Image segmentation for measurement of particulate contamination in hydraulic fluid

An image segmentation method is proposed to identify the particulate contamination in hydraulic fluid. The procedure of segmentation consists of three steps. First, a partial contour of the interest object is derived by an adaptive edge focusing algorithm. Second, the derived edge points are considered as seedpoints of the threshold surface, then a special surface interpolating technique is employed to generate the threshold surface. In the last step, the image is segmented by the threshold surface. The experimental results show that when background noise and objects are very similar in both shape and gray level, the proposed algorithm can get better results than the other segmentation methods studied. Thus, the algorithm has been used in many kinds of particulate contamination measurement in hydraulic fluid as a very important image preprocessing step.<>