透過您的圖書館登入
IP:18.221.129.19
  • 學位論文

植基於立體視覺之IC腳位尺寸量測

IC Lead Inspection Based on Stereo Vision

指導教授 : 陳金聖

摘要


為了檢測IC的外觀品質,本研究應用立體視覺技術發展一套高檢測精確度及高檢測速度的SOP IC尺寸量測系統,檢測項目包含腳間距(pitch)、腳跨距(spread)、腳排彎(sweep)、腳高(stand off)以及腳共平面(coplanarity)。在傳統IC尺寸量測中,腳位大部分都由2D影像所獲得的資訊經由連結體分析(blob analysis)配合重心法(mass of center)來做運算處理,而本研究改以立體視覺的概念,配合次像素之運算達到高速高精度之要求。量測之技術包含:1) 2D至3D之影像校正,2) IC尺寸量測。首先,於影像校正部分,經由Harris特徵點擷取法及本論文所提的強健性對應點搜尋法獲得2D影像對之特徵,然後配合具懲罰函數的映射函數法來提升校正精度及速度。影像校正後,用上述同樣方式獲得待量測之IC於2D影像對之特徵,然後利用影像校正步驟所獲得之2D至3D影像轉換矩陣,將影像對之資訊轉換為3D資訊,再配合幾何特性量測出腳間距(pitch)、腳跨距(spread)、腳排彎(sweep)以及腳高(stand off)。在腳共平面度中,為了提昇檢測速度,我們以包絡線法取代傳統自然落下法來進行量測。 從整體系統來做評估,在檢測IC時,希望其重現性及檢測精確度皆能符合業界的要求並能達到高檢測速度以降低檢測時間的成本,本論文最後將透過實驗驗證本系統具有高的實用價值。

並列摘要


This thesis proposes a high speed and high accuracy inspection system of small outline package (SOP) IC based on stereo vision. This system can efficiently measure the following specifications of an inspected IC: lead pitch, lead spread, lead sweep, lead stand off, and lead coplanarity. In the presented literatures, the techniques of blob analysis and mass of center based on 2D images are employed to inspect the leads of a SOP IC. This thesis proposes a novel IC lead inspection system based on stereo vision and the subpixel algorithm to fit the high speed and high accuracy requirement. This inspection system includes two stages: 1) image calibration from 2D to 3D image, 2) IC lead inspection. In the first stage, the algorithms of Harris corner detector and robust corresponding point are firstly applied to capture the features in the pair of 2D images. Then, the mapping function with penalty is used to improve the speed and accuracy of image calibration from 2D to 3D image. In the second stage, we use the same algorithms to capture the features of the inspected IC in the pair of 2D images. The 3D information will be calculated by the transform matrix obtained in the first stage. Then the specifications of lead pitch, lead spread, lead sweep, and lead stand off can be easily measured by the IC’s geometry and the corresponding 3D information. In the other hand, we used the heuristic algorithm to replace the traditional algorithm to quickly measure the coplanarity. The proposed inspection system has been evaluated by many experiments to verify the repeatability and accuracy.

參考文獻


[26] 曾凱銘,半導體封裝製程之銀膠厚度與晶片傾斜度檢測,碩士論文,私立中原大學機械工程研究所,桃園,2006。
[25] 蔡佑男,齒模三維輪廓量測,碩士論文,私立中原大學機械工程研究所,桃園,2002。
[24] 林高輝,映射函數三維空間影像量測技術研究,博士論文,私立中原大學機械工程研究所,桃園,2002。
[20] 戴呈霖,雙照相機座標量測系統,碩士論文,國立成功大學機械工程研究所,台南,2004。
[19] 陳柏安,利用電腦視覺作自走車之障礙物定位與環境掃描,碩士論文,國立成功大學工程科學系,台南,2003。

被引用紀錄


蔡治平(2013)。影像視覺之光譜照度分析與物件特徵辨識〔碩士論文,國立虎尾科技大學〕。華藝線上圖書館。https://doi.org/10.6827/NFU.2013.00061

延伸閱讀