A Vessel Extraction Algorithm Based on Point Balance

The article presents a concept of balance among image points based on the correlation of points and then develops a new binarization method for gray scale image, which can be used in vessel extraction of DSA images. A point is in balance only when its resistance counteracts the influence from the neighborhood, while noise breaks down such balance. By restoring the balance of every single point, the image can be a balance system, and most noise can been removed. The balance image turns into a binary image after segmentation by threshold. Experiments show that the algorithm has satisfactory effect on brain DSA images