為了檢測光學檢測軟性顯示器的耐用性，本研究創建了一個自動光學檢測系統(optical detection system, ODS)，該系統包括一個類神經網路作為對準控制演算法，可在XYZ平台和一個帶有可旋轉平台四個自由度上的影像擷取儀進行軟性顯示器的光學影像量測。此外，本文提出了一種自動全面性摺疊測試系統(automatic whole-folding testing system, AWTS)，可以為整個樣本的摺疊或彎曲操作調整摺疊或彎曲半徑。在軟性顯示器的全面性摺疊測試期間，軟性顯示器在不同尺寸下需要軟性顯示器的相應旋轉角度以進行垂直法線方向之光學檢測。根據AWTS上不同尺寸大小的軟性顯示器的光學影像量測，本研究利用ODS進行正交對準，並通過類神經網路視覺對準演算法自動將軟性顯示器對齊到顯示面的中心進行光學自動化檢測。最後，本研究利用類神經網路來控制四個自由度之ODS，其最大控制誤差百分比為2%。此外，本研究還針對軟性膽固醇液晶顯示器進行了全面性摺疊測試。結果發現衰減光學特性隨著摺疊次數的增加而增加，並且在全面性摺疊測試之後此軟性膽固醇液晶顯示器的部分區域光學灰度值的最大衰減百分比為10%。因此，根據本研究之量測結果和本研究所創建的類神經網路視覺對位演算法之自動光學檢測方法，其適用於基於全面性摺疊測試來評估軟性顯示器的性能，以提供改善軟性顯示器的軟性特性。

In order to inspect the durability of flexible displays, this study creates an automatic optical detection system (ODS), which includes a neural network as an alignment control algorithm, an XYZ platform with a rotatable stage, and the image capture instrument performed optical image measurement of flexible displays in four degrees of freedom. In addition, this study creates an automatic whole-folding testing system (AWTS), which can adjust the folding or bending radii for the whole sample folding or bending operation. During the whole-folding test of the flexible display, the flexible display requires the corresponding rotation angle of the flexible display in different sizes to perform the optical inspection in the vertical normal direction. According to the optical image measurement of flexible displays based on different sizes on the AWTS, this study uses ODS for orthogonal alignment, and automatically aligns the flexible display to the center of the display surface by using a neural network vision alignment algorithm for automatic optical inspection. Finally, this study uses a neural network to control the ODS of four degrees of freedom, and the maximum control error percentage is 2%. In addition, this study measures the optical characteristics of the flexible cholesteric liquid crystal display for a whole-folding test. As a result, it is found that the attenuation optical characteristics increased as the number of folds increased, and the maximum attenuation percentage of the optical gray value in some local areas after the whole-folding test is 10%. Therefore, based on the measurement results of this study and the neural network vision alignment algorithm created by this study, the automatic optical inspection method is suitable for evaluating the performance of flexible displays based on whole-folding tests to provide the improved flexible characteristics of flexible displays.