基于最小总体偏差和区域信息Snake模型的图像分割

Image Segmentation of Snake Model Based on Minimal Total Deviation and Region Information

提出基于最小总体偏差和区域信息Snake模型的图像分割方法。根据最小总体偏差准则,用包含区域信息的变力替换在气球力Snake模型中的恒定力,并应用于图像分割。实验结果表明,该模型与初始轮廓曲线位置无关,能自动分割模糊边界。对于带椒盐噪声的图像,应用该模型也能取得满意的分割效果。

A novel Snake model for image segmentation is given based on minimal total deviation and region information, This model can exchange the constant force in the balloon Snake model into the variable force for incorporating foreground and background information, Experimental results prove that the model is robust to initial contour places and it can automatically segment ambiguous edge images, The segmentation on the sale and impulse noisy image is correct。