Unsupervised learning of a finite gamma mixture using MML: application to SAR image analysis

The technique of image histogram modification can be applied to the problem of image enhancement. Global and local histogram equalisation and local area histogram equalisation are two well-known techniques for the same purpose. In this research work, a new method is proposed to enhance the contrast of bimodal SAR images using histogram specification with Gamma distribution. The method is aimed to read the original image and calculate its original histogram. Then apply the maximum likelihood with Gamma distribution (ML) technique on the original histogram to get accurate statistical parameters as the means and prior probabilities of the two modes. The two estimated modes are thus separated by shifting the first mode to the left or shifting the second mode to the right or performing both shifts. After shifting, a new histogram called 'desired histogram' is generated from a mixture of Gamma distributions using updated data. By applying a histogram specification method, a high contrast SAR image will be produced. The new method of contrast enhancement of SAR image has been applied on bimodal SAR images and showed good results.

In Minimum Message Length (MML) clustering (unsupervised classification, mixture modelling) the aim is to infer a set of classes that best explains the observed data items. There are cases where parts of the observed data do not need to be explained by the inferred classes but can be used to improve the inference and resulting predictions. Our main contribution is to provide a simple and flexible way of using such context data in MML clustering. This is done by replacing the traditional mixing proportion vector with a new context matrix. We show how our method can be used to give evidence regarding the presence of apparent long-term trends in climate-related atmospheric pressure records. Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) solutions for our model have also been implemented to compare with the MML solution.

This article, written by Special Publications Editor Adam Wilson, contains highlights of paper SPE 190087, “Unsupervised Statistical Learning With Integrated Pattern-Based Geostatistical Simulation,” by Q. Li and R. Aguilera, SPE, University of Calgary, prepared for the 2018 SPE Western Regional Meeting, Garden Grove, California, USA, 22–27 April. The paper has not been peer reviewed. This paper presents a new geostatistics modeling methodology that connects geostatistics and machine-learning methodologies, uses nonlinear topological mapping to reduce the original high-dimensional data space, and uses unsupervised-learning algorithms to bypass problems with supervised-learning algorithms. The algorithm presented is a neural topology-preserving pattern-based geostatistical simulation algorithm that integrates the self-organizing map (SOM) concept and its updated version—growing self-organizing map (GSOM)—with an unsupervised competitive learning structure. Introduction In oil and gas reservoir modeling, any model construction faces challenges of limited data to some extent. The heuristic behind all geostatistical techniques is the implicit existence of statistical relationships among available data. “Data” here is a broad term; it could be discrete points, such as porosity or permeability at certain locations, but it also could be training images (TIs), which are used in this work. Using TIs as input data originated with multiple-point geostatistics. The aim was to overcome the limitations of using traditional two-point statistical variograms to describe geological continuity, especially in the case of curvilinear structures, which are quite common in nature, such as in fracture networks and geological fluvial structures. The authors write that geostatistics could benefit from the machine-learning or statistical-learning areas. Machine-learning tasks can be divided into two protocols, supervised learning and unsupervised learning. The difference depends on whether the input data have correct labels or not. For the investigation considered in this paper, after retrieving image patches from TIs, machine-learning algorithms were used to cluster those image patches into different classifications. If the correct clusters to which the image patches belong is known beforehand, the data can be said to have correct labels. Further, a large amount of these labeled data could be used for a model to learn. Supervised learning occurs when the model is finely tuned by guidance of these correct labels using an error-correction process. Unsupervised learning, on the other hand, occurs when neither how many clusters exist nor the correct clusters to which the image patches belong is known. Thus, without correct labels for guidance, this problem is identified as an unsupervised-learning problem. This is the first major drawback of applying machine-learning algorithms to geostatistical simulations. Two other important issues are encountered when performing pattern-based geostatistical simulations. The first is related to the large number of image patches. This large image-patch database contains high pattern redundancy, which could make pattern similarity comparison inefficient. The second issue is the high dimensionality of each image patch, which could typically be described by low-dimensional nonlinear-structure manifolds. To model and visualize these high-dimensional data, a nonlinearity dimensionality reduction should be sought.

Accurate optical and synthetic aperture radar (SAR) image registration is crucial to multisensor remote sensing applications. Though several algorithms have been proposed, the practical implementation that directly matches large and high-resolution optical and SAR images remains underexplored by the community. Equipped with the satellite positioning parameters, optical and SAR images can be roughly registered based on geographic coordinates. However, the relative positioning accuracy is still dozens, even hundreds of pixels due to the inaccuracies of sensor parameters and elevation. Consequently, we propose a robust registration algorithm, which consists of two stages, where the horizontal positioning errors can be reduced in the first stage, and then, we fine-tune the correspondences in the second stage. Specifically, we propose a novel template matching method based on the dilated convolutional feature (DCF) and epipolar-oriented phase correlation. DCF is constructed by a depthwise-separable dilated convolution with multichannel gradients, which are generated by the Sobel operator for optical images and a ratio of exponentially weighted averages (ROEWA) operator for SAR images. Due to the large reception field of dilated convolution, DCF can retain invariance even for large relative positioning errors. The epipolar-oriented rectangle template, which stretches along the epipolar line, is then proposed to capture more overlapping areas compared to square templates. Furthermore, the outlier removal is implemented in the coordinate system of optical images to avoid the effect of range compression in SAR images. Inliers are finally used to refine the rational polynomial coefficients (RPCs) based on the bundle adjustment technique. Experimental results on high-resolution optical and SAR image products of various scenarios demonstrate the effectiveness of the proposed registration framework. The relative positioning errors of the refined RPCs can be reduced from hundreds of pixels to the subpixel level.

Images segmentation is an important issue for many applications as pattern recognition and computer vision. Thresholding is an important and fast technique used in most applications. Gaussian Otsu's method is a thresholding technique based on between class variance. Gamma distribution models data more than Gaussian distribution. In this paper, we developed a new formula using Otsu's method for estimating the optimal threshold values based on Gamma distribution. Our method applied on bimodal and multimodal images. Also It uses an iteratively rather than sequentially to decrease the number of operations. Further, using Gamma distribution give satisfying thresholding results in low-high contrast images where modes are symmetric or non-symmetric. For our results, we compared it with the original Gaussian Otsu's method.

Ship detection from synthetic aperture radar (SAR) images is a hot topic, but the difficulty in collecting labeled SAR images may hinder the development of deep-learning-based detection methods. Inspired by the idea of domain adaptation, in this article, we propose a hierarchical similarity alignment neural network (HSANet) for ship detection in SAR images, which is a domain adaptive (DA) approach with optical remote sensing images as training samples. The kernel target of HSANet is to mine and align both the global structure and the local instance information between SAR and optical images, where two modules, structural alignment module (SAM) and prototype alignment module (PAM), are designed to, respectively, conduct two hierarchies of alignment process. In general, SAM attempts to extract the global structure similarity which exists in image-level feature representation, while PAM tends to extract the local shape similarity which is instance-level representation. To be specific, SAM is developed by Fourier-based feature alignment, which tries to describe the similar structural relationship between optical and SAR images. Meanwhile, PAM is proposed based on the conjoint confidence analysis where the instance-level ship representations of the source and target domains is aligned. SAM and PAM work together to construct a hierarchical domain adaptation network for SAR ship detection. Experiments on several public datasets may indicate the effectiveness of the proposed method.

Reliable and accurate land use/land cover (LULC) map is a crucial data source for the understanding of coupled human–environment systems, monitoring changes, timely low-cost planning, and management of natural resources. Improvements in sensor technologies and machine learning capabilities have shifted the attention of remote sensing community to data complementarity through fusion of multi-sensor data for accurate feature extraction and mapping. Amalgamation of optical and synthetic aperture radar (SAR) images has shown promising advantages in enhancing the accuracy of extracting LULC as such method allows exploitation of information in sensors. This study investigated the potential of using freely available multisource Sentinel images to extract LULC maps in semi-arid environments through multi-scale geographic object-based image analysis (GEOBIA). A multi-scale classification framework that integrates GEOBIA, correlation-based feature selection (CFS), and random forest (RF)-supervised classification was adopted to extract LULC from assimilation of Sentinel multi-sensor products. First, Sentinel-1 and -2 images were pre-processed. Second, optimum multi-scale segmentation levels were selected using F-score segmentation quality measures. Third, 70 features of various spectral indices and derivatives and geometrical features from optical data and multiple ratios and textural features from dual-polarization SAR images were computed, and a CFS based on wrapper approach was used to select the most significant features at multi-scale levels. Finally, a single and multi-scale RF classifier was used to extract LULC classes using the most relevant features extracted from Sentinel SAR and optical images. Results of multi-scale image segmentation optimization showed that scale parameter (SP) values of 40, 60, and 150 were optimal for extraction of LULC classes. Results of feature selection showed that 22, 24, and 27 features were selected at scale SP values of 40, 60, and 150, respectively. Half of the features were common among the three scales. Single RF classification yielded overall accuracy (OA) values of 92.10%, 93%, and 91% and kappa coefficients of 0.901, 0.912, and 0.89 at scale values of 150, 60, and 40, respectively. Multiscale RF classification from scale values of 150 and 60 produced better LULC classification with OA 96.06% and kappa coefficient of 0.95 compared with other scale SP values. The integrated approach demonstrated an effective and promising method for high-quality LULC extraction from coupling optical and SAR images. Overall, multi-sensor Sentinel images along with the adopted approach feature a remarkable potential for improving LULC extraction and can effectively be used to update geographic information system layers for various applications.

Synthetic aperture radar (SAR) and optical imaging are different remote sensing methods. Given a SAR image, is it possible to predict what the observed scene looks like in an optical image? Transfer between SAR data and optical data seems to be impossible. However, this article shows examples that by applying deep learning techniques on high resolution airborne SAR images and GoogleEarth optical images, the SAR images and optical images can transfer with each other. The transferring help us to better understand the relationship between SAR and optical image, and can be potentially used to transfer detection or classification algorithms for optical image straightforwardly to be applied on SAR image.

Segmentation is a low-level processing aimed at the partition of an image in disjoint regions, each one homogeneous with respect to some properties like intensity, texture, shape, etc. Such a task is needed in many high-level processing and applications in such diverse fields as remote-sensing, medical imaging, video coding, and industrial automation, just to name a few. In this work of thesis, we first deal with the problem of color based image segmentation, that is based on some form of homogeneity in the color/spectral properties of the image, and then move to the more complex task of texture based segmentation, where the aim is to recognize complex structures in the image which are typically non homogeneous in terms of spectral properties. Concerning color based segmentation, we focus on a specific class of Markov Random Field (MRF) probability models, namely the Tree-Structured Markov Random Fields (TS-MRF), recently proposed to provide a hierarchical multiscale description of images. Based on such a model, the unsupervised image segmentation is carried out by means of a sequence of nested class splits, where each class is modeled as a local binary MRF. As a first contribution, we first extend the model to generic tree structures removing the unnecessary binary constraint, and then devise a new TS-MRF unsupervised segmentation technique which improves upon the original algorithm by selecting a better tree structure and eliminating spurious classes. Such results are obtained by using the Mean-Shift procedure to estimate the number of pdf modes at each node, and to obtain a more reliable initial clustering for subsequent MRF optimization. To this end, we devise a new reliable and fast clustering algorithm based on the Mean-Shift technique, which makes use of a variable-bandwidth strategy based on the k-Nearest Neighbors (k-NN) technique, and is implemented with a speed-up strategy that cuts significantly the computational complexity. Experimental results on synthetic data and real life images from the remote sensing domain prove the potential of the proposed method. Turning to texture based segmentation, we present a novel multiscale texture model, and a related algorithm for the unsupervised segmentation of color images. Elementary textures are characterized by their spatial interactions with neighboring regions along selected directions. Such interactions are modeled in turn by means of a set of Markov chains, one for each direction, whose parameters are collected in a feature vector that synthetically describes the texture. Based on the feature vectors, the texture are then recursively merged, giving rise to larger and more complex textures, which appear at different scales of observation: accordingly, the model is named Hierarchical Multiple Markov Chain (H-MMC). The Texture Fragmentation and Reconstruction (TFR) algorithm, addresses the unsupervised segmentation problem based on the H-MMC model. The ``fragmentation'' step allows one to find the elementary textures of the model, while the ``reconstruction'' step defines the hierarchical image segmentation based on a probabilistic measure (texture score) which takes into account both region scale and inter-region interactions. The performance of the proposed method was assessed through the Prague segmentation benchmark, based on mosaics of real natural textures, and also tested on real-world natural images from the Berkeley dataset. Finally, an application of the TFR algorithm for the unsupervised segmentation of multiresolution remotely sensed images is proposed, based on the significant presence of textural content in last generation high-resolution satellite images. The obtained results further assess the potential of the proposed method for such real life applications.

Online clustering via finite mixtures of Dirichlet and minimum message length

Sea ice as a disaster has recently attracted a great deal of attention in China. Its monitoring has become a routine task for the maritime sector. Remote sensing, which depends mainly on SAR and optical sensors, has become the primary means for sea-ice research. Optical images contain abundant sea-ice multi-spectral information, whereas SAR images contain rich sea-ice texture information. If the characteristic advantages of SAR and optical images could be combined for sea-ice study, the ability of sea-ice monitoring would be improved. In this study, in accordance with the characteristics of sea-ice SAR and optical images, the transformation and fusion methods for these images were chosen. Also, a fusion method of optical and SAR images was proposed in order to improve sea-ice identification. Texture information can play an important role in sea-ice classification. Haar wavelet transformation was found to be suitable for the sea-ice SAR images, and the texture information of the sea-ice SAR image from Advanced Synthetic Aperture Radar (ASAR) loaded on ENVISAT was documented. The results of our studies showed that, the optical images in the hue-intensity-saturation (HIS) space could reflect the spectral characteristics of the sea-ice types more efficiently than in the red-green-blue (RGB) space, and the optical image from the China-Brazil Earth Resources Satellite (CBERS-02B) was transferred from the RGB space to the HIS space. The principal component analysis (PCA) method could potentially contain the maximum information of the sea-ice images by fusing the HIS and texture images. The fusion image was obtained by a PCA method, which included the advantages of both the sea-ice SAR image and the optical image. To validate the fusion method, three methods were used to evaluate the fused image, i.e., objective, subjective, and comprehensive evaluations. It was concluded that the fusion method proposed could improve the ability of image interpretation and sea-ice identification.

Different approaches have been proposed in the literature for anomaly detection in image/video area. Traditional methods such as trajectory or spatio-temporal based techniques rely on hand-crafted features. The occlusion problems and high complexity in crowded scenes are the vital drawbacks behind using these methodologies. Deep learning structures proved recently to be useful for defining effective solutions for anomaly detection where the high level features are learnt and selected automatically. However, the block-wise methods such as CNNs are computationally slow. On the other hand, they are totally supervised learning methodologies, while the video-based anomaly detection is an unsupervised problem. Auto- Encoders (Convolutional AE, vibrational AE, etc.) can be considered as an alternative option. This paper presents a stateof- the-art deep learning algorithm to be applied in such an unsupervised problem. Using the basic concepts behind the Auto-Encoders as a well-known unsupervised learning algorithm, we propose a novel methodology to detect and localize the anomalies in a video scene. The presented network is trained based on the normal patterns during training phase. The proposed structure enables the system to capture the 2D structure in image sequences during the learning process. The working hypothesis is that a deep network is able to learn normal events in videos, and, therefore, the difference of normal and anomalous frames can be used for devising an anomaly score. The simulation results on well-known data sets such as UCSD confirm that the proposed methodology achieves high performance results in case of accuracy and total processing time compared with counterpart approaches.

A new [formula omitted]-deformed parametric model for the size distribution of wealth

In this paper, the impact of non-Markovian distributed primary radio (PR) on/off activities on the opportunity time for secondary radio (SR) transmission is studied. The licensed spectrum is partitioned into N frequency bins (FBs) and each primary user is assumed to require only one FB for transmission. The PR activity in each FB is modeled as an independent on/off process. The sojourn time for PR activity in the ‘on’ state and in the ‘off’ state is described by the Pareto distribution and the hyper-Erlang distribution, respectively. Numerous cases are examined in this paper. In the first case, we assume the PR activity factors statistically identical in all the FBs. For this case, we examine the scenarios in which the PR traffic is light and also when it is heavy. In the second case, we assume that the PR activity factors between any two FBs are statistically different. Through the performance of extensive simulations, the probability density function (PDF) of the opportunity time is modeled and studied for different number of FBs, which is varied from N=2 to N=10. The goodness-of-fit of the statistical model to the simulated opportunity time is verified by the Kolmogorov-Smirnov test at 5% level of significance. The results show that when the PR activities are statistically identical, the PDF of the opportunity time can be described by a Gamma distribution. However, when the PR activities are statistically different, a hyper-Gamma distribution (mixture of Gamma distributions) provides a very good fit to the PDF of the opportunity time. We highlight the inadequacy of the Exponential and hyper-Exponential distribution fits for the PDF of the opportunity time in these cases. From these modeling results, the opportunity time can be reproduced with relative ease, and its statistics can be exploited for future SR performance studies.

Due to the differences in radiation and geometric characteristics of optical and synthetic aperture radar (SAR) images, there is still a huge challenge for accurate matching. In this paper, we propose a patch-matching network (PM-Net) to improve the matching performance of optical and SAR images. First, a multi-level keypoints detector (MKD) with fused high-level and low-level features is presented to extract more robust keypoints from optical and SAR images. Second, we use a two-channel network structure to improve the image patch matching performance. Benefiting from this design, the proposed method can directly learn the similarity between optical and SAR image patches without manually designing features and descriptors. Finally, the MKD and two-channel net-work are trained separately on GL3D and QXS-SAROPT data sets, and the PM-Net is tested on multiple pairs of optical and SAR images. The experimental results demonstrate that the proposed method outperforms four advanced image matching networks on qualitative and quantitative assessments. The quantitative experiment results show that using our method correct matching points numbers are increased by more than 1.15 times, the value of F1-measure is raised by an average of 7.4% and the root mean squared error (RMSE) is reduced by more than 15.3%. The advantages of MKD and the two-channel network are also verified through ablation experiments.