Analysis of noise image and research for noise image identification

In this paper, we propose a new Non-marking algorithm to calculate the Multi-degree connectivity in Binary Images with noise. In this method we first divide the Image into several grids, by marking the color of the grids. We can filter out the noise from the Image. When scanning the Image, we read the grids instead of pixels. Then, we add the areas which connect each other into the same assemble. By optimizing the process of scan and analysis, this method overcomes the problem of repeating scan and can't process noise Images in the old methods. Besides, the method we proposed can find the connected area in any shape faster.