BGA基板視覺檢測

Abstract

在過去，傳統檢測是以人眼來檢查產品是否有瑕疵，但費時、不可靠且昂貴。 提供高速度、高品質和低成本的視覺檢測系統，在今天的工業界已經廣泛地被採用。 在論文中，我們提出一個視覺檢查方法來檢查BGA基板的表面瑕疵，其檢查方法如下:

首先， 我們需要選擇無缺陷影像並用它來做樣板影像。然後，我們將樣板影像切割成Bond finger區和Power/Ground ring兩區。把每一個Bond finger都分成三個檢測區:重要檢測區,次要檢測區,和短路檢測區。並儲存從樣板影像提取的重要參數。在樣板影像建立好以後，輸入待測影像並以樣板影像來做影像校正。在檢查過程中，把這個待測影像和樣板影像比對後，決定待測影像中物件是否有缺陷。最後實驗的結果表示我們的方法能檢查出在BGA基板檢測標準中定義的所有缺陷。

In the early days, traditional inspection was done by using eyes of human operators, and therefore time consuming ,unreliable and expensive.Visual inspection system provides high speed, high quality and low labor cost, and has been widely adopted in today's industry. In the thesis, we propose a visual inspection method to inspect the surface of BGA substrates. The inspection method we developed is described as follows.

First of all, we need to select a defect free image and use it to define a template image. Then we segment the template image into the bond finger regions and the power/ground ring region. Each single bond finger is divided into three regions : the major inspection region , the minor inspection region, and the short circuit inspection region. The parameters extracted from the template image are stored. After the template image builded, the inspected image is captured, and aligned with the template image. In the inspection process, the inspected image is inspected and compared to the template image to determine whether the object in the inspected image is defective or not. The outcome of the final experiment showed that our method could detect all defects defined in the BGA substrate criteria.