Konstantinos Karantzalos

Recognition-Driven Two-Dimensional Competing Priors Toward Automatic and Accurate Building Detection

In this paper, a novel recognition-driven variational framework, toward multiple building extraction from aerial and satellite images, is introduced. To this end, competing shape priors are considered, and building extraction is addressed through an image segmentation approach that involves the use of a data-driven term constrained from the prior models. The proposed framework extends previous approaches toward the integration of multiple shape priors into the level-set segmentation. In particular, it estimates the number of buildings as well as their pose from the observed data. Therefore, it can address multiple building extraction from a single optical image, a highly demanding task of fundamental importance in various geoscience and remote-sensing applications. Furthermore, it can be easily extended to deal with other remote-sensing data through a simple modification of the image term. Very promising experimental results and the performed qualitative and quantitative evaluation demonstrate the potential of our approach.

Improving edge detection and watershed segmentation with anisotropic diffusion and morphological levellings

Edge preserving smoothing and image simplification is of fundamental importance in a variety of remote sensing applications during feature extraction and object detection procedures. The construction of a pre‐processing filtering tool for edge detection and segmentation tasks is still an open matter. Towards this end, this paper brings together two advanced nonlinear scale space representations, anisotropic diffusion filtering and morphological levellings, forming a processing scheme by their combination. The proposed scheme was applied to edge detection and watershed segmentation tasks. The experimental results showed that the developed scheme generated an effective pre‐processing tool for automatic olive tree detection and solving watershed over‐segmentation problems.

Building detection in very high resolution multispectral data with deep learning features

The automated man-made object detection and building extraction from single satellite images is, still, one of the most challenging tasks for various urban planning and monitoring engineering applications. To this end, in this paper we propose an automated building detection framework from very high resolution remote sensing data based on deep convolutional neural networks. The core of the developed method is based on a supervised classification procedure employing a very large training dataset. An MRF model is then responsible for obtaining the optimal labels regarding the detection of scene buildings. The experimental results and the performed quantitative validation indicate the quite promising potentials of the developed approach.

Monitoring urban changes based on scale-space filtering and object-oriented classification

Abstract This paper introduces a multi-temporal image processing framework towards an efficient and (semi-) automated detection of urban changes. Nonlinear scale space filtering was embedded in an object-based classification procedure and the resulted simplified images provided a more compact and reliable source in order to generate image objects in various scales. In this manner the multiresolution segmentation outcome was constrained qualitatively. Multivariate alteration detection (MAD) transformation was applied afterwards on the simplified data to facilitate the detection of possible changes. The altered image regions along with the simplified data were further analyzed through a multilevel knowledge-based classification scheme. The developed algorithm was implemented on a number of multi-temporal data acquired by different remote sensing sensors. The qualitative and quantitative evaluation of change detection results performed with the help of the appropriate ancillary ground truth data. Experimental results demonstrated the effectiveness of the developed scale-space, object-oriented classification framework.

A Region-based Level Set Segmentation for Automatic Detection of Man-made Objects from Aerial and Satellite Images

A region-based level set segmentation was developed for the automatic detection of man-made objects from aerial and satellite images. The essence of the approach is to optimize the position and the geometric form of an evolving curve, by measuring information within the regions that compose a particular image partition based on their statistical description. The present region-based variational model is fully automated without the need to manually specify the position of the initial contour. Furthermore, it converges after a small number of iterations, allowing real-time applications. The developed algorithm was tested for the detection of roads, buildings and other man-made objects in a number of aerial and satellite images. The effectiveness of the algorithm is demonstrated by the experimental results and the performed qualitative and quantitative evaluation.

Automatic detection and tracking of oil spills in SAR imagery with level set segmentation

Automatic detection and monitoring of oil spills and illegal oil discharges is of fundamental importance in ensuring compliance with marine legislation and protection of the coastal environments, which are under considerable threat from intentional or accidental oil spills, uncontrolled sewage and wastewater discharged. In this paper, the level set based image segmentation was evaluated for the real‐time detection and tracking of oil spills from SAR imagery. The processing scheme developed consists of a pre‐processing step, in which an advanced image simplification takes place, followed by a geometric level set segmentation for the detection of possible oil spills. Finally, a classification was performed for the separation of look‐alikes, leading to oil spill extraction. Experimental results demonstrate that the level set segmentation is a robust tool for the detection of possible oil spills, copes well with abrupt shape deformations and splits and outperforms earlier efforts that were based on different types of thresholds or edge detection techniques. The developed algorithms efficiency for real‐time oil spill detection and monitoring was also tested.

Large-Scale Building Reconstruction Through Information Fusion and 3-D Priors

In this paper, a novel variational framework is introduced toward automatic 3-D building reconstruction from remote-sensing data. We consider a subset of building models that involve the footprint, their elevation, and the roof type. These models, under a certain hierarchical representation, describe the space of solutions and, under a fruitful synergy with an inferential procedure, recover the observed scenes geometry. Such an integrated approach is defined in a variational context, solves segmentation both in optical images and digital elevation maps, and allows multiple competing priors to determine their pose and 3-D geometry from the observed data. The very promising experimental results and the performed quantitative evaluation demonstrate the potentials of our approach.

Automatic Descriptor-Based Co-Registration of Frame Hyperspectral Data

Frame hyperspectral sensors, in contrast to push-broom or line-scanning ones, produce hyperspectral datasets with, in general, better geometry but with unregistered spectral bands. Being acquired at different instances and due to platform motion and movements (UAVs, aircrafts, etc.), every spectral band is displaced and acquired with a different geometry. The automatic and accurate registration of hyperspectral datasets from frame sensors remains a challenge. Powerful local feature descriptors when computed over the spectrum fail to extract enough correspondences and successfully complete the registration procedure. To this end, we propose a generic and automated framework which decomposes the problem and enables the efficient computation of a sufficient amount of accurate correspondences over the given spectrum, without using any ancillary data (e.g., from GPS/IMU). First, the spectral bands are divided in spectral groups according to their wavelength. The spectral borders of each group are not strict and their formulation allows certain overlaps. The spectral variance and proximity determine the applicability of every spectral band to act as a reference during the registration procedure. The proposed decomposition allows the descriptor and the robust estimation process to deliver numerous inliers. The search space of possible solutions has been effectively narrowed by sorting and selecting the optimal spectral bands which under an unsupervised manner can quickly recover hypercube’s geometry. The developed approach has been qualitatively and quantitatively evaluated with six different datasets obtained by frame sensors onboard aerial platforms and UAVs. Experimental results appear promising.

A Scalable Geospatial Web Service for Near Real-Time, High-Resolution Land Cover Mapping

A land cover classification service is introduced toward addressing current challenges on the handling and online processing of big remote sensing data. The geospatial web service has been designed, developed, and evaluated toward the efficient and automated classification of satellite imagery and the production of high-resolution land cover maps. The core of our platform consists of the Rasdaman array database management system for raster data storage and the open geospatial consortium web coverage processing service for data querying. Currently, the system is fully covering Greece with Landsat 8 multispectral imagery, from the beginning of its operational orbit. Datasets are stored and preprocessed automatically. A two-stage automated classification procedure was developed which is based on a statistical learning model and a multiclass support vector machine classifier, integrating advanced remote sensing and computer vision tools like Orfeo Toolbox and OpenCV. The framework has been trained to classify pansharpened images at 15-m ground resolution toward the initial detection of 31 spectral classes. The final product of our system is delivering, after a postclassification and merging procedure, multitemporal land cover maps with 10 land cover classes. The performed intensive quantitative evaluation has indicated an overall classification accuracy above 80%. The system in its current alpha release, once receiving a request from the client, can process and deliver land cover maps, for a 500-

Vineyard Detection and Vine Variety Discrimination from Very High Resolution Satellite Data

\text{km}^2