Abstract: Aiming at the wear debris automatic identifying for visual on-line ferrograph sensor, digital image capturing system is built and the ferrograph image processing methods are investigated. The image preprocessing is firstly investigated. Original image graying processing is performed with the 2 color models of YUV and RGB, and image sharpening with the differential moulds is followed. Furthermore, the wear debris segmentation of the preprocessed images is investigated with automatic threshold method and background subtracting method. Finally, a statistics description suitable for on-line ferrograph image is put forward. As main results, the luminance component of the YUV color model shows better effects in image graying processing over the RGB color model. The differential method with suitable model performs well in debris boundary sharpening without loss of image smoothness. The background subtracting method shows inconvenience in wear debris segmentation processing because of its artificial threshold selection, while the automatic threshold method performs well with the self-adapting of the threshold value. The index of particle coverage area performs well as an index of statistical wear debris content in the well segmented ferrograph images.

Key words: Image processing, On-line ferrograph, Wear debris