李天峰
摘 要: 針對(duì)基于自標(biāo)定的三維人物圖像動(dòng)態(tài)重構(gòu)方法不能準(zhǔn)確獲取人物運(yùn)動(dòng)形態(tài)的三維位置,重構(gòu)結(jié)果存在較高的偏差的問(wèn)題,提出基于多媒體技術(shù)的三維人物圖像動(dòng)態(tài)重構(gòu)方法,采用單目圖像重構(gòu)算法完成人體三維姿態(tài)重構(gòu)。對(duì)多媒體技術(shù)中單攝像機(jī)拍攝的單目人體運(yùn)動(dòng)圖像進(jìn)行重構(gòu)時(shí),首先基于人體樹(shù)狀模型,采用基于部件檢測(cè)器的二維肢體檢測(cè)算法完成人體二維肢體檢測(cè);再采用基于關(guān)節(jié)點(diǎn)圖像坐標(biāo)的三維姿態(tài)重構(gòu)算法,依據(jù)人體二維姿態(tài)檢測(cè)結(jié)果,通過(guò)預(yù)測(cè)關(guān)節(jié)點(diǎn)反投影誤差的退火粒子濾波算法完成人體三維姿態(tài)的跟蹤,實(shí)現(xiàn)三維人體圖像動(dòng)態(tài)重構(gòu)。實(shí)驗(yàn)結(jié)果說(shuō)明,所提方法可準(zhǔn)確實(shí)現(xiàn)三維人物圖像動(dòng)態(tài)重構(gòu),具有較高的重構(gòu)精度和效率。
關(guān)鍵詞: 多媒體技術(shù); 三維人物圖像; 三維姿態(tài)重構(gòu); 退火粒子濾波算法; 重構(gòu)精度; 姿態(tài)跟蹤
中圖分類(lèi)號(hào): TN911.73?34; TP181 文獻(xiàn)標(biāo)識(shí)碼: A 文章編號(hào): 1004?373X(2018)09?0068?04
Abstract: The traditional three?dimensional (3D) character image dynamic reconstruction method based on self?calibration can′t acquire the 3D position of characters motion morphology accurately, and has high deviation of reconstruction results. According to the above problems, a 3D character image dynamic reconstruction method based on multimedia technology is put forward, in which the monocular image reconstruction algorithm is used to reconstruct the 3D posture of human body. For the reconstruction of monocular human movement image taken by a single camera in multimedia technology, the 2D limbs detection algorithm based on component detector is used to detect the 2D limbs of human body according to the tree?shaped model of the human body. The 3D posture reconstruction algorithm based on image coordinate of joint points is used, and the annealing particle filtering algorithm for the prediction of joint point back projection error is adopted to track the 3D posture of human body accor?ding to detection results of human body′s 2D posture, so as to realize the 3D character image dynamic reconstruction. The experimental results indicate that the proposed method can realize the dynamic reconstruction of 3D character image accurately, and has high reconstruction accuracy and efficiency.
Keywords: multimedia technology; 3D character image; 3D posture reconstruction; annealing particle filtering algorithm; reconstruction accuracy; posture tracking
0 引 言
隨著計(jì)算機(jī)動(dòng)畫(huà)和計(jì)算機(jī)視覺(jué)等多媒體技術(shù)的快速發(fā)展,對(duì)運(yùn)動(dòng)人體三維姿態(tài)進(jìn)行準(zhǔn)確的動(dòng)態(tài)跟蹤,在監(jiān)控、體育、醫(yī)療以及影視制作等領(lǐng)域具有重要的應(yīng)用價(jià)值。人體三維姿態(tài)重構(gòu)是基于單視角以及多視角拍攝數(shù)字圖像以及視頻等多媒體資源,獲取包括RGB?D信息以及輪廓信息等原始數(shù)據(jù),對(duì)這些數(shù)據(jù)進(jìn)行分割和特征采集,采用重建算法獲取人物圖像動(dòng)態(tài)重構(gòu)結(jié)果,恢復(fù)出目標(biāo)人物圖像的三維空間信息的過(guò)程[1]。
傳統(tǒng)基于自標(biāo)定的三維人物圖像動(dòng)態(tài)重構(gòu)方法不能準(zhǔn)確獲取人物運(yùn)動(dòng)形態(tài)的三維位置,重構(gòu)結(jié)果存在較大的偏差。因此,本文提出基于多媒體技術(shù)的三維人物圖像動(dòng)態(tài)重構(gòu)方法,獲取高精度的三維人物圖像重構(gòu)結(jié)果。
1 多媒體技術(shù)的三維人物圖像動(dòng)態(tài)重構(gòu)
本文主要對(duì)人體的姿態(tài)進(jìn)行分析,采用多媒體技術(shù)中的單目圖像重構(gòu)算法,實(shí)現(xiàn)人體三維圖像動(dòng)態(tài)重構(gòu)。
1.1 單目圖像的人體二維姿態(tài)估計(jì)
對(duì)多媒體技術(shù)中單攝像機(jī)拍攝的單目圖像中的人體二維姿態(tài)進(jìn)行估計(jì)時(shí),采用基于部件檢測(cè)器的二維肢體檢測(cè)算法,通過(guò)優(yōu)化的形狀上下文采集人體不同肢體的特征,再對(duì)獲取的特征實(shí)施離線(xiàn)學(xué)習(xí),訓(xùn)練對(duì)應(yīng)的分類(lèi)器,檢測(cè)目標(biāo)人體的姿態(tài),獲取相關(guān)部位的檢測(cè)概率,預(yù)測(cè)出人體不同肢體部位的位置。通過(guò)依據(jù)部件的方法,基于人體樹(shù)狀模型,將人體看成是15個(gè)肢體部位形成的總體,分別對(duì)15個(gè)部位實(shí)施訓(xùn)練和檢測(cè)[2],獲取準(zhǔn)確的人體二維姿態(tài)預(yù)測(cè)結(jié)果。將人體各肢體部位看成是一個(gè)部件,其狀態(tài)描述為[li=pi,θi,si]。[pi=(xi,yi)]用于描述第[i]個(gè)部件的中心在圖像