[发明专利]基于跨域字典学习的人体动作识别方法

说明书

技术领域

本发明涉及计算机视觉任务技术领域，特别涉及一种基于跨域字典学习的人体动作识别方法。

背景技术

最近，针对稀疏表达的字典学习吸引了大量的注意力。其已经成功的应用到了很多的计算机视觉任务中，比如脸部识别(参见文献“J.Wright,A.Y.Yang,A.Ganesh,S.S.Sastry,and Y.Ma.Robust face recognition via sparse representation.Pattern Analysis and Machine Intelligence,IEEE Transactions on,31(2):210–227,2009.”)和图像降噪(参见文献“M.Zhou,H.Chen,J.Paisley,L.Ren,G.Sapiro,and L.Carin.Non-parametric Bayesian dictionary learning for sparse image representations.2009.”)。利用一个过完备的字典，稀疏信号模型能够通过字典基的稀疏线性表达来估计输入信号。根据不同的准则，许多方法(参见文献“H.Lee,A.Battle,R.Raina,and A.Y.Ng.Efficient sparse coding algorithms.Advances in neural information processing systems,19:801,2007.”，“J.Wang,J.Yang,K.Yu,F.Lv,T.Huang,and Y.Gong.Locality-constrained linear coding for image classification.In Proc.IEEE Conf.Computer Vision and Pattern Recognition,pages3360–3367.IEEE,2010.”，“J.Wright,A.Y.Yang,A.Ganesh,S.S.Sastry,and Y.Ma.Robust face recognition via sparse representation.Pattern Analysis and Machine Intelligence,IEEE Transactions on,31(2):210–227,2009.”)被提出来学习这样一个字典。K-奇异值分解(K-SVD)算法(参见文献“M.Aharon,M.Elad,and A.Bruckstein.K-svd:An algorithm for designing overcom-plete dictionaries for sparse representation.Signal Processing,IEEE Transactions on,54(1):4311–4322,2006.”)是一个经典的字典学习算法，同时也是k-means聚类算法的泛华。K-SVD算法专注于对信号的重建能力，但是在另一方面，因为学习过程是非监督的，其判别性能并没有获得足够的关注。这样一来，许多将判别性准则加入字典学习的方法在(参见文献“Q.Zhang and B.Li.Discriminative k-svd for dictionary learning in face recognition.Proc.IEEE Conf.Computer Vision and Pattern Recognition,pages2691–2698,2010”，“J.Yang,K.Yu,and T.Huang.Supervised translation-invariant sparse coding.Proc.IEEE Conf.Computer Vision and Pattern Recognition,pages3517–3524,2010.”，“J.Marial,M.Leordeanu,F.Bach,M.Hebert,and J.Ponce.Discriminative sparse image models for class-specific edge detection and image interpretation.European Conference on Computer Vision,pages43–56,2008.”，“J.Mairal,F.Bach,J.Ponce,G.Sapiro,and A.Zisserman.Supervised dictionary learn-ing.Advances in Neural Information Processing Systems,page10,2009.”，“Y.Boureau,F.Bach,Y.LeCun,and J.Ponce.Learning mid-level features for recogni-tion.Proc.IEEE Conf.Computer Vision and Pattern Recognition,pages2559–2566,2010.”)被提出。除了字典学习的判别性能力以外，其他在字典学习原型上设计的目标函数包括多字典学习(参见文献“W.Zhang,A.Surve,X.Fern,and T.Dietterich.Learning non-redundant codebooks for classifying complex objects.In International Conference on Machine Learning,pages1241–1248.ACM,2009.”)，指定类的字典学习(参见文献“L.Yang,R.Jin,R.Sukthankar,and F.Jurie.Unifying discriminative visual codebook generation with classifier training for object category recognition.In Proc.IEEE Conf.Computer Vision and Pattern Recognition,pages1–8.IEEE,2008.”)等等。不同于大多数分别学习字典和分类器的方法，(参见文献“Q.Zhang and B.Li.Discriminative k-svd for dictionary learning in face recognition.Proc.IEEE Conf.Computer Vision and Pattern Recognition,pages2691–2698,2010.”，“Z.Jiang,Z.Lin,and L.S.Davis.Learning a discriminative dictionary for sparse coding via label consistent k-svd.Proc.IEEE Conf.Computer Vision and Pattern Recognition,pages1697–1704,2011.”)的作者将两个学习过程融合到了一个单一的收监管的优化过程中，并且同时学习一个判别性的字典和与其相应的分类器。更进一步，Qiu等人(参见文献“”)和Zheng等人(参见文献“J.Zheng,Z.Jiang,J.Phillips,and R.Chellappa.Cross-view action recognition via a transferable dictionary pair.In Procedings of the British Machine Vision Conference,pages125–1,2012.”)设计了满足训练数据和测试数据具有不同分布条件的字典学习。前者提出一个能够将从一个域中学习得到的字典转移到另一个域中的总体联合优化函数，并且将其应用到姿态对齐，姿势和亮度估计以及脸部识别的应用中。后者在跨视角动作识别问题上通过不同视角间对应信息所建立的字典对取得了很好的结果。为了利用那些与目标域数据不太相关的数据，Raina等人(参见文献“R.Raina,A.Battle,H.Lee,B.Packer,and A.Y.Ng.Self-taught learning:transfer learning from unlabeled data.In international conference on Machine learning,pages759–766.ACM,2007.”)[21]提出了一种对未标记的数据进行稀疏编码来将源域中的大量数据分解为基本模块(比如图像分类中的图像边界)。这样一来，信息能够从底层传递到高层。