Ugur Murat Leloglu

Surface reconstruction from multiple aerial images in dense urban areas

Accurate 3D surface models of dense urban areas are essential for a variety of applications, such as cartography, urban planning and monitoring mobile communications, etc. Since manual surface reconstruction is very costly and time-consuming, the development of automated algorithms is of great importance. While most of existing algorithms focus on surface reconstruction either in rural or sub-urban areas, we present an approach dealing with dense urban scenes. The approach utilizes different image-derived cues, like multiview stereo and color information, as well as the general scene knowledge, formulated in data-driven reasoning and geometric constraints. Another important feature of our approach is simultaneous processing of 2D and 3D data. Our approach begins with two independent tasks: stereo reconstruction using multiple views and region-based image segmentation, resulting in generation disparity and segmentation maps, respectively. Then, the information derived from the both maps is utilized for generation of a dense elevation map, through robust verification of planar surface approximations for the detected regions and imposition of geometric constraints. The approach has been successfully tested on complex residential and industrial scenes.

Tuz Gölü: new CEOS reference standard test site for infrared visible optical sensors

Tuz Gölü, a dry salt lake located in the central plateau of Anatolia (38.50°N, 33.20°E, center latitude longitude) in Turkey, was established as a standard reference test site for the vicarious calibration of land surface imagers. The site was first selected on the basis of the spatial uniformity results from the analysis of Moderate Resolution Imaging Spectroradiometer (MODIS) images from July–August 2004 and July–August 2006. This proved its suitability for the calibration of high- and medium-resolution optical sensors. During the 2008 and 2009 measurement campaigns, the site was characterized in terms of its solar spectral surface reflectance factor (RF) and surface bidirectional reflectance factor (BRF) characteristics and atmospheric conditions. The average surface reflectance factor is between 0.4 and 0.6 in the visible and near infrared (VNIR) and about 0.2 in the short-wave infrared (SWIR), which makes this site suitable mainly for the VNIR spectral region. The site is spatially uniform to within 2%–4% normalized standard deviation in the VNIR with large homogeneous areas, which are suitable for the calibration of medium-resolution optical sensors. The aerosol optical thickness at 550 nm has values higher than 0.1, and the related Ångström wavelength exponent has values over 1.3 for most of the days during the 2008 and 2009 measurement campaigns, indicating the presence of anthropogenic aerosols. According to the meteorological data (rain, solar irradiance) and the availability of large, dry homogeneous surfaces in August, the site can be used for postlaunch calibration of land surface imagers at this time of the year. It is one of the eight instrumented reference standard test sites (LandNet) that is characterized on a yearly basis by TÜBİTAK UZAY (TU), Turkey, with traceability to the International System of Units (SI) standards through the National Physical Laboratory (NPL), United Kingdom, according to the satellite overpasses. The data can be provided to interested space agencies on request.

Automated region segmentation on cartridge case base

One of the significant problems encountered in criminology studies is the successful automated matching of fired cartridge cases, on the basis of the characteristic marks left on them by firearms. An intermediate step in the solution of this problem is the segmentation of certain regions that are defined on the cartridge case base. This paper describes a model-based method that performs segmentation of the cartridge case using surface height image of a center fire cartridge case base. The proposed method detects the location of the cartridge case base center and specific circular contours around it iteratively by projecting the problem to a one-dimensional feature space. In addition, the firing pin impression region is determined by utilizing an adaptive threshold that differentiates impression marks form primer region surface. Letters on the cartridge case base are also detected by using surface modeling and adaptive thresholding, in order to render the surface comparison operation robust against irrelevant surface features. Promising experimental results indicate the eligibility of the proposed method to be used for automated cartridge case base region segmentation process.

First LEO Satellite Built in Turkey: RASAT

BiLSAT* project was the first step to initiate and to improve satellite technologies in Turkey and it covered all major steps to establish the necessary infrastructure in Turkey to design, build and test a LEO satellite entirely in the country. As a second step, RASAT project has been initiated in year 2004. Except for some systems; like imaging system and solar panels, the technology of which are not yet available in Turkey; RASAT will totally be built and integrated at TUBITAK-UZAY facilities. The imaging system will have ground sampling distances of 7.5 m for panchromatic imaging, and 15 m for multispectral (red, green, blue) imaging which is an improvement compared to the imaging system of BiLSAT. What is so important about RASAT is that various new modules will also be flown for the first time on this satellite and will be tested in space for flight heritage. These modules will later form the core of future Turkish satellite, should they be successfully demonstrated in orbit. RASAT, similar to its predecessor BiLSAT, will be launched into a sun synchronous orbit at an altitude of about 700 km and an LTAN of 10:00AM or 10:30AM depending on the launch vehicle availability. The satellite will be commissioned and controlled from TUBITAK-UZAY ground station. Once the commissioning is over, the imagery data acquired from the satellite will be archieved and will be available to users. The experimental payloads will also be tested in space and thus TUBITAK-UZAY will spend a significant amount of operation time fory these experiments. The experience acquired from the RASAT project will later be used to develop further subsystems to fly in space in the future satellite projects.

Tuz Gölü site Characteristics

The Tuz Gölü site does demonstrate some challenges for being used as an absolute calibration site. These include: spatial and temporal variability on the order of +/− 0.02, atmospheric transmittance variability on the order of between 0.52 and 0.86 at 380 nm, lower spectral reflectance levels of less than 0.10 in the longer wavelengths, and limitations on the site availability (only two month period in summer). But, with each successful campaign made at this location, these challenges are being overcome and site characteristics have become better understood and techniques for compensation are being devised. With proper techniques, instrumentation, and coincident data collection this site remains a viable candidate as a vicarious calibration site for optical imaging sensors.

Noise estimation for hyperspectral imagery using spectral unmixing and synthesis

Most hyperspectral image (HSI) processing algorithms assume a signal to noise ratio model in their formulation which makes them dependent on accurate noise estimation. Many techniques have been proposed to estimate the noise. A very comprehensive comparative study on the subject is done by Gao et al. [1]. In a nut-shell, most techniques are based on the idea of calculating standard deviation from assumed-to-be homogenous regions in the image. Some of these algorithms work on a regular grid parameterized with a window size w, while others make use of image segmentation in order to obtain homogenous regions. This study focuses not only to the statistics of the noise but to the estimation of the noise itself. A noise estimation technique motivated from a recent HSI de-noising approach [2] is proposed in this study. The denoising algorithm is based on estimation of the end-members and their fractional abundances using non-negative least squares method. The end-members are extracted using the well-known simplex volume optimization technique called NFINDR after manual selection of number of end-members and the image is reconstructed using the estimated endmembers and abundances. Actually, image de-noising and noise estimation are two sides of the same coin: Once we denoise an image, we can estimate the noise by calculating the difference of the de-noised image and the original noisy image. In this study, the noise is estimated as described above. To assess the accuracy of this method, the methodology in [1] is followed, i.e., synthetic images are created by mixing end-member spectra and noise. Since best performing method for noise estimation was spectral and spatial de-correlation (SSDC) originally proposed in [3], the proposed method is compared to SSDC. The results of the experiments conducted with synthetic HSIs suggest that the proposed noise estimation strategy outperforms the existing techniques in terms of mean and standard deviation of absolute error of the estimated noise. Finally, it is shown that the proposed technique demonstrated a robust behavior to the change of its single parameter, namely the number of end-members.

The 2010 Tuz Gölü field campaign—An overview

Tuz GGölü, a salt lake situated in the central Anatolia, which dries during summer resulting in a spatially homogeneous high reflective surface is one of the LANDNET Sites [4, 6]. As a relatively new site and with support from the European Space Agency (ESA) and Space Technologies Research Institute of The Scientific and Technological Research Council of Turkey (TUBITAK UZAY) it was decided that this would be a good candidate to carry out an international comparison of this nature. This paper provides an overview of the CEOS international comparison carried out at Tuz Gölü in 13 – 27 August 2010.

Hierarchical approach to hydrological model calibration

Hydrological models have been widely used for water resources management. Successful application of hydrological models depends on careful calibration and uncertainty analysis. Spatial unit of water balance calculations may differ widely in different models from grids to hydrological response units (HRU). The Soil and Water Assessment Tool (SWAT) software uses HRU as the spatial unit. SWAT simulates hydrological processes at sub-basin level by deriving HRUs by thresholding areas of soil type, land use, and slope combinations. This may ignore some important areas, which may have great impact on hydrological processes in the watershed. In this study, a hierarchical HRU approach was developed in order to increase model performance and reduce computational complexity simultaneously. For hierarchical optimization, HRUs are first divided into two-HRU types and are optimized with respect to some relevant influence parameters. Then, each HRU is further divided into two. Each child HRU inherits the optimum parameter values of the parent HRU as its initial value. This approach decreases the total calibration time while obtaining a better result. The performance of the hierarchical methodology is demonstrated on two basins, namely Sarisu-Eylikler and Namazgah Dam Lake Basins in Turkey. In Sarisu-Eylikler, we obtained good results by a combination of curve number (CN2), soil hydraulic conductivity, and slope for generating HRUs, while in Namazgah use of only CN2 gave better results.

Land use/land cover classification for Göktürk-2 satellite

In this study, land-use/land-cover method for Gokturk-2 satellite was developed. Moreover, the developed method can be adapted to any satellite. The aim is to classify pixels in images captured by Gokturk-2 into four basic categories (soil, green areas, water bodies, and others), and to form a persistent mathematical model for this purpose. “Partial Least Squares for Discriminant Analysis (PLS-DA)” method is employed. This method forms a new sample space in such a way that it maximizes the covariance between samples and observations (in this study, each observation represents the class which the corresponding sample belongs to). Axes for the new sample space represent four classes rather than four physical bands (red, green, blue, and NIR) Gokturk-2 has. In this new space; two normal distributions from training data (one for class samples, and one for non-class samples) for each class is calculated in order to find an acceptance threshold. These thresholds are used in classification process. Test samples picked from the test dataset (three satellite images) show that the developed method reached average accuracy rate of 96.51%; while the Maximum Likelihood Estimator, which is frequently used in land use/land cover classification, (also included in commercial products such as ENVI, ERDAS) attained 86.13%.

A compact imager design concept for micro/nanosatellites using scanning mirrors

In this work, a compact imager design concept for micro and nano satellites is introduced. Although the performances of these platforms are increasing, the imagers cannot keep up with the platforms due to mass and volume limitations. Hence payloads require novel designs to have better imagers on board of these platforms, in terms of spectral resolution, spatial resolution, signal to noise ratio and field of view. The concept that we present here involves an object-space flat mirror that can be rotated around two orthogonal axes. One of the movements helps scanning the Earth in cross-track direction while the other movement helps increasing the integration time by forward motion compensation. When used in combination with a frame sensor covered with parallel filter strips, the camera can fulfill all of the above mentioned requirements simultaneously with an optimal performance. To demonstrate the effectiveness of the concept, an example design is shown that can fit into a three unit cubesat.