本文選題：人體姿態(tài)估計(jì) + 混合部件模型��； 參考：《光學(xué)精密工程》2017年04期

【摘要】：基于混合部件模型的人體姿態(tài)估計(jì)方法忽視了人體結(jié)構(gòu)的對(duì)稱位姿約束關(guān)系,從而導(dǎo)致對(duì)稱部件容易被重復(fù)檢測(cè)、人體姿態(tài)估計(jì)準(zhǔn)確率較低,為此,提出一種基于位姿約束與軌跡尋優(yōu)的姿態(tài)估計(jì)新方法。首先估計(jì)人體單部件和對(duì)稱部件在單幀圖像中的多個(gè)合理位置,利用對(duì)稱部件之間的位姿約束關(guān)系構(gòu)建標(biāo)識(shí)部件。然后根據(jù)單部件和標(biāo)識(shí)部件各自的目標(biāo)優(yōu)化函數(shù),通過動(dòng)態(tài)規(guī)劃算法反復(fù)迭代獲得初始軌跡候選集,再結(jié)合軌跡的全局特征剔除檢測(cè)得分較低的運(yùn)動(dòng)軌跡。最后引入樹形合約模型,聯(lián)系時(shí)空上下文信息,準(zhǔn)確求解出視頻序列光滑且兼容的最優(yōu)軌跡。在N-best、Outdoor Pose和Scene數(shù)據(jù)集中的實(shí)驗(yàn)結(jié)果表明,對(duì)于存在背景復(fù)雜、運(yùn)動(dòng)模糊、部件遮擋等問題的視頻序列中,該方法平均姿態(tài)估計(jì)準(zhǔn)確率達(dá)87%以上,有效減少了對(duì)稱部件的誤判,提高了視頻中人體姿態(tài)估計(jì)的準(zhǔn)確率。
[Abstract]:The method of human body attitude estimation based on hybrid component model neglects the symmetrical pose constraint relation of human body structure, which leads to the symmetry component being easily detected repeatedly, and the accuracy of human body attitude estimation is low. A new attitude estimation method based on pose constraint and trajectory optimization is proposed. Firstly, the reasonable position of human body single component and symmetrical component in a single frame image is estimated, and the identification components are constructed by using the pose constraint relationship between symmetric components. Then according to the objective optimization function of the single component and the marking component, the initial trajectory candidate set is iterated through dynamic programming algorithm, and then the motion track with low score is eliminated by combining the global feature of the trajectory. Finally, a tree contract model is introduced to accurately solve the smooth and compatible optimal trajectory of video sequences with temporal and spatial context information. The experimental results in N-best outdoor Pose and Scene data sets show that for video sequences with complex background, motion blur and component occlusion, the average attitude estimation accuracy of this method is over 87%, which effectively reduces the misjudgment of symmetric components. The accuracy of human pose estimation in video is improved.
【作者單位】： 河海大學(xué)物聯(lián)網(wǎng)工程學(xué)院;常州市傳感網(wǎng)與環(huán)境感知重點(diǎn)實(shí)驗(yàn)室;
【基金】：國(guó)家自然科學(xué)基金(41301448) 江蘇省重點(diǎn)研發(fā)計(jì)劃(BE2016071)
【分類號(hào)】：TP391.41

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 韓貴金;朱虹;;一種基于圖結(jié)構(gòu)模型的人體姿態(tài)估計(jì)算法[J];計(jì)算機(jī)工程與應(yīng)用;2013年14期

2 王浩;劉則芬;方寶富;陳金金;;基于約束樹形圖結(jié)構(gòu)外觀模型的人體姿態(tài)估計(jì)[J];計(jì)算機(jī)科學(xué);2014年03期

3 蘇延超;艾海舟;勞世z，

本文編號(hào)：1991223