结合边界和区域的水平集超声图像分割算法

针对距离正则化的水平集演化（DRLSE）模型难以处理弱边缘图像、初始轮廓敏感以及曲线演化方向单一等问题，提出一种结合边缘和区域信息的变分水平集超声图像分割模型。该模型采用改进的四阶偏微分方程进行滤波，实现在去除噪声的同时保护图像边缘信息；构造了自适应加权系数，实现曲线自适应地向内或者向外演化；引入CV模型的外部能量项，将图像的边缘信息和区域信息相结合，提高了全局分割能力。实验结果表明：该方法在分割超声图像时，具有演化结果稳定，边缘定位准确的特点，可以较好地提取超声图像中的目标。

Ultrasound image segmentation method based on Level set combined with boundary and region

As the distance regularized level set evolution model is difficult to deal with weak boundary image and sensitive to the location of initial curve and has a single direction of curve evolution, a ultrasound image segmentatian madel based on varitional level set combined with boundary and region is proposed in this paper. This model adopts improved fourth - order partial differential equations filter which can remove the noise while preserving edge information. It constructs an adaptive weighting coefficients to realize the curve adaptive evolution inwards or outwards. It introduces external energy term of CV model, combing the boundary information and region information of image, which can improve the ability of global segmentation. Experimental results show that the method has the stable evolution results and is accurate to locate the position of the boundary in the segmentation of ultrasound images. Also, it can extract the object ideally from ultrasonic image.