摘要： 基于人体部件的树形模型表达直观且计算高效,被广泛应用在人体姿态估计中。然而模型本身在部件特征表达上的不足限制了姿态估计结果的准确度,为此,提出一种基于图结构模型和新型纹理特征的人体姿态估计算法。采用改进后的外观模型,从训练集中获得部件位置的先验知识,联系相邻部件之间的关系,并将其应用于测试图像的外观模型建模阶段。应用Haar 型局部二值模式(HLBP)纹理特征,提取部件的纹理信息,对图像进行分 块处理,并为每一块赋予不同的权重。实验结果表明,带权重的HLBP 特征能更有效地提取部件的纹理特征,与HLBP 特征、归一化HLBP 特征和颜色特征相比能获得更高的准确度。

关键词: 计算机视觉, 人体姿态估计, 外观模型, 特征提取, 纹理特征, 加权Haar 型局部二值模式特征

Abstract: Tree structured model based on body parts is widely used in human pose estimation filed due to its intuitive expression and effective calculation. In order to address the problem of low accuracy of pose estimation caused by insufficient feature expression of body part models,a pose estimation method based on pictorial structure and novel text feature is proposed. Better appearance model is adopted,prior and latent relationships of different body parts are learned from annotated images then help in estimating better appearance models on test images. Haar characteristics Local Binary Pattern(HLBP) text feature is used to extract the text information of body parts. Furthermore,an image is block processed and different weight is assigned to different block. Experimental results show that when compared with HLBP,normed HLBP and color feature,Weighted HLBP(WHLBP) captures text feature more effectively and gains higher accuracy.

Key words: computer vision, human pose estimation, appearance model, feature extract, texture feature, Weighted Haar characteristics Local Binary Pattern(WHLBP)

中图分类号: 

• TP391