透過您的圖書館登入
IP:18.218.61.16
  • 學位論文

應用機器視覺檢驗油封之尺寸及表面瑕疵

Dimensional and Surface Defects Inspection of Oil Seal Using Machine Vision System

指導教授 : 江行全
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


本研究利用視覺系統檢驗油封之內外徑尺寸及表面瑕疵。 因此研究方法 主要分成兩部份, 第一部分為內外徑尺寸檢驗,首先計算影像的分劃值 , 利用分劃值與邊緣追蹤法則可將油封內外徑的邊緣點位置依序取出, 邊緣點位置取出後利用高斯平滑法修正,最後將修正後的邊緣點位置代入 面積最小誤差公式即可取得油封內外徑。要進行第二部分油封之表面瑕疵 檢驗之前, 首先必須用第一部分的演算法計算圓心。 利用圓心可取得待 檢驗區域內的圓周點上之像素點灰階值, 將像素點灰階值以「改良式之 移動全距與個別值管制圖法」判斷檢測油封之灰階值是否為異常。 然而 視覺系統極易受到環境如打光設備、油漬、灰塵等影響,使得單由灰階異 常判斷, 實在很難達到檢測需求,故本研究再進一步判斷臨近像素點之 異常情形; 即檢查待檢驗區域內所有的圓周點後, 將灰階度異常的像素 點資料存入矩陣中,透過分析矩陣資料, 作為判斷待檢區域是為瑕疵區 ,如此將可降低光源、灰塵等非瑕疵因素的影響,以提昇實務檢測之適用 性。本研究成果可運用於油封的自動檢驗,藉以減少人工抽檢作業。 可 對進行產品全檢, 實際反應產品品質狀況,進而提昇品質之穩定度,使 得製程具備更多的彈性及生產效率。

並列摘要


In this research the Diemnsions and surface defects of the oil seal are inspected by using machine vision system. The inspection can be divided into two part: Dimensions and Surface defects inspection. The dimensions of the oil seal are inner and outer diameter. There are four steps to get the dimensions. First, get the threshold value of the image. Second, using the threshold value and boundary tracking method to extract the boundary point position. Third, let boundary point position fit the real positio n by using Gaussian smoothing method. Fourth, use the boundary points and the equation of Minimize Errors in Area method to calculate the dimensions. In this surface defects inspection, the center position must be calculated first. With the center, we can get the grey levels of the circumferential pixels on the inspection regions. The inspection result is decided by applying improved moving range and individual observation control chart method to checking the grey levels. If the defect inspection is judged only by the grey level data, the inspection result will be widely influenced by noises, such as light reflection, oil slick, dust, and so on. In this research we add a judge procedure to store the neightborhood pixels inspection results in a feature array. The surface defect inspectoin is decided by analyzing the feature array, and then the influence of noises will be reduced. The result of this research can be applied to the oil-seal-inspect automation, and a 100% inspection on the products can be performed. That is, the result of the research can make the manufacture procedure more flexibility and more efficiency.

延伸閱讀