中文摘要：

针对传统的参数活动轮廓（Snake）模型对初始轮廓曲线非常敏感,提出融合区域信息Snake模型图像分割。该方法将气球力Snake模型中的恒定气球力替换为包含区域信息的变力,以目标和背景两区域平均灰度值距离最大为准则,引导轮廓曲线进化。实验结果表明,该模型对初始轮廓位置不敏感,能实现自动分割;并且对于带噪声图像,该模型也能取得满意分割效果。

英文摘要：

Traditional parametric active contour （Snake） model is very sensitive to initial contour curve position. In light of this, an ima segmentation method based on Snake model incorporated with regional information was proposed in the paper. The model replaces constant h~ loon force in balloon force Snake model with variable force in it the regional information is included, guides counter curves to evolve with criterion of maximizing the distance of average grey values of two regions of the target and the background. Experimental results show that proposed model is insensitive to initial contours position and can segment image automatically. In addition, the testing for segmentation of noise-corrupted image has a satisfied effect as well.