Abstract: In intensity inhomogeneity image segmentation, the Adaptive Distance Preserving Level Set Evolution(ADPLS) algorithm can get a result with high speed, low accuracy and has no relation to initial contour, on the other hand, the Local Binary Fitting(LBF) algorithm can get a result with high accuracy, low speed and its result is sensitive to initial contour. Thus, a novel and adaptive fusing level set method is proposed to make use their advantages respectively, which can automatically adjust the proportion of ADPLS and LBF in the fusing method according to image information. Experiment results show that the comprehensive performance indicators, such as accuracy, speed and stability can be significantly improved in the fusing method.

Key words: image segmentation, level set fusing algorithm, Adaptive Distance Preserving Level Set Evolution(ADPLS) algorithm, Local Binary Fitting(LBF) algorithm

摘要： 在处理不均匀图像时，自适应距离保持水平集演化(ADPLS)算法速度快、不受初始轮廓影响，但精度较低；LBF算法精度高，但速度较慢同时易受初始轮廓影响。针对上述2种算法的优缺点，提出一种新的自适应融合算法。该算法根据图像信息自动调整ADPLS与局部二值拟合算法在融合算法中所占比重，实现不同算法的优势互补。实验结果证明，该融合算法在分割精度、速度及稳定性等方面有明显提高。

关键词: 图像分割, 水平集融合算法, 自适应距离保持水平集演化算法, 局部二值拟合算法

CLC Number: 

• TP391