基于小波变换的权重自适应图像分割模型

针对灰度不均匀且含噪声图像的分割问题,提出了全局和局部灰度信息的权重参 数自适应水平集分割模型。首先,利用图像的全局和局部灰度信息构造全局能量项和局部能量 项;然后,利用小波变换和小波阈值去噪方法,构造对噪声不敏感的边缘信息刻画矩阵,定义包含 图像边缘信息的自适应权重系数矩阵;最后,利用定义的权重系数矩阵组合全局和局部能量项, 得到分割模型的能量泛函。使用变分法得到了水平集函数演化方程,利用有限差分法实现数值 求解。实验结果表明,该模型兼有 Chan-Vese 模型和 Local Binary Fitting 模型的优点,能够有效 地分割灰度不均匀含噪图像,并对活动轮廓曲线的初始位置和初始形状具有很强的鲁棒性。     

引入分数阶微分的局部高斯分布拟合能量模型

局部高斯分布拟合能量(LGDF)模型缺乏全局信息,对初始轮廓曲线选取较敏感,特别在分割弱边缘和弱纹理区域图像时,容易陷入局部极值,对噪声的鲁棒性不好.针对上述问题,文中提出引入分数阶微分的LGDF模型.在LGDF模型中引入全局的Grümwald-Letnikov(G-L)分数阶梯度拟合项,增强弱边缘和弱纹理区域的梯度信息,提高对初始轮廓曲线和噪声的鲁棒性.采用自适应权重函数确定全局项和局部项的系数,提高对灰度不均匀图像的分割效率和分割精度.根据图像的梯度模值、信息熵和对比度构建自适应分数阶阶次的函数,提高分割效率.理论分析和实验均表明,文中模型可以用于灰度不均匀、弱纹理、弱边缘图像的分割.合成图像和真实图像的实验表明文中模型可以提高图像的分割精度和效率.     

基于改进活动轮廓模型的图像分割

针对图像分割中的灰度不均匀和轮廓初始化问题,提出一种基于区域的活动轮廓模型。将图像的全局信息和局部信息作为能量项驱动活动轮廓向目标边缘演化,以有效分割灰度不均匀图像,为保证图像分割的速度和精度,在能量方程中加入长度项和惩罚项,并采用梯度下降法得到该模型的最小化能量方程。实验结果表明,和局部二值拟合模型、局部图像拟合模型相比,该模型能分割灰度不均匀的图像,对初始轮廓曲线大小和位置更不敏感,且分割图像所需的迭代次数、迭代时间更少。     

用于灰度不均图像分割的自适应灰度拟合模型

针对灰度不均图像的分割问题,提出了一个结合全局信息的局部区域自适应灰度拟合模型。首先,分别利用图像的局部和全局信息构造了局部拟合项和全局拟合项;其次,利用像素点邻域内灰度的极差反映该点邻域内灰度的偏差程度,并以此定义了一个自适应权值函数;最后,利用定义的权值函数为局部项和全局项自适应赋权值,得到所提模型的能量泛函,并使用变分法推导出模型的水平集函数迭代方程。数值实现采用有限差分法。实验结果表明,与区域可变灰度拟合（RSF）模型和局部和全局灰度拟合（LGIF）模型相比,所提模型不仅能够稳定、准确地分割多种灰度不均图像,而且对演化曲线初始轮廓的位置、大小和形状具有更强的鲁棒性。     

一种区域自适应主动轮廓模型的图像分割方法

为了有效地分割灰度不均匀图像,提出了一种区域自适应主动轮廓模型,在该模型中,定义了一个包含全局能量项和局部能量项的能量泛函。在算法的初期,全局能量项占主导地位,它具有收敛速度快、对初始轮廓不敏感的优点。在算法的后期,局部能量项占主导地位,它具有定位精度高的优点。理论分析和实验结果表明,该模型具有收敛速度快、分割精度高、对初始轮廓不敏感等优点。     

基于演化曲线自适应的图像分割模型的研究

利用图像梯度和几何曲率等信息可以准确定位分割图像的边缘。基于此,在对图像分割典型变分模型有效性及所存在问题分析和讨论的基础上,提出了一种演化曲线自适应驱动的图像分割水平集模型。模型通过调整演化曲线长度项和面积项的权重函数,使演化曲线能够根据图像当前的状态自适应地调整演化幅度和方向,不仅提高了图像分割的准确度,还大大缩减了图像分割时间;模型在利用图像局部区域信息的同时,也利用全局化的正则函数来兼顾模型能量泛函的全局性,使模型有了对异质区域边界的捕捉能力。经试验验证,文章所提出的新模型有效可靠。     

结合全局和双核局部拟合的活动轮廓分割模型

针对可缩放区域拟合（RSF）模型对初始轮廓敏感的缺点,提出了一种结合全局和局部图像信息的变分水平集活动轮廓模型。该模型设计了一个灰度域上的核函数,将其与RSF模型空域核的线性组合作为局部能量项,弥补了采样权值仅与空间距离有关的缺陷,提高了分割精度;构造了带有自适应全局指示函数的面积项,作为全局拟合力,提高了模型的收敛速度且避免陷入局部极小值;使用了高斯滤波方法,规则水平集函数,使其保持光滑,并避免了复杂的重新初始化过程。实验结果表明,该模型初始化灵活,对灰度不均匀图像有很好的分割效果     

结合局部熵能量泛函与非凸正则项的图像分割

为了克服灰度不均匀对图像分割的影响，结合CV模型的全局能量项和LBF模型的局部能量项，引入图像局部熵信息和非凸正则项，构造新的能量泛函，提出了结合局部熵的局部能量泛函与非凸正则项的图像分割算法。该算法首先采用CV模型中的全局能量泛函得到图像的大致演化轮廓；通过构建具有局部熵信息的局部能量泛函，实现对图像的精确分割。然后，利用非凸正则项作为图像演化过程中零水平集逼近目标的又一驱动力驱动曲线演化和边缘保护。该算法利用变分水平集方法将这一新构建的能量泛函进行最小化，通过迭代更新水平集函数，完成曲线演化。最后，对比实验表明，所提出的算法可以高效、准确地分割灰度不均匀图像。     

一种图像分割模型的自适应参数选择方法

针对灰度分布不均匀的图像特征,提出一种基于局部和全局高斯分布拟合能量的自适应权重参数选择方法。基于图像的局部和全局区域信息,以高斯分布作为拟合函数建立能量泛函。基于水平集方法,随着活动轮廓的演化,局部和全局区域信息在能量泛函中的权重会相应地变化,有利于提高图像分割的质量和效率。数值实验验证了该方法的有效性。     

基于核函数与局部信息的凸优化分割模型*

本文针对CV模型不能准确分割非同质和高噪声的图像,且计算效率比较低的特点,做出如下改进：对于区域中的每一点,本文利用该点所在区域的平均灰度值和其邻域内其它点的灰度值的核函数度量定义局部能量项,然后对图像域上所有点的局部能量进行积分定义全局能量项,由于局部信息和核函数的引入使得区域均值的更新具有较强的抗噪能力,提高分割鲁棒性;然后将该模型转化为全局凸分割模型,同时引入边界边缘检测函数加权的总变差范数(total variation,TV)更加准确地获取目标的边界位置,以提高模型的分割精度;最后,使用split Bregman迭代进行数值求解。实验结果表明,该模型能够有效的分割非同质和高噪声图像,与CV模型,RSF模型,DRLSE模型相比,在运行速度和分割精度上有了很大的提升。     

结合全局信息的局部图像灰度拟合模型

针对局部图像拟合（LIF）模型对初始轮廓大小、形状和位置敏感的问题,提出一个结合全局信息的局部图像灰度拟合模型。首先,构造了一个基于全局图像信息的全局项;其次,将该全局项与LIF模型中的局部项线性组合;最后,得到了一个以偏微分方程形式存在的图像分割模型。数值实现采用有限差分法,同时采用高斯滤波器正则化水平集函数以确保水平集函数的光滑作用。在分割实验中,当选取不同的初始轮廓时,该模型均能得到正确的分割结果,且分割时间仅为LIF模型的20%到50%。实验结果表明,所提模型既对演化曲线初始轮廓的大小、形状和位置都不敏感,又能够有效地分割灰度不均图像,且分割速度较快。此外,在无初始轮廓的情形下,该模型能快速分割一些真实图像和人造图像。     

基于核特征距离的局部活动轮廓模型

提出了一种基于核特征距离局部活动轮廓分割模型。在模型中使用核特征距离来构造局部拟合能量,从而可以获取精确的局部图像特征,可以分割存在灰度不均匀的图像。并通过引入水平集规范项以避免水平集演化的重新初始化,提高了分割的效率。实验结果表明,本模型可以很好地克服灰度不均匀性,同时在分割精度上有了较大的提升,特别是分割速度比LBF模型快1.3～1.5倍。     

结合全局与局部信息活动轮廓的非同质图像分割

目的 通过对现有基于区域的活动轮廓模型能量泛函的Euler-Lagrange方程进行变形,建立其与K-means方法的等价关系,提出一种新的基于K-means活动轮廓模型,该模型能有效分割灰度非同质图像。方法 结合图像全局和局部信息,根据交互熵的特性,提出新的局部自适应权重,它根据像素点所在邻域的局部统计信息自适应地确定各个像素点的分割阈值,排除灰度非同质分割目标的影响。结果 采用Jaccard相似系数-JS(Jaccard similarity)和Dice相似系数-DSC(Dice similarity coefficient)两个指标对自然以及合成图像的分割结果进行定量分析,与传统及最新经典的活动轮廓模型相比,新模型JS和DSC的值最接近1,且迭代次数不多于50次。提出的模型具有较高的计算效率和准确率。结论 通过大量实验发现,新模型结合图像全局和局部信息,利用交互熵特性得到自适应权重,对初始曲线位置具有稳定性,且对灰度非同质图像具有较好地分割效果。本文算法主要适用于分割含有噪声及灰度非同质的医学图像,而且分割结果对初始轮廓具有鲁棒性。     

Active contour model for simultaneous MR image segmentation and denoising

In this paper, a new region-based active contour model is proposed for magnetic resonance image segmentation and denoising based on the global minimization framework and level set evolution. A new region fitting energy based on Nadaraya–Watson estimator and local image information is defined to enforce the curve evolution. By this improved region fitting term, the images with noise and intensity un-uniformity can be segmented and denoised. Inspired by the Perona–Malik diffusion equation, an edge-preserving regularization term is defined through the duality formulation to penalize the length of region boundaries. By this new regularization term, the edge information is utilized to improve the contour?s ability of capturing the edge and remaining smooth during the evolution. The energy functional of the proposed model is minimized by an efficient dual algorithm avoiding the inefficiency of the gradient descent method. Experiments on medical images demonstrate the proposed model provides a hybrid way to perform image segmentation and image denoising simultaneously.     

基于凸优化的自适应CV模型*

针对CV模型分割精度不高、分割速度缓慢和易陷于局部最优等缺点,提出了一种新的基于凸优化的自适应CV模型。首先,引入了自适应权重项,对拟合中心的计算采用加权平均,提高了拟合中心计算的准确性;然后,在模型中加入了凸优化技术,以获取模型的全局最优解;最后,采用了Split Bregman方法进行快速求解,有效地提高了分割效率。实验结果表明,基于凸优化的自适应CV模型有效地提高了分割精度和效率,对初始化也具有较好的鲁棒性。     

Nonlocal variational image segmentation models on graphs using the Split Bregman

Variational functionals such as Mumford-Shah and Chan-Vese methods have a major impact on various areas of image processing. After over 10 years of investigation, they are still in widespread use today. These formulations optimize contours by evolution through gradient descent, which is known for its overdependence on initialization and the tendency to produce undesirable local minima. In this paper, we propose an image segmentation model in a variational nonlocal means framework based on a weighted graph. The advantages of this model are twofold. First, the convexity global minimum (optimum) information is taken into account to achieve better segmentation results. Second, the proposed global convex energy functionals combine nonlocal regularization and local intensity fitting terms. The nonlocal total variational regularization term based on the graph is able to preserve the detailed structure of target objects. At the same time, the modified local binary fitting term introduced in the model as the local fitting term can efficiently deal with intensity inhomogeneity in images. Finally, we apply the Split Bregman method to minimize the proposed energy functional efficiently. The proposed model has been applied to segmentation of real medical and remote sensing images. Compared with other methods, the proposed model is superior in terms of both accuracy and efficient.     

An efficient local Chan-Vese model for image segmentation

In this paper, a new local Chan-Vese (LCV) model is proposed for image segmentation, which is built based on the techniques of curve evolution, local statistical function and level set method. The energy functional for the proposed model consists of three terms, i.e., global term, local term and regularization term. By incorporating the local image information into the proposed model, the images with intensity inhomogeneity can be efficiently segmented. In addition, the time-consuming re-initialization step widely adopted in traditional level set methods can be avoided by introducing a new penalizing energy. To avoid the long iteration process for level set evolution, an efficient termination criterion is presented which is based on the length change of evolving curve. Particularly, we proposed constructing an extended structure tensor (EST) by adding the intensity information into the classical structure tensor for texture image segmentation. It can be found that by combining the EST with our LCV model, the texture image can be efficiently segmented no matter whether it presents intensity inhomogeneity or not. Finally, experiments on some synthetic and real images have demonstrated the efficiency and robustness of our model. Moreover, comparisons with the well-known Chan-Vese (CV) model and recent popular local binary fitting (LBF) model also show that our LCV model can segment images with few iteration times and be less sensitive to the location of initial contour and the selection of governing parameters.     

A novel region-based active contour model based on kernel function for image segmentation

It is a difficult task to accurately segment images with intensity inhomogeneity, because most of existing algorithms are based upon the assumption of the homogeneity of image intensity. In this paper, we propose a novel region-based active contour model, referred to as the K-GLIF, which utilizes both global and local image intensity fittings with kernel functions. The model consists of an intensity fitting term and a new regularization term. The intensity fitting term of the level set function is the gradient descent flow that minimizes the global binary fitting energy functional. The local intensity fitting value based on the generalized Gaussian kernel function is then incorporated into the global intensity fitting value to form the weighted intensity fitting value on the two sides of the contour. Owing to the kernel function, the intensity information in local regions is extracted to guide the motion of the contour, which enables the model to effectively segment images with intensity inhomogeneity and smooth noise. A new regularization term is used to control the smoothness of the level set function and avoid complicated re-initialization. Experimental results and comparisons with other models of inhomogeneous images, synthetic images, medical images, multi-object images, natural and infrared images show that the proposed K-GLIF model improves the quality of image segmentation in terms of accuracy and robustness of initial contours.