Fatih Nar

Region based target detection approach for synthetic aperture radar images and its parallel implementation

Automatic target detection (ATD) methods for synthetic aperture radar (SAR) imagery are sensitive to image resolution, target size, clutter complexity, and speckle noise level. However, a robust ATD method needs to be less sensitive to the above factors. In this study, a constant false alarm rate (CFAR) based method is proposed which can perform target detection independent of image resolution and target size even in heterogeneous background clutter. The proposed method is computationally efficient since clutter statistics are calculated only for candidate target regions and a single execution of the method is sufficient for different types of targets having different shapes and sizes. Computational efficiency is further increased by parallelizing the algorithm using OpenMP and NVidia CUDA implementations.

Sparsity-Driven Despeckling for SAR Images

Speckle noise inherent in synthetic aperture radar (SAR) images seriously affects the result of various SAR image processing tasks such as edge detection and segmentation. Thus, speckle reduction is critical and is used as a preprocessing step for smoothing homogeneous regions while preserving features such as edges and point scatterers. Although state-of-the-art methods provide better despeckling compared with conventional methods, their resource consumption is higher. In this letter, a sparsity-driven total-variation (TV) approach employing l0-norm, fractional norm, or l1-norm to smooth homogeneous regions with minimal degradation in edges and point scatterers is proposed. Proposed method, sparsity-driven despeckling (SDD), is capable of using different norms controlled by a single parameter and provides better or similar despeckling compared with the state-of-the-art methods with shorter execution times. Despeckling performance and execution time of the SDD are shown using synthetic and real-world SAR images.

A robust nonlinear scale space change detection approach for SAR images

In this paper, we propose a change detection approach based on nonlinear scale space analysis of change images for robust detection of various changes incurred by natural phenomena and/or human activities in Synthetic Aperture Radar (SAR) images using Maximally Stable Extremal Regions (MSERs). To achieve this, a variant of the log-ratio image of multitemporal images is calculated which is followed by Feature Preserving Despeckling (FPD) to generate nonlinear scale space images exhibiting different trade-offs in terms of speckle reduction and shape detail preservation. MSERs of each scale space image are found and then combined through a decision level fusion strategy, namely “selective scale fusion” (SSF), where contrast and boundary curvature of each MSER are considered. The performance of the proposed method is evaluated using real multitemporal high resolution TerraSAR-X images and synthetically generated multitemporal images composed of shapes with several orientations, sizes, and backscatter amplitude levels representing a variety of possible signatures of change. One of the main outcomes of this approach is that different objects having different sizes and levels of contrast with their surroundings appear as stable regions at different scale space images thus the fusion of results from scale space images yields a good overall performance.

Sparsity-Driven Change Detection in Multitemporal SAR Images

In this letter, a method for detecting changes in multitemporal synthetic aperture radar (SAR) images by minimizing a novel cost function is proposed. This cost function is constructed with log-ratio-based data fidelity terms and an 11-norm-based total variation (TV) regularization term. Log-ratio terms model the changes between the two SAR images where the TV regularization term imposes smoothness on these changes in a sparse manner such that fine details are extracted while effects like speckle noise are reduced. The proposed method, sparsity-driven change detection (SDCD), employs accurate approximation techniques for the minimization of the cost function since data fidelity terms are not convex and the employed £1-norm TV regularization term is not differentiable. The performance of the SDCD is shown on real-world SAR images obtained from various SAR sensors.

Automatic and semi-automatic extraction of curvilinear features from SAR images

Extraction of curvilinear features from synthetic aperture radar (SAR) images is important for automatic recognition of various targets, such as fences, surrounding the buildings. The bright pixels which constitute curvilinear features in SAR images are usually disrupted and also degraded by high amount of speckle noise which makes extraction of such curvilinear features very difficult. In this paper an approach for the extraction of curvilinear features from SAR images is presented. The proposed approach is based on searching the curvilinear features as an optimum unidirectional path crossing over the vertices of the features determined after a despeckling operation. The proposed method can be used in a semi-automatic mode if the user supplies the starting vertex or in an automatic mode otherwise. In the semi-automatic mode, the proposed method produces reasonably accurate real-time solutions for SAR images.

Fast target detection in radar images using Rayleigh Mixtures and Summed Area Tables

Abstract As the first step of automatic image interpretation systems, automatic detection of targets should be accurate and fast. For Synthetic Aperture Radar (SAR) images, Constant False Alarm Rate (CFAR) is the most popular framework used for target detection. In CFAR, modeling of the clutter is crucial since the decision threshold is calculated based on this model. In this study, we propose to model the background statistics using a Rayleigh Mixture (RM) model. Such an approach facilitates modeling of complex statistics, including but not limited to those involved in heavy tailed distributions, which are shown to be good fits especially for high resolution SAR images. We also propose an efficient method to evaluate CFAR thresholds according to the proposed model by use of Summed Area Tables (SAT). SAT provides a remarkable efficiency as the Rayleigh distribution is represented by only one parameter that can be estimated using simple moments. Tiling and parallel implementation is also utilized for fast computation of results. The outcome is a highly-accurate, extremely fast, and adaptive target detection approach that can be seamlessly used with a variety of complex SAR scenes. Our experiments compare the proposed approach with existing target detection methods and demonstrate its effectiveness as well as the benefits it provides.

Fast feature preserving despeckling

Synthetic Aperture Radar (SAR) images contain high amount of speckle noise which causes edge detection, shape analysis, classification, segmentation, change detection and target recognition tasks become more difficult. To overcome such difficulties, smoothing of homogenous regions while preserving point scatterers and edges during speckle reduction is quite important. Besides, due to huge size of SAR images in remote sensing applications efficiency of computational load and memory consumption must be further improved. In this paper, a parallel computational approach is proposed for the Feature Preserving Despeckling (FPD) method which is chosen due to its success in speckle reduction. Speckle reduction performance, execution time and memory consumption of the proposed Fast FPD (FFPD) method is shown using spot mode SAR images.

Scattered data estimation on medical images for cranioplasty applications

Cranioplasty is a surgical operation to repair hole or defects on skull. 3 dimensional computed tomography (CT) images are used for automatic determination of the shape of implant which is used for repairing defect. The designing implant by mathematical model and manufacturing it before operation lowers the operation cost. In this paper, previous studies are examined, applications are realized by radial basis functions (RBF) and insufficient sections of previous studies are revealed.

GPU efficient SAR image despeckling using mixed norms

Speckle noise which is inherent to Synthetic Aperture Radar (SAR) imaging obstructs various image exploitation tasks such as edge detection, segmentation, change detection, and target recognition. Therefore, speckle reduction is generally used as a first step which has to smooth out homogeneous regions while preserving edges and point scatterers. Traditional speckle reduction methods are fast and their memory consumption is insignificant. However, they are either good at smoothing homogeneous regions or preserving edges and point scatterers. State of the art despeckling methods are proposed to overcome this trade-off. However, they introduce another trade-off between denoising quality and resource consumption, thereby higher denoising quality requires higher computational load and/or memory consumption. In this paper, a local pixel-based total variation (TV) approach is proposed, which combines l2-norm and l1-norm in order to improve despeckling quality while keeping execution times reasonably short. Pixel-based approach allows efficient computation model with relatively low memory consumption. Their parallel implementations are also more efficient comparing to global TV approaches which generally require numerical solution of sparse linear systems. However, pixel-based approaches are trapped to local minima frequently hence despeckling quality is worse comparing to global TV approaches. Proposed method, namely mixed norm despeckling (MND), combines l2-norm and l1-norm in order to improve despeckling performance by alleviating local minima problem. All steps of the MND are parallelized using OpenMP on CPU and CUDA on GPU. Speckle reduction performance, execution time and memory consumption of the proposed method are shown using synthetic images and TerraSAR-X spot mode SAR images.

Sparsity-driven weighted ensemble classifier

In this study, a novel sparsity-driven weighted ensemble classifier (SDWEC) that improves classification accuracy and minimizes the number of classifiers is proposed. Using pre-trained classifiers, an ensemble in which base classifiers votes according to assigned weights is formed. These assigned weights directly affect classifier accuracy. In the proposed method, ensemble weights finding problem is modeled as a cost function with the following terms: (a) a data fidelity term aiming to decrease misclassification rate, (b) a sparsity term aiming to decrease the number of classifiers, and (c) a non-negativity constraint on the weights of the classifiers. As the proposed cost function is non-convex thus hard to solve, convex relaxation techniques and novel approximations are employed to obtain a numerically efficient solution. Sparsity term of cost function allows trade-off between accuracy and testing time when needed. The efficiency of SDWEC was tested on 11 datasets and compared with the state-of-the art classifier ensemble methods. The results show that SDWEC provides better or similar accuracy levels using fewer classifiers and reduces testing time for ensemble.