Level Set Method for SAR Image Coregistration

In this letter, a novel approach for the coregistration of synthetic aperture radar (SAR) images is proposed based on the level set method. The features of regions are detected by segmenting the images, and the images are coregistered by matching the detected features. The energy functional of level sets is formulated with respect to detecting and matching features. The coregistration is achieved by minimizing the energy functional. Compared with the conventional tie-patch method, the results on a series of simulated experiments and real SAR data demonstrate the feasibility of the proposed approach.      

基于交互式分割技术和决策级融合的SAR图像变化检测

为免去去除斑点噪声的预处理操作及克服选择分布模型的限制,本文结合差异图的特点和一种不涉及分布模型的交互式分割方法,产生不同“种子点”下的变化检测结果后,再利用投票策略对其进行决策级的融合给出最终的变化检测结果。分割中,将每个像素的特征设为差异图及由静态小波变换分解差异图再丢弃高频系数后重构得到的各层表示中,其对应位置上的灰度值构成的矢量。此特征及决策级融合的策略使本文的变化检测技术对SAR图像中的斑点噪声具有一定的鲁棒性。在无需对SAR图像做预处理的情况下,对真实SAR图像数据集的变化检测结果,其效果优于其他相关技术的。     

局部统计活动轮廓模型的SAR图像海岸线检测

首次将局部统计活动轮廓模型引入SAR图像海岸线检测问题中,提出了一种基于局部统计活动轮廓模型的SAR 图像海岸线检测方法。首先利用C-V模型进行粗分割,消除局部统计活动轮廓模型对初始轮廓线设置要求严格的限制,然后提出了一种基于G⁰分布的局部统计活动轮廓模型,进行精细分割。该模型采用G⁰分布对轮廓线上每一点的邻域进行统计建模,增强了模型数据拟合能力,提高了海岸线检测精度,加入水平集函数惩罚项,消除了重新初始化过程。实测SAR图像实验表明,本文方法可用于精确海岸线检测。     

SAR图像河流分割的加权指数区域能量模型

传统主动轮廓模型很难实现精确的SAR图像河流分割。针对这一问题,本文提出了一种加权指数区域能量主动轮廓模型,以精确地提取SAR图像中的河流。该模型在Chan-Vese(CV)模型能量泛函中引入了指数区域能量,能更好地衡量分割图像和原始图像的差异程度,提高模型的分割准确性。此外,利用目标区域和背景区域内像素灰度的最大绝对差取代模型中常值区域能量权重,自适应地调节目标区域和背景区域的能量比重,加速曲线运动到目标区域的边缘,获得更高的分割效率。针对实际河流SAR图像进行了分割试验,结果表明:与传统主动轮廓模型相比,本文提出的模型能更快速、精确地分割SAR图像中的河流,在分割结果和分割效率两方面具有优势。     

An unsupervised segmentation method based on MPM for SAR images

An unsupervised segmentation method for synthetic aperture radar (SAR) images is proposed. It alternately approximates the maximization of the posterior marginals estimate of the pixel class labels and estimates all model parameters except the number of classes during segmentation. In this method, a multilevel logistic (MLL) model for the pixel class labels and Gamma distribution for the marginal distribution of each class in the observed SAR image are employed. In our implementation, the expectation-maximization algorithm is used to estimate parameters of the Gamma distributions, and the iterative conditional estimation algorithm is used to estimate the MLL model parameters. The segmentation results for synthetic and real SAR images show that the proposed method has a good performance.     

一种有效的MSTAR SAR图像分割方法

提出了一种有效的MSTAR SAR图像分割方法。该方法首先对待处理图像进行过分割操作,得到过分割图像区域,然后对过分割后的图像进行图像区域级和像素级的特征提取,得到用于表示图像的特征向量,接着对MSTAR SAR图像使用空间隐含狄利克雷分配模型(sLDA)和马尔科夫随机场(MRF)建立本文所提出的模型,得到能量泛函,最后运用Graph-Cut算法和Branch-and-Bound算法对能量泛函进行优化,得到最终的分割结果。通过使用MSTAR SAR图像进行分割实验比较,仿真结果表明了方法的有效性。     

利用倒数灰度熵和改进Chan-Vese模型进行SAR河流图像分割

为了进一步提高合成孔径雷达(SAR)图像中河流分割的精度和速度,提出了一种基于人工蜂群优化的倒数灰度熵多阈值选取与改进Chan-Vese(CV)模型相结合的分割方法。考虑SAR图像中河流目标和背景类内灰度的均匀性,提出了基于蜂群优化的倒数灰度熵多阈值选取方法,以此对河流图像进行粗分割;针对基本CV模型收敛速度低、对初始条件敏感的问题,利用图像边缘强度取代Dirac函数,将粗分割结果作为改进CV模型的初始条件,对河流图像进行细分割。大量试验结果表明,所提出的分割方法无须设置初始条件,运行速度快,分割精度高。     

基于MPM准则的无监督SAR图像分割

提出了一种基于MPM(maximization of the posterior marginals)准则的SAR图像无监督分割方法，并给出了对模拟和真实SAR图像的分割结果。     

利用纹理融合与广义高斯模型的高分辨率SAR影像变化检测

为了充分利用高分辨率SAR影像的纹理特征,提出一种纹理信息融合与广义高斯模型相结合的SAR影像变化检测方法。通过灰度共生矩阵计算影像的纹理特征进而构造纹理差异影像,利用离散平稳小波变换,融合灰度差异影像和纹理差异影像。然后利用广义高斯模型进行统计建模,估计融合后差异影像上变化类和未变化类的概率分布,利用KI阈值准则获取最佳分割阈值,实现多时相SAR影像的非监督变化检测。选取两组TerraSAR-X数据进行实验,结果表明融合纹理信息与广义高斯模型的变化检测方法可行,其中融合逆差距纹理信息的检测性能最优。     

SAR影像中海洋浮油膜特征分割的Level Set方法

介绍了一种全新的区域影像分割技术--基于迎风格式偏微分方程（PDE）的Level Set方法进行海洋浮油膜特征提取.在该方法中,海洋浮油膜特征表示为扩散界面,影像灰度的梯度决定了界面扩散的方向和强度.界面边缘在影像灰度差异动力和曲率流的共同作用下不断向外扩散,能够有效地克服尖锐突起和裂缝等,在特征边缘趋于稳定和光滑.同时,该方法对高噪声具有一定抑制作用,适用于低对比度高噪声图像,特别是SAR图象中特征提取.以不同地区ERS-2 SAR图像中海洋浮油膜提取为例,进行方法验证,同时对传统的影像分割技术进行对比.     

Using SAR Images to Detect Ships From Sea Clutter

An innovative constant false alarm rate (CFAR) algorithm was studied for ship detection using synthetic aperture radar (SAR) images of the sea. Two advances were achieved. An alpha-stable distribution rather than a traditional Weibull or -distribution was used to model the distribution of sea clutter. The distribution of sea clutter in a SAR image was typically heterogeneous, caused mainly by variable wind and current conditions. Image segmentation was carried out to improve the homogeneity of the distribution in each subimage or region. In comparison with ship detection using the CFAR algorithms based on the Weibull or K -distribution, our algorithm detected the most number of ships with the smallest number of false alarms.     

Accounting for image uncertainty in SAR-based flood mapping

Operational flood mitigation and flood modeling activities benefit from a rapid and automated flood mapping procedure. A valuable information source for such a flood mapping procedure can be remote sensing synthetic aperture radar (SAR) data. In order to be reliable, an objective characterization of the uncertainty associated with the flood maps is required.This work focuses on speckle uncertainty associated with the SAR data and introduces the use of a non-parametric bootstrap method to take into account this uncertainty on the resulting flood maps. From several synthetic images, constructed through bootstrapping the original image, flood maps are delineated. The accuracy of these flood maps is also evaluated w.r.t. an independent validation data set, obtaining, in the two test cases analyzed in this paper, F-values (i.e. values of the Jaccard coefficient) comprised between 0.50 and 0.65. This method is further compared to an image segmentation method for speckle analysis, with which similar results are obtained. The uncertainty analysis of the ensemble of bootstrapped synthetic images was found to be representative of image speckle, with the advantage that no segmentation and speckle estimations are required.Furthermore, this work assesses to what extent the bootstrap ensemble size can be reduced while remaining representative of the original ensemble, as operational applications would clearly benefit from such reduced ensemble sizes.     

指数型多尺度影像序列的C-V模型SAR岸线快速自动分割

为提高SAR影像岸线自动分割的精度和效率,针对传统二进制(影像序列生成的金字塔步长底数为a=2)多尺度C-V模型对初始条件敏感、收敛速度低的问题,提出指数型(影像序列生成的底数为a≥1)多尺度影像序列生成方法,本方法将传统多尺度影像序列的生成方式的底数2量化为a≥1的任意数,并应用筛选因素进行自动地快速识别海岸线。从海岸线分割结果和所需时间方面与已有传统二进制C-V模型算法进行对比,实验表明本文算法在保证精度的条件下单次迭代逼近海岸线的计算量上小于传统即二进制多尺度C-V模型的单次迭代计算量,总迭代次数有所减少,时间效率有所提高,提高了岸线自动分割的精度和效率。     

区域Gamma混合模型的SAR图像分割

针对传统Gamma混合模型用于SAR图像分割时忽略像素间空间相关性,导致分割结果不连续并产生大量误分割的现象,提出了区域Gamma混合模型的SAR图像分割算法。首先对图像进行分水岭分割,得到过分割区域块,然后将其作为输入样本进行基于Gamma混合模型的聚类,在模型的参数估计过程中进一步考虑区域间的空间相关性,设计邻域因子融入到迭代过程,得到邻域加权类分布概率。该算法充分利用像素间的空间相关性,能够降低噪声对分割结果的影响。通过合成图像和真实SAR图像的实验表明,本文算法能够实现SAR图像的准确分割。     

Automatic Determination of Number of Homogenous Regions in SAR Images Utilizing Splitting and Merging Based on a Reversible Jump MCMC Algorithm

This paper presents an algorithm dealing with initial segmentation of speckled Synthetic Aperture Radar (SAR) intensity images in order to automatically determine the number of homogeneous regions. Taking this problem into account, segmentation procedure utilizing splitting and merging is designed, iteratively. The proposed approach is based upon Bayesian inference, a maximum likelihood gamma distribution parameter estimator, and a Reversible Jump Markov Chain Monte Carlo (RJMCMC) algorithm. By using of image splitting operation, SAR image is partitioned into finite regions iteratively, until all individual regions are coherent. Then each region is assigned a unique label to indicate the class to which the homogeneous region belongs. The intensities of pixels in each coherent region are assumed to satisfy identical and independent gamma distribution. Then an RJMCMC scheme is designed to simulate the posterior distribution in order to estimate the number of components and delineate an initial segmentation. Thus, the main purpose of this research is to define the number of homogeneous regions rather than a perfect segmentation, i.e. model outputs can be served for unsupervised segmentation methodologies as prior information. The results obtained from Radarsat-1/2 of SAR intensity images show that the proposed algorithm is both capable and reliable in defining the accurate number of homogeneous regions in a wide variety of SAR intensity images, comprising a high level of speckle noise.     

SAR change detection based on intensity and texture changes

In this paper, a novel change detection approach is proposed for multitemporal synthetic aperture radar (SAR) images. The approach is based on two difference images, which are constructed through intensity and texture information, respectively. In the extraction of the texture differences, robust principal component analysis technique is used to separate irrelevant and noisy elements from Gabor responses. Then graph cuts are improved by a novel energy function based on multivariate generalized Gaussian model for more accurately fitting. The effectiveness of the proposed method is proved by the experiment results obtained on several real SAR images data sets.     

Robust river boundaries extraction of dammed lakes in mountain areas after Wenchuan Earthquake from high resolution SAR images combining local connectivity and ACM

River boundaries extraction from SAR imagery is valuable for flood monitoring and damage assessment. Several rivers, parts of which include dammed lakes caused by landslides and rock avalanches triggered by the 2008 Wenchuan Earthquake, were taken as a case study for robust extraction. In this paper, a novel state-of-the-art approach for automated river boundaries extraction using high resolution synthetic aperture radar (SAR) intensity imagery is presented. The key of our approach lies in the combined usage of local connectivity feature of the river and a region-based active contours model (ACM) in a variational level set framework to differentiate between river and the background. First, sub-patched intensity thresholding segmentation is applied to SAR imagery. Pixels with intensities below the threshold are selected as potential river pixels while the others are potential background pixels. Second, potential river pixels are divided into several connected regions, considering that the river is a big connected region, only relatively bigger regions with similar contrast value are retained as the regions of interest (ROI) while others are noise due to pixel-level decision approach in the first step or shadows due to mountains terrain. Third, the ROI and their contours are regarded as local region and the initial contours to refine the river boundaries, which are used to reduce the scene complexity of ACM and its sensitivity to initial situation, respectively. A novel ACM driven by local image fitting (LIF) energy is presented and used for river boundaries extraction for the first time, which is not only robust against inhomogeneity widely spread in SAR imagery but also can work with efficiency without the need of re-initialization during iteration compared to traditional ACM. The proposed approach was tested on numerous high resolution airborne SAR images containing connected rivers or dammed lakes obtained by Chinese domestic radar system after Wenchuan Earthquake. For the overall dataset, the average commission error, omission error and root mean squared error were 6.5%, 3.3%, and 0.51, respectively. The average computational time for 4000 by 4000 image size was 21 min using a PC-based MATLAB platform. Our experimental results demonstrate that the proposed approach is robust and effective.     

混合智能优化算法的SAR图像特征选择

为提高SAR图像自动目标识别的准确率及实时性,提出了一种基于混合智能优化的SAR图像特征选择算法。首先,采用分形特征对SAR图像进行增强,基于分割后的图像提出了一种基于图像矩的方位角估计方法。然后基于未校正和校正后的图像分别提取Zernike矩、Gabor小波系数和灰度共生矩阵构成候选特征集合,使用遗传算法结合二值粒子群的混合优化算法实现SAR图像特征选择。最后,采用MSTAR数据库验证本文算法的有效性。实验结果表明,优化后的特征集合具有一定泛化能力,一方面提高了SAR目标识别的准确率,另一方面减小了SAR图像目标识别的时间。     

边缘约束下的分形网络分割算法

分形网络演化算法（fractal net evolution approach，FNEA）是一种有效的多尺度影像分割算法，但对于具有斑点噪声、局部区域对比度低等特点的高分辨率合成孔径雷达（synthetic aperture radar，SAR）图像，直接应用FNEA算法得到的分割结果难以用于后续的面向对象影像分析。提出了基于边缘约束的FNEA（edge restricted FNEA，eFNEA）算法，通过加入边缘信息和构建异质性规则来为分割融入更多信息，提高分割效果。实验结果表明，对于微弱边缘和噪声污染严重等情形，eFNEA算法的分割结果均优于FNEA算法。