Ultrasound Image Segmentation Based on the Early Vision Model and Watershed Transform

Image segmentation is one of the basic steps for ultrasound image analysis and medical diagnosis. However, due to the intrinsic noisy nature, it remains to be a very difficult problem. A novel segmentation method based on watershed transform and early vision model is presented in this paper. The early vision model improves the antinoise and texture-sensitive capability of our algorithm and watershed transform provides a simple way to get one pixel wide and closed boundaries of ultrasound images. Furthermore, a local marker method is also put forward to solve the over-segmentation problem of watershed transform. The effectiveness of proposed scheme is demonstrated through experimental results on ultrasound images