| 摘要: | 在工業4.0掀起全球製造業巨大浪潮之中，高速、非接觸且非破壞性的光學三維造影與檢測方式，已成為國內外工業產業之重要感測技術與發展驅勢。目前數位全像術運用數位化波前重現方式完整記錄與重建物體三維資訊，成為新世代三維造影與精準量化分析的新利器，而結構光條紋投影技術則為非接觸式全域光學形貌量測技術，其運用形變條紋方式以解析物體三維輪廓。由於檢測物件往往同時具有漫射與鏡射的複合表面特性，因此，具三維複合表面物體所需快速而有效的全場造影與量測方法目前仍待開發；另外，近年來全球智慧製造中少量多樣的生產比例提升，產業面對新生產模式需更具智慧化與彈性化，檢測智慧化將廣泛應用在各領域以提升生產效率和品質，具備人工智慧之光學檢測與資訊處理已成為提升製造競爭力的重要議題。有鑑於此，基於先前研究結果，本專題計畫研擬研究主題為「結構光數位全像三維造影及其人工智慧資訊處理應用」，擬規劃兩年計畫期程研發整合運用數位全像術暨結構光條紋投影分析之表面造影技術及其相關原理、模擬、實驗、系統實作與實例應用等相關研究工作，並將人工智慧導入研發系統，達成智慧化資訊處理之目標。 第一年度計畫研究重點：結構光數位全像於三維複合表面造影，基於精密量測與微型光學元件量測往往同時具有漫射與鏡射的複合表面組成，運用數位全像技術對具鏡射區域物體進行取像研究，並結合先前所提出高動態範圍結構光條紋投影于強/弱漫射表面之研究成果，再進行複合表面之鏡射與漫射區域之對位融合等資訊處理，完成全幅的三維複合表面造影及其輪廓重建工作；第二年度計畫研究重點：人工智慧資訊處理應用，旨在持續第一年度計畫所建立三維複合表面造影技術及其實驗數據，擬透過人工智慧之深度學習模式進行雜訊消減、三維影像對位與融合應用等資訊處理應用，以提升整體光學檢測系統之準確度與效能。具體而言，本計畫工作之主要價值在於創新整合數位全像與結構光條紋投影技術克服目前複合表面三維立體實物複雜漫射與鏡射的光學檢測限制，以獲得物體完整全幅的三維複合表面的造影與輪廓量測而實現全場三維複合表面造影及其智慧資訊處理應用新模式。
High-speed, non-contact and non-destructive optical three-dimensional (3D) imaging and inspection are the crucial sensing technology and development trend required by domestic and international industries in significant impact on global manufacturing of Industry 4.0. Digital holography is a potential tool of advanced optical 3D imaging and accurate quantitative analysis based on the digital wavefront reconstruction method, which can completely record and reconstruct 3D information of objects. Fringe projection profilometry is full-field and non-contact optical 3D measurement technology that projects digital fringe patterns on objects to produce surface deformation and then extract 3D contours of objects. With the requirements of complex industrial processes and objects, the targets are often combined with diffused and specular composite surfaces. However, a faster and more efficient full-field imaging method required for measuring composite surface objects is still waiting! On the other hand, the proportion of small and diverse production in global smart manufacturing has increased, and the industry needs to be more intelligent and flexible in the face of new production models. Inspection intelligence will be widely used to improve production efficiency and quality. Optical inspection with artificial intelligence has become an important issue to enhance manufacturing competitiveness. With that in mind, based on previous research results, the theme of this special project entitled “Structured light digital holography for three-dimensional imaging and its application on information processing with artificial intelligence” will launch a two-year extensive study on composite surface metrology. In this project, we are planning to perform the research and development for integrating digital holography and fringe projection systems in term of related principles, simulations, experiments, system implementation and applications. Two techniques are integrated to solve key problems such as high dynamic range fringe distortion, fringe shaping, speckle noise, image fusion, to achieve high-quality, full-field 3D composite surface imaging and measurement. Further, we will introduce artificial intelligence into the proposed system to achieve the goal of intelligent optical inspection as well. The first year's project will focus on: structured light digital holography for 3D imaging, according to the precision measurement of micro-optics/electronics components, it often has a composite surface with diffused and specular properties. Digital holography is used to imaging for specular region and fringe analysis is for diffused region. The integrated structured light digital holography can solve the arduous problem for composite surfaces. The second year’s project will focus on the information processing applications by artificial intelligence for the developed 3D imaging system: based on 1st year results, we will plan to apply convolutional neural network (CNN) model for information processing in noise reduction, image registration and fusion of 3D composite surface. It will improve the overall accuracy and efficiency of 3D composite object in optical inspection system. In summary, the main value of this project is the innovative integration of digital holography and fringe projection, related information processing to overcome the optical detection and recording limitations of the complex diffused and specular composite surfaces of 3D objects, and to realize a new model for full-field optical 3D imaging and intelligent information processing applications. |
