结合KL散度和RSF模型的主动轮廓图像分割方法

Active contour driven by region-scalable fitting and Kullback-Leibler divergence for image segmentation

摘要:

针对主动轮廓模型在进行图像分割时计算复杂度较高的问题，提出一种基于区域的变分水平集主动轮廓模型图像分割方法.新模型将Kullback-Leibler（KL）散度信息加入到RSF（region-scalable fitting）模型中，在新模型的能量项中通过RSF能量项计算区域内某点和该区域"中心"之间的拟合距离来表示目标区域的相似性，同时通过最大化KL能量项使模型能更容易分离图像中的不同灰度区域，进而使图像分割的计算时间显著降低.该模型可以很好地处理图像的模糊边界和图像噪声等问题，并适用于合成图像和实际图像的分割.通过实验结果的对比可以看出，本模型在保证分割精度的前提下，加快了边缘的收敛速度，提高了图像分割的效率.

Abstract:

To overcome the problem of high computational cost of active contour model, a new local region-based active contour model in a variational level set formulation for image segmentation is proposed. An energy function based on the region-scalable fitting (RSF) term and the Kullback-Leibler divergence term is formulated. The existing methods construct the energy function for segmentation through computing the distances among the intraregion points and the ″ center″ fitting this region, representing similarity of object region. An energy term including the disparity measured by Kullback-Leibler divergence between regions to be segmented is added to the energy function of the RSF model in the proposed model. The model can handle blurry boundaries and noise problems. The proposed method is applied to segment synthetic and real images, and the experimental results show that KL-RSF can improve the effectiveness of segmentation while ensuring the accuracy through accelerating the minimization of the energy function.