关键词: 图像分割, 降雪模型, 随机游走, 高斯核函数

Abstract: In order to improve accuracy of image segmentation, reduce the effect of noise on the cutting edge of image segmentation as much as possible, a new image segmentation method based on the model of the snowfall is proposed. Snowfall model and snow surface effect are analyzed in detail, the snow model is applied to image segmentation with strong adaptability, and it mixs the traditional random walk image segmentation algorithm with adaptive snow model characteristics, generates a new algorithm, makes performance simulation using virtual and real images algorithm, the results show that the image segmentation performance is better than the common NCut and the traditional random walk algorithm for image segmentation, and it has certain research value.

Key words: image segmentation, snowfall model, random walk, Gauss kernel function