抄録
This study treats the defect detection of the gear parts with image processing. The gear was put on the turn table and its side part was lightened by the LED light equipped on the USB camera. The obtained image was converted to the binalization one. Here, we focused on the fact that the reflection points appeared at the top of the gear. When the gear was rotated on the turn table, the black and white oblong domains on the banalization image were alternately appeared due to the relation between the gear convex-concave shapes. It was divided into three domains to find the reflection points. As a result, it became easy to find the edge line of the gear because the reflected points were always located at the extension of the edge line.
