在本篇研究之中,我們改進了一般3D模組貼圖技術的麻煩處,讓使用者不必再另行手動點選特徵點來進行貼圖以及提供了一個自動化3D人物材質恢復之系統。在系統中首先針對一個3D立體人物模型將它讀入,接著利用改良後之SDF演算法計算出人物之SDF值。接著,我們以這些資訊用GMM演算法將3D人物物件做初始分群並且針對分群結果進行修補。再來,將此分群之結果和人物材質依照分群後之各個部位進行分開比對。最後,我們以此比對結果來對於3D人物模組進行材質恢復。藉由這個技術,我們只需要取得一個3D人物模組和不同人物之不同角度的材質,我們就可以自動地針對3D人物做材質貼上。 In this research, we proposed an approach for 3D object texture mapping and an automatic recovery 3D object surface system. Users don’t need to select feature points for mapping by themselves, it’s the most difficult challenge of 3D object texture mapping in this paper., the first step, we load a 3D object model and compute the SDF value by SDF algorithm. And then, we use GMM algorithm to segment the 3D object model and modify the results without noise. After above steps, we compare the different parts between clustering results and textures. Finally, we can recover the 3D object model according to these matching results. With this technology, we just need a 3D model and texture with different views to get a new 3D model with our own person’s photos. This method is total automatically in texture mapping for 3D model.