現今社會3D建模技術越來越廣泛應用，也成為時下流行的趨勢，例如3D列印表機可以透過三維掃描儀得到三維資訊列印出來並加以應用等等。本論文以條紋投影輪廓儀(Projection Fringe Profilometry，簡稱PFP)為應用，此優點在於非接觸式測量，可避免碰觸物體導致物體受損，再來操作時間短，架設方便，可即時取得三維資訊。但在過去的研究中發現，所得到的條紋圖像背景有顆粒狀的光斑成隨機分布，因此本論文主題著重在後端演算裡增加消除雜訊演算法來降低雜訊，在進行相位展開之前，選用改良版中值濾波，針對截斷線上做紋理分析，保留住該有特性，並在相位展開之後加入對數螺線演算法消除雜訊，使還原結果更接近原始深度資訊，最後用RMSE比較結果。 3D modeling technology is widely used in decades, the applications including the most popular 3D printing that prints out the object scanned by a 3D scanner. This research work employs the fringe projection techniques to construct the 3D model. The fringe projection technique has the advantages of non-contact measurement to avoid touching objects that may damage the objects and fast operation time. However the fringe image may have random distributed granular spots (noise). This research work tries to reduce the noise caused by the fringe projection techniques when constructing a 3D object. First, this research uses a median filter to filter out some noises, which can preserve the characteristics of the phase wrapped map before the procedure of phase unwrapping. To eliminate the noise further, after phase unwrapping we apply our proposed logarithmic spiral algorithm to the unwrapped figure. Our approach can make the reconstructed 3D figure much better than that reconstructed by other approaches.
