Watershed Segmentation Based on Inertial Product Energy

An image segmentation method is presented based on the inertial product energy to construct a good watershed structure for gradient level image and improve watershed segmentation. The inertial product energy can weaken the noises and tiny edges in gradient level image. Then a non-linear processing classification and filtering based on fusion and adaptive are employed to reduce the influence of a lot of local minimum pixels which are less sensible in the inertial product energy image. Thus the inertial product energy image has a good watershed structure. Experimental results show that the proposed method significantly decrease the over-segmentation and provides a reliable basis for further processing.