基于改进的水平集局部二值拟合模型车牌分割算法

摘要

This paper proposes an improved model of contour evolution LBF energy function, which combines the CV model energy term to accelerate evolution rate, and combines local mean LBF model information while introducing a global image of the local variance and variance information. Experimental results show that this method can provide accurate smooth closed boundary,and that precision can reach sub-pixel level,and that the recognition accuracy rate is high.%本文提出了改进的LBF模型的轮廓演化能量函数，它结合了CV模型的能量项加快了演化速度，并在结合LBF模型中的局部均值信息的同时，引入图像的局部方差和全局方差信息。实验结果证明该方法能够提供准确的光滑的闭合的边界，精度可以达到亚像素级，系统的识别准确率高。