影像量測技術於近年來開始廣泛地被應用於地震工程結構實驗。影像量測過程中發生的相機微量偏移是一個難以避免的誤差來源。在實驗的過程中,相機應保持穩固不動以避免在三角定位時產生誤差。而可能讓相機晃動的原因有-戶外實驗的強風、三角架的不穩固,以及任何非預期的人為碰觸。在長達數小時的量測中,即使是相當微量的相機偏移都將導致後續影像分析的嚴重誤差。 本研究在實驗進行時在地面上貼設數個不動點,做為相機偏移及相片校正的依據。本文利用第一張照片的不動點影像座標做為初始依據,計算後續照片的移動量及旋轉量。接著對於偏移後照片進行重新取樣,產生修正後的照片,以利進行後續的影像分析。本研究以國家地震工程研究中心執行的單柱搖擺實驗,驗證相機偏移之誤差修正的實用性。本文將探討雙相機量測系統與FPGA (Field-Programmable Gate Array)系統整合的可行性評估。
Image-based measurement is getting widely used in earthquake engineering structural experiments. Slight camera movement during an experiment is sometimes an inevitable source of measurement errors, which may result in considerable error of stereo triangulation. Camera movement is typically induced by slight tripod deformation, unexpected touching of imaging equipment, outdoor wind, and other factors. This work developed an adjustment method to minimize the error induced by camera movement. This method requires several fixed reference points to be captured in the images. This method estimates the translational and rotational movement of images based on the translational movement of the fixed reference points in the images. Adjusted images are then generated based on estimated movement, and are used for follow-up image analysis process. An RC-rocking experiment conducted at National Center for Research on Earthquake Engineering was used to test the adjustment method. Finally, the feasibility of integrating stereo triangulation image-based analysis and Field-Programmable Gate Array (FPGA) system is discussed in this work.