摘要
基于曲线演化的水平集算法近来已被广泛应用于医学图像分割中,根据分割医学图像的鲁棒性实时性的要求,提出一种新的基于改进窄带法的图像分割方法INBM(Improved narrow band method)。INBM首先将均匀采样的图像映射到对数极坐标系中,由视网膜空间分辨率机制可知,注视点都在图像兴趣区,由此形成初始轮廓,然后用改进的窄带水平集(Level Set)方法演化曲线得到最终分割结果.改进窄带法是通过降低窄带区域内的水平集函数求解个数,来减少计算时间。实验结果表明,该方法能够快速、准确地得到医学图像的结果。
Curve evolution based Level Set has been widely used in medical image segmentation. However, the high computational cost excludes its use in robust real-time medical image segmentation. This paper presents a novel segmentation method based on Improved Narrow Band Method （INBM）. Firstly, images are transformed from Cartesian space to log-polar coordinate space. The space invariant theory of human vision system guarantees the focus locates only on the interested region. Then an initial contour of the region is formed before we use INBM to get the segmentation results. INBM＇ s adopts reducing level set function solution＇s number of narrow band region to decrease the computation. Our experiment results show a significant decrease in computation time, and thus an effective speed up of medical image segmentation.
出处
《生命科学仪器》
2007年第7期28-33,共6页
Life Science Instruments
基金
国家自然科学基金资助项目(编号:60675015)
