Method of self-reinforcement of independent-edges

We proposed the method of self-reinforcement of independent-edges. We use repetitive random searches to obtain various possible independent-edge trajectories, use self-reinforcement of search trajectories to replace reinforcement based on fixed points in relaxation labeling, and use accumulation to replace iteration. Experiments show that our method can solve the problem that the existing methods are very sensitive to noise and local errors, and can reinforce edges well and effectively suppress noise.