针对距离规则水平集演化(DRLSE)模型存在易陷入虚假边界、对噪声敏感、收敛速度慢以及容易从弱边缘处泄露等不稳定问题,提出了面积项能量加强的水平集演化函数对水平集方法进行改进.首先提出了一个自适应边缘指示函数,其根据图像信息来调整函数参数,从而控制演化速度以及对噪声敏感度,使水平集演化更加快速稳定.同时结合区域生长方法,将图像处理成一个二值矩阵,并据此矩阵增加一加强项,使得面积项能量得到加强,令水平集函数随着距离目标远近而自动调整能量大小,降低计算成本,有效解决对噪声敏感、易陷入虚假边界等问题.为验证模型的有效性,采用多张实图进行分割实验并与 DRLSE 等模型进行对比,实验结果表明,提出的模型能有效解决存在问题,有更高的计算效率和准确率.%The distance regularized level set evolution (DRLSE) model has lots of shortcomings when applied in image segmentation. It is easy to stuck around false boundaries, is sensitive to noise, have slow convergence speed and may not detect weak edges. To solve these problems, this paper present a level set function whose area term is enhanced. Firstly, an adaptive edge indicator function is proposed, which can adjust some parameters automatically based on image information, to control evolution speed and the sensitivity to noise. Furthermore, combining the model with region growing method and processing the image into a binary matrix, a reinforcement term is added based on the binary matrix to intensify energy of the area term. So the energy functional could be automatically adjusted according to the distance between evolution curves and targets. This reduces the computational cost, makes the model insensitive to noise and isn't easy to fall into false boundaries. Experiments on some synthetic and real images prove that the proposed model is not only robust, but also achieves higher segmentation accuracy and efficiency.
展开▼
