本研究利用影像處理技術對於物件的規格進行視覺量測,使用高畫素之精密量測解決物件不良的問題。本研究對於汽車零組件進行數位化處理,利用影像物件之幾何特徵並加以篩選,同時符合既定條件之特性,以程式化與自動化處理影像物件為目標。 在影像擷取系統中,本研究運用影像處理結合NI MAX系統擷取物件之影像畫面參數,配合LabVIEW軟體量測物件。硬體部分則是利用高解析度的CMOS (Complementary Metal-Oxide-Semiconductor)攝影機配合影像擷取卡運用Giga-E介面擷取影像。另外,光源系統利用不同類型的光源設備配合,調整出最適合物件的光源參數。本文量測的對象是汽車零組件,透過CMOS拍攝出的物件進行匹配程式內參數,針對影像進行物件萃取、篩選、量測、報告輸出等程式化之結構,將量測結果顯示在電腦上。實驗結果之準確度:(1)橢圓直徑短邊為99.273%;(2)橢圓直徑長邊為99.027%;(3)內圓直徑為95.774%;(4)外圓直徑為91.114%。透過每次所測得的數據可以得知該物件的規格,以便利往後資料庫建立與系統比對,節省檢測所需的時間,令影像處理可以在工程應用予以實現。
This research measures the object standard by using image processing technique to conduct vision measurement, and solves the problem of unqualified object by high pixel precision measurement. We use the auto parts to conduct digital processing, and use object image’s geometry characteristic to filter the result in this research. In addition, to comply with the characteristic of condition, we consider programmable and automatic processing of image objects as our goal. In image capturing system, this research uses image processing to combine NI MAX image capturing system of image parameter to match LabVIEW software measuring object. The hardware uses high resolution CMOS (Complementary Metal-Oxide-Semiconductor) camera to combine frame grabber with Giga-E to capture images. Besides, lighting system is combined with different types of lightening equipment to adjust the most appropriate object lightening parameter. The measurement of this research is the auto parts, through the image that CMOS was taken to conduct and match the parameter of the programs. Then, it was directed against the image conducted to extract, filter, measure, and report the structure of programming on the computer. The experiment results of the accuracy showing that: (1) short egde of the oval diameter is 99.273%, (2) long egde of the oval diameter is 99.027%, (3) the inner diameter is 95.774% and (4) outer diameter is 91.114%. Through the database, we can know the standard of the object every time and can compare with the database system, so as to save the measuring time. And the proposed system can realize the auto image processing in engineering applications.