基于数据场和水平集演化的图像分割

为了充分挖掘图像内不同区域间的隐含关联性，并解决图像分割中自适应阈值选择问题，提出了一种基于数据场和水平集的图像分割方法。利用数据场能够有效地表示图像像素间的相互作用，根据势值能够得到对应的势值等势线，可以根据该等势线的分布情况，采用梯度下降法来得到图像的二值化分割结果。为了得到更为精确的分割边缘，还引入基于拉普拉斯边缘检测函数的水平集演化方法来对二值化结果进行边缘曲线演化。将二值化分割结果与传统的基于数据场的图像分割算法对比结果显示，所提出的算法在分割准确性上表现良好，并且能够使得图像边缘更加精确。实验结果表明，提出的方法能够较好地分割目标，且对噪声图像具有较强的鲁棒性。

Image segmentation based on data field and level set evolution

In order to solve the problem of adaptive threshold selection in image segmentation and explore the implicit correlation between different regions within the image, an image segmentation method based on data field and level set is proposed. Using the data field can effectively represent the interaction between the image pixels. According to the distribution of equipotential lines by the potential value, we can use the gradient descent method to obtain the binary segmentation of the image. Particularly, in order to obtain the more accurate segmentation edge, a level set evolution method based on the Laplaction edge detection function is introduced to do the edge curve evolution from the binarization result. The binary segmentation results are compared with the traditional image segmentation algorithm based on the data field theory. The proposed algorithm has a good performance in segmentation accuracy and can make the image edge more accurate. Experiments show that the proposed algorithm can better segment the target and has strong robustness to the noise image.