法利用SUSAN算子获取帧间匹配特征点,并提出基于金字塔改进光流模型,实现抖动视频序列的全局运动矢量计算。为克服机载图像中局部运动目标和移动背景的影响,利用RANSAC算法去除误匹配点,提出稳像参考帧的动态更新机制。针对
3、矿下机载摄像抖动的不确定性,运用均值滤波求解运动补偿矢量,在保留主动运动的同时过滤随机抖动。理论分析和实验结果表明,提出的算法不仅保留了光流法获取运动矢量的精度,而且提高了算法的实时性,对于煤矿井下机载抖动视频序列具有较好的稳像效果。关键词:电子稳像;SUSAN算子;光流中图分类号:TD679文献标志码:A文章编号:0253-9993(2015)03-0707-06AnelectronicimagestabilizationalgorithmbasedonimprovedopticalflowmethodCHENGDe-qiang,GUOZheng,LIUJie,Q
4、IANJian-sheng,ZHANGYuan-yuan,ZHANGJia-lin(SchoolofInformationandElectricalEngineering,ChinaUniversityofMiningandTechnology,Xuzhou221008,China)Abstract:Toimprovethequalityofimagescapturedbymovingundergroundcameraswiththecharacteristicsoflowresolution,lowilluminationandbeingeasilyaffecte
5、dbynoise,anewelectronicimagestabilizationalgorithmbasedonimprovedopticalflowwasproposed.Firstly,byextractinginter-framematchingfeaturepointswiththeSUSANalgo-rithm,animprovedopticalflowmethodbasedonpyramidwasproposedtocalculatetheglobalmotionvectors.Further-more,inordertoovercometheinfl
6、uenceofmovingobjectsandmovingbackgroundwithinimages,theRANSACalgo-rithmwasusedtoeliminatetheerrormatchingpoints,andadynamicupdatemechanismforreferenceframewaspro-posed.Consideringthefeatureofuncertainjitterfromundergroundmovingcarriers,thevectorsofmotioncompensationweresolvedbythemeanf
7、ilteralgorithm,whichcanretaintheactivemovementwhilefilteringtherandomjitter.Theo-reticalanalysisandexperimentalresultsshowthattheproposedmethodcannotonlyretaintheprocessingprecisionformotionvectorbypyramid-basedopticalflowmethod,butalsoimprovethereal-timepropertyofthealgorithm.Ithasa
此文档下载收益归作者所有