Image Segmentation Method based on Lifting Wavelet and Watershed Arithmetic

The overlapping phenomenon often appears in medical cell images. Traditional watershed based image segmentation methods have the drawback of over-segmentation. A method combining lifting wavelets with watersheds is presented to solve this problem. Lifting wavelet transform is firstly used to create multi-resolution images, then the watershed arithmetic based on distance transform is applied to segment the low-resolution image. At last, the inverse lifting wavelet transform is used to get the high-resolution image. Experimental results show that this arithmetic can successfully separate the cell clusters.