數位正弦條紋結構光於三維高度資訊重建之研究

The study of the digital sinusoidal structured light on three-dimensional height information reconstruction

余光翔(Guang-Xiang Yu)

指導教授 ： 蔡曜陽

國立臺灣大學/工學院/機械工程學研究所/碩士(2018年)

https://doi.org/10.6342/NTU201803682

摘要

在視覺系統中，所擷取到的影像皆為二維平面之影像，與實際三度空間物體仍差一深度資訊，本研究中，將以主動非接觸式量測方式取得二維平面之影像點資料，進行運算及分析，進而求出物體的三維高度資訊，以完成逆向工程中的三維輪廓重建技術。
在本研究的實驗中，將以一台投影機與一CCD元件組成三維非接觸式量測系統，透過電腦產生可變週期的弦波條紋結構光，並以微型投影機投射結構光於待測物件上，接著以攝相機拍攝影像。利用四步相位移法與傅立葉轉換法兩種演算過程，計算物體在每個像素點上的真實絕對相位，有相機的像素座標可轉換成二維點資訊，並以相位值計算出物體的每個像素點之高度值，即可獲取物體的三維座標，完成三維表面輪廓重建。
實驗結果顯示，利用四步相位移法與傅立葉轉換法來計算得出三維高度資訊皆可順利重建完成，而在傅立葉轉換法重建過程中，若有使用中值濾波以去除空間雜訊，可有效使重建出的三維高度資訊邊緣平滑化，經去除空間雜訊後較不易出線高度辨識錯誤的現象。同時也比較不同條紋密度對高度資訊獲得的差異性，發現條紋密度較高可以較為有效使細節呈現，但密度過高時(在166mm×166mm的視野下正弦條紋角頻率大於π/12)會使相機辨識產生困難，進而產生相位計算錯誤。實驗驗證此數位結構光投影量測系統可應用於工業實物之三維形貌量測。

關鍵字

數位結構光 ； 影像處理 ； 四步相位移 ； 傅立葉轉換 ； 三維高度資訊

並列摘要

In the vision system, the images captured are two-dimensional images, which lack height information from the actual three-dimensional objects. In this study, the two-dimensional image will be acquired by active non-contact measurement system. And it will be calculated and analyzed, then obtain the 3D height information of the object to complete the 3D contour reconstruction technology in reverse engineering.
In the experiment of this study, a three-dimensional non-contact measurement system consisting of a projector and a CCD component. the computer will generate variable period sinusoidal structured light and the structure light is projected on the object by the projector, then take the image by the camera. Using the four-step phase shift method and the Fourier transform method to calculate the absolute phase of the object at each pixel point. The height value of the pixel points can acquire the height information and reconstruct the three-dimensional surface contour.
The experimental results show that the four-step phase-shifting method and the Fourier transform method can be used to calculate the three-dimensional height information, and the reconstruction can be completed smoothly. In the reconstruction process of the Fourier transform method, if median filtering is used to remove spatial noise, it can be effective, it will smooth the reconstructed 3D height information edge. Besides, the difference between the stripe density and the height information is that the higher the stripe density can effectively make the details appear, but when the density is too high (the sinusoidal fringe angle frequency is larger than π/12 in the 166mm×166mm field of view). Camera identification is difficult, resulting in phase calculation errors. Experiments verify that this digital structure light projection measurement system can be applied to the three-dimensional shape measurement of industrial objects.

並列關鍵字

Digital structured light ； Image processing ； Four-step phase shifting ； Fourier transform ； Three-dimensional height information

參考文獻

[1] Srinivasan, V., H. C. Liu, and M. Halioua, "Automated phase-measuring profilometry of 3-D diffuse objects," Applied Optics, vol.23, pp.3105-3108, 1984.

[2] P. Jordi, J. Salvi, G. Refael and M. Carles, “Overview of coded light projection techniques for automatic 3D profiling.” IEEE International Conference, vol. 1, pp. 133-138, 2003.

[3] Pavageau, S., R. Dallier, N. Servagent, and T. Bosch, "A new algorithm for large surfaces profiling by fringe projection," Sensors and Actuators, vol.115, no.2-3, pp.178-184, 2004.

[4] Quan, C., X. Y. He, and C. F. Wang, "Shape measurement of small objects using LCD fringe projection with phase shifting," Optics-81-Communications, vol.189, no.1-3, pp.21-29, 2001.

[5] L .Douglas, and G. Taubin, "Build your own 3D scanner: 3D photography forbeginners, " ACM SIGGRAPH 2009 Courses, no.8, 2009.