The images of some marine microalgae were segmented by a C-V active contour model of Chan and Vese whose improved model was described in this paper. When the curvature of the main boundaries of microalga images was changed greatly, namely the main boundaries were of " steep" , it is very difficult to obtain the boundary of the microalgae images by C-V active contour model directly. In the improved C-V active contour model, the rough initial boundaries were drawn through the man-machine interactive pattern at the zero level set, the symbolic function was introduced in the initial zero level set to define internal and external energy, and appropriate parameters were adjusted to execute the process of the evolution of the image boundary. The comparison between the two models for the typical marine microalga image boundaries revealed that for the microalga images with "steep" boundary it is difficult to obtain their boundaries or their boundaries slowly by C-V active contour model. The boundaries of the microalga images were quickly obtained as well as a lot of information of the boundaries when the improved C-V active contour model was used for the microalga image with "steep" boundary. The results showed die effectiveness of die improved C-V active contour model, indicating diat this provides a new skill for the segmentation of microalga images.%采用Chan和Vese的C-V主动轮廓模型以及本文中改进的C-V主动轮廓模型对几类典型的海洋微藻图像进行了分割.当微藻图像的主要边界曲率变化较大,即主边界“陡峭”时,直接使用C-V主动轮廓模型难以获得微藻图像的边界.在改进的C-V主动轮廓模型中,通过人机交互绘制粗略的初始边界,并将其设定为初始零水平集,将符号函数引入到初始水平集中定义内外能量,再通过适当的参数调整进行图像边界的演化.将采用两种模型算法获取典型的海洋微藻图像边界的过程进行对比可知,对于带“陡峭”边界的微藻图像,采用C-V主动轮廓模型难以获得或以较慢速度获得图像边界,而采用改进的C-V主动轮廓模型不仅图像边界获取速度快,而且边界信息量大.实验结果验证了改进的C-V主动轮廓模型算法的有效性,为微藻图像的分割提供了新的技术手段.
展开▼
