由於電腦科技日新月異，電腦能提供的運算功能、記憶體以及各種軟體的應用正以極快的速度增加，因此越來越多的產業將其生產自動化。其中機器覺扮演了很重要的角色，而影像對位更是其中一項重要技術。 本論文將建立一套具視覺導引對位的自動化系統，藉以同時達到高品質低成本且挑戰性的目標。此系統涵蓋機械定位與影像辨識兩大子系統。機械定位系統為二軸伺服馬達所架構出的X-Y Table，經過路徑規劃補償後由伺服馬達控制卡來驅動馬達進行運動定位。影像辨識系統架構則是透過CCD(Charge-Coupled Device,電荷耦合元件)，配合照明技術及影像擷取卡擷取影像。影像處理的部分則使用影像正交投影法(Projection Profile)進行料盒位置的粗略搜尋，接著以正規化相關匹配法(Normalized Cross Correlation,NCC)進行SMD料帶盒上模影像區域細緻搜尋比對(Pattern Match)，最後輸出修正量到機械定系統進行修正。經實驗證明，本系統可達到快速且準確對位的預期目標。 Due to the fast development of computer technologies, faster operation speed, larger memory space, and multifunctional software can be easily acquired. It is a trend that in industry that machine vision will play more and more important role in industrial automation. One of the applications in industrial automation is the alignment utilizing computer vision techniques. In this thesis, a model equipped with vision-guiding automation is built to accomplish the challenging goals of both high quality and low cost outputs. The proposed system is composed of two sub-systems including mechanical alignment and vision identification. In mechanical alignment sub-system, 2-axis servomotors are employed to actuate the X-Y Table. The table will be activated by precise route calculation and compensation. The vision sub-system will capture the image of each plastic embossed cavity of SMD (Surface Mounting Device) packaging carrier via the CCD camera and the illumination source. Then get the initial position by using the image projection profile method. Next, the task of fine search will be accomplished by using the method of NCC (Normalized Cross Correlation) to match the pattern of every target cavity with the original. The compensation value will be obtained and inputted to the mechanical system for final route adjustment. Experimental results demonstrate that the proposed automated vision system can achieve very fast and precise solution when applying to the mechanical alignment problem.
