Workpiece localization with shadow detection and removing

This paper presents a new approach to detect and remove the shadows for workpiece localization, which is with an extensive application in automatic assembly system. However, the shadows of workpiece will badly affect this procedure as the contour of the shadow has the same shape with the workpiece itself in the image. The localization system treats the shadow as a part of the workpiece and make incorrect decision. So removing the shadow in the image before localization is meaningful. Our approach use CAD model to estimate the pose of workpiece, and the contour of object can be drawn in the image. Gray and texture features are used to detect and remove the shadow around the workpiece, and the workpiece is localized without the disturbance of the shadow in image. Experiments have been designed and performed. The experimental results demonstrate the effectiveness of the proposed method.