Auto-threshold Confirming Segmentation for Wear Particles in Ferrographic Image

In machine condition monitoring, wear particles formed in rubbing are a source of valuable information on the wear mechanism and severity, while ferrographic image is an information carrier of wear particles. For the significance of image segmentation for wear particle feature extraction and recognition, defects of some traditional methods that can convert color image into binary are introduced and analyzed in this paper. After this, based on image object area and its difference, an auto-threshold confirming segmentation algorithm for ferrographic image is presented. Experimental results show that this algorithm can segment wear particles accurately and automatically. Especially, it is efficient for bright object with black background in micro image with reflex but no transmission light.