【摘要】： 基于樣例的表情生成是指利用一對(duì)樣例表情人臉,通過表情計(jì)算生成一致的目標(biāo)表情人臉。通過基于樣例人臉的表情生成方法研究,探討人臉建模、表情映射和紋理貼圖等關(guān)鍵技術(shù),對(duì)深入研究逼真的表情合成方法具有重要的意義。此外,表情生成方法在智能人機(jī)交互、計(jì)算機(jī)動(dòng)漫、數(shù)字娛樂等領(lǐng)域具有廣泛的應(yīng)用價(jià)值。? 論文的主要工作和創(chuàng)新如下:? 首先,本文采用CANDIDE 3模型作為標(biāo)準(zhǔn)的人臉網(wǎng)格模型。通過基于OpenGL的特征點(diǎn)拾取技術(shù)的人機(jī)交互方式對(duì)CANDIDE 3三角網(wǎng)格模型上頂點(diǎn)的進(jìn)行調(diào)整后,實(shí)現(xiàn)了CANDIDE 3標(biāo)準(zhǔn)模型與特定人臉的匹配,最后得到了一個(gè)符合特定人臉特征的三角網(wǎng)格模型。? 其次,對(duì)樣例人臉的三角網(wǎng)格模型進(jìn)行傾斜度校正等預(yù)處理后,本文實(shí)現(xiàn)的基于樣例的表情映射算法把樣例人臉的表情運(yùn)動(dòng)數(shù)據(jù),映射到目標(biāo)人臉網(wǎng)格模型上,計(jì)算出目標(biāo)人臉相應(yīng)表情的三角網(wǎng)格模型。由于表情映射后網(wǎng)格模型上存在一些奇異點(diǎn),本文對(duì)這個(gè)網(wǎng)格模型添加了兩個(gè)約束,一是對(duì)模型上網(wǎng)格點(diǎn)的運(yùn)動(dòng)范圍進(jìn)行約束,二是依據(jù)人臉的對(duì)稱性對(duì)網(wǎng)格模型上的網(wǎng)格點(diǎn)添加對(duì)稱性約束。? 第三,基于OpenGL紋理映射技術(shù)的圖像變形算法,對(duì)表情映射生成的三角網(wǎng)格模型進(jìn)行紋理貼圖。本算法依據(jù)目標(biāo)人臉的中性表情圖像、中性表情對(duì)應(yīng)的人臉三角網(wǎng)格模型和表情映射后的三角網(wǎng)格模型數(shù)據(jù),得到了目標(biāo)人臉的逼真的不走樣的表情圖像。? 最后,本文對(duì)所提出的表情生成算法的相關(guān)步驟逐步進(jìn)行了一系列實(shí)驗(yàn),最后對(duì)整體算法給出了綜合實(shí)驗(yàn)和交叉驗(yàn)證實(shí)驗(yàn),驗(yàn)證了算法的有效性。?
[Abstract]:Example-based expression generation refers to the use of a pair of sample facial expressions, through expression calculation to generate consistent target facial expressions. Through the research of expression generation method based on sample face, this paper discusses the key technologies such as face modeling, expression mapping and texture mapping, which is of great significance to the further study of realistic expression synthesis method. In addition, expression generation method has a wide range of application value in intelligent human-computer interaction, computer animation, digital entertainment and other fields. The main work and innovations of the paper are as follows: Firstly, the CANDIDE 鈮，

本文編號(hào)：2483675