融合统计范数度量的局部区域图像分割

Local region image segmentation by fusion of statistical norm metrics

主动轮廓模型(Active Contour Model,ACM)已成为最常用的图像分割工具之一,但在处理灰度不均匀的图像时,现有的算法比较耗时而且会导致分割精度急剧下降。为此,本文结合局部图像信息,提出了一种统计范数用于图像分割。首先,利用新的偏置场模型来建模图像,将图像的灰度不均匀性分解为观测图像的一个分量。与传统的乘性偏置场相比,加性偏置场模块使能量泛函能够从新的维度提取图像的纹理信息。其次,采用局部信息融合的策略,计算轮廓内外的特征拟合图。最后,利用统计范数刻画特征拟合图与原始特征图的相似度。因此,新设计的能量泛函通过利用分层局部特征、全局空间一致性和多尺度抽象表示等来处理具有复杂背景的图像。实验结果表明:分割非同质医学图像,本文模型仅仅需要迭代50次,其他模型均在100次以上;算法运行时间仅仅需要8 s,但其余模型远远高于8 s。同时,采用客观评价指标评价所提出的算法:DC指标的平均值是0.9851,FP指标的平均值是0.0052,JCS指标的平均值是0.9706,P指标的平均值是0.9947,TP指标的平均值是0.9757。本文模型能够提取更多的纹理结构信息,并且对灰度不均匀和初始轮廓具有较强的鲁棒性。

representation.The experimental results show that for segmenting non-homogeneous medical images,the model in this paper requires only 50 iterations,while the other models are all over 100;the algorithm takes only 8 seconds to run,but the rest of the models are much higher than 8 seconds.At the same time,the proposed algorithm was evaluated using objective evaluation indicators:the average value of the DC indicator is 0.9851,the average value of the FP indicator is 0.0052,the average value of the JCS indicator is 0.9706,the average value of the P indicator is 0.9947,and the average value of the TP indicator is 0.9757.The model in this paper is able to extract more information about the texture structure and is robust to gray scale inhomogeneity and initial contours.