基于角点的蛇模型在图像分割中的研究

主要应用基于MIC角点检测的snake模型进行图像分割研究。传统的snake算法只考虑曲线本身的特征,并没有充分的利用被拟合图像本身的特征,而此算法首先检测被拟和目标的角点,然后以这些角点为基准进行曲线初始化,最后根据snake算法进行目标拟和;较好地解决了传统snake模型对初始化位置敏感、对有角点和凹陷的图像收敛效果差的缺点。实验表明此算法达到了很好的效果。

Study of snake based on corner on image segmentation

The image segmentation is studied according the snake model based on the comer. The traditional snake model only applies the features of the curve, which did not adequately apply the feature of the image. But in the algorithm, the comers of the detected object is detected based on the MIC operator; the initialized curve is set based on the comers; the object is fitted according on the snake model. The traditional snake model highly depended on the initialized position of the curve and can not get well fitted effort for the image which has the corners and concaves. The proposed algorithm preferably solve the drawbacks. The experiment proves that the algorithm has very good segmentation efforts.