• Home
  • Panagiotis Sidiropoulos
Panagiotis Sidiropoulos

Panagiotis Sidiropoulos
FLOX Limited · Artificial Intelligence

PhD
Head of Department at FLOX Limited, Honorary Lecturer University College London

About

84
Publications
14,776
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
803
Citations
Additional affiliations
September 2017 - present
University College London
Position
  • Lecturer
July 2013 - March 2017
University College London
Position
  • PhD Student
October 2005 - October 2006
Aristotle University of Thessaloniki
Position
  • Research Assistant
Education
January 2010 - July 2012
University of Surrey
Field of study
  • Machine learning, Multimedia Processing
October 2007 - February 2014
Aristotle University of Thessaloniki
Field of study
  • Mathematics
October 2004 - September 2007
Aristotle University of Thessaloniki
Field of study
  • Informatics

Publications

Publications (84)
Article
Full-text available
We address the problem of automatically co-registering planetary images to a common baseline, introducing a novel generic technique that achieves an unprecedented robustness to different image inputs, thus making batch-mode co-registration achievable without requiring the usual parameter tweaking. We introduce a novel image matching technique, whic...
Article
Full-text available
This work studies the chord length distribution, in the case where both ends lie on a N -dimensional hypersphere (N ≥ 2). Actually, after connecting this distribution to the recently estimated surface of a hyperspherical cap [3], closed-form expressions of both the probability density function and the cumulative distribution function are straightfo...
Article
Full-text available
In this work, we address the problem of fast and accurate extraction of points that correspond to the same location (named tie-points) from pairs of large-sized images. Firstly, we conduct a theoretical analysis of the performance of the full-image matching approach, demonstrating its limitations when applied to large images. Subsequently, we intro...
Article
This paper proposes a method for binary image retrieval, where the black-and-white image is represented by a novel feature named the adaptive hierarchical density histogram, which exploits the distribution of the image points on a two-dimensional area. This adaptive hierarchical decomposition technique employs the estimation of point density histog...
Article
Full-text available
In this work a novel approach to video temporal decomposition into semantic units, termed scenes, is presented. In contrast to previous temporal segmentation approaches that employ mostly low-level visual or audiovisual features, we introduce a technique that jointly exploits low-level and high-level features automatically extracted from the visual...
Preprint
Full-text available
We introduce a novel neural network architecture -- Spectral ENcoder for SEnsor Independence (SEnSeI) -- by which several multispectral instruments, each with different combinations of spectral bands, can be used to train a generalised deep learning model. We focus on the problem of cloud masking, using several pre-existing datasets, and a new, fre...
Article
We introduce a novel neural network architecture—Spectral ENcoder for SEnsor Independence (SEnSeI)—by which several multispectral instruments, each with different combinations of spectral bands, can be used to train a generalised deep learning model. We focus on the problem of cloud masking, using several pre-existing datasets, and a new, freely av...
Article
Full-text available
The accurate split of large areas of land into discrete fields is a crucial step for several agriculture-related remote sensing pipelines. This work aims to fully automate this tedious and resource-demanding process using a state-of-the-art deep learning algorithm with only a single Sentinel-2 image as input. The Mask R-CNN, which has forged its su...
Article
Full-text available
We present here a dataset of nearly 5000 small craters across roughly 1700 km2 of the Martian surface, in the MC-11 East quadrangle. The dataset covers twelve 2000-by-2000 pixel Context Camera images, each of which is comprehensively labelled by six annotators, whose results are combined using agglomerative clustering. Crater size-frequency distrib...
Preprint
Data uniformity is a concept associated with several semantic data characteristics such as lack of features, correlation and sample bias. This article introduces a novel measure to assess data uniformity and detect uniform pointsets on high-dimensional Euclidean spaces. Spatial uniformity measure builds upon the isomorphism between hyperspherical c...
Article
Full-text available
Cloud masking is of central importance to the Earth Observation community. This paper deals with the problem of detecting clouds in visible and multispectral imagery from high-resolution satellite cameras. Recently, Machine Learning has offered promising solutions to the problem of cloud masking, allowing for more flexibility than traditional thres...
Conference Paper
The surface of Mars has been imaged in visible wavelengths for more than 40 years since the first flyby image taken by Mariner 4 in 1964. With higher resolution from orbit from MOC-NA, HRSC, CTX, THEMIS, and HiRISE, changes can now be observed on high-resolution images from different instruments, including spiders (Piqueux et al., 2003) near the so...
Article
Full-text available
The surface of Mars has been imaged in visible wavelengths for more than 40 years since the first flyby image taken by Mariner 4 in 1964. With higher resolution from orbit from MOC-NA, HRSC, CTX, THEMIS, and HiRISE, changes can now be observed on high-resolution images from different instruments, including spiders (Piqueux et al., 2003) near the so...
Presentation
Full-text available
This presentation is about the ongoing work on the validation of Sentinel-3 products using Deep Learning for the detection of cloud masks. In greater detail, the aim is to validate the cloud flagging data provided with Sentinel-3 (SLSTR) products using multi-band combination from Landsat-8 satellite imagery.
Article
Full-text available
The first high-resolution Digital Terrain Model (DTM) of the entire South Pole of Mars has been produced. A modified version (Kim and Muller, 2009) of a NASA-VICAR-based pipeline developed by DLR (German Aerospace Centre) and JPL (Jet Propulsion Laboratory) has been employed with image matching based on the Gotcha (Gruen-Otto-Chau) algorithm (Shin...
Preprint
Full-text available
In this paper, we introduce a variation of a state-of-the-art real-time tracker (CFNet), which adds to the original algorithm robustness to target loss without a significant computational overhead. The new method is based on the assumption that the feature map can be used to estimate the tracking confidence more accurately. When the confidence is l...
Article
Web-based planetary image dissemination platforms usually show outline coverage of available data and offer querying for metadata as well as preview and download. While the usual form of presenting multiorbit data sets is to merge the data into a larger mosaic, for change detection purposes it is essential to maintain the individual images as an im...
Preprint
Full-text available
In this paper, we examine the real-time recovery of a video tracker from a target loss, using information that is already available from the original tracker and without a significant computational overhead. More specifically, before using the tracker output to update the target position we estimate the detection confidence. In the case of a low co...
Article
Full-text available
We present our research on compositional mapping of the Martian South Polar Residual Cap (SPRC), especially the detection of organic signatures within the dust content of the ice, based on hyperspectral data analysis. The SPRC is the main region of interest for this investigation, because of the unique CO2 ice sublimation features that cover the su...
Article
Change detection in satellite images is a key concern of the EarthObservation field for environmental and climate change monitoring. Satellite images also provide important clues to both the past and present surface conditions of other planets, which cannot be validated on the ground. With the volume of satellite imagery continuing to grow, the ina...
Article
Digital Terrain Model (DTM) creation is essential to improving our understanding of the formation processes of the Martian surface. Although there have been previous demonstrations of open-source or commercial planetary 3D reconstruction software, planetary scientists are still struggling with creating good quality DTMs that meet their science need...
Article
Full-text available
This work presents the coregistered, orthorecti�ed and mosaiced high-resolution products of the MC11 quadrangle of Mars, which have been processed using novel, fully automatic, techniques. We discuss the development of a pipeline that achieves fully automatic and parameter independent geometric alignment of high-resolution planetary images, startin...
Article
Full-text available
There has been a revolution in 3D surface imaging of Mars over the last 12 years with systematic stereoscopy from HRSC. Digital Terrain Models (DTMs) and OrthoRectified Images (ORIs) have been produced for almost 50% of the Martian surface. DLR, together with the HRSC science team, produced 3D HRSC mosaic products for large regions comprising aroun...
Conference Paper
Full-text available
The European FP7 iMars project has developed tools and 3D models of the Martian surface through the co-registration of NASA and ESA mission data dating from the Viking missions of the 1970s to the present day, for a much more comprehensive interpretation of the geomorphological and climatic processes that have taken and do take place. We present th...
Conference Paper
Full-text available
There has been a revolution in 3D surface imaging of Mars over the last 12 years with systematic stereoscopy from HRSC and the production for almost 50% of the Martian surface of DTMs and ORIs. The iMars project has been exploiting this unique set of 3D products as a basemap to co-register NASA imagery going back to the 1970s. DLR have produced 3D...
Conference Paper
The surface of Mars has been an object of interest for planetary research since the launch of Mariner 4 in 1964. Since then different cameras such as the Viking Visual Imaging Subsystem (VIS), Mars Global Surveyor (MGS) Mars Orbiter Camera (MOC), and Mars Reconnaissance Orbiter (MRO) Context Camera (CTX) and High Resolution Imaging Science Experime...
Conference Paper
Understanding planetary atmosphere-surface and extra-terrestrial-surface formation processes within our Solar System is one of the fundamental goals of planetary science research. There has been a revolution in planetary surface observations over the last 15 years, especially in 3D imaging of surface shape. This has led to the ability to be able to...
Conference Paper
Understanding planetary atmosphere-surface and extra-terrestrial-surface formation processes within our Solar System is one of the fundamental goals of planetary science research. There has been a revolution in planetary surface observations over the last 15 years, especially in 3D imaging of surface shape. This has led to the ability to be able to...
Article
Full-text available
Polycyclic aromatic hydrocarbons (PAHs) are considered to be important in theories of abiogenesis (Allamandola, 2011) . There is evidence that PAHs have been detected on two icy Saturnian satellites using the Visual and Infrared Mapping Spectrometer (VIMS) on the Cassini spacecraft (Cruikshank et al., 2007). The hypothesised presence of PAHs in Mar...
Article
Full-text available
Polycyclic aromatic hydrocarbons (PAHs) are considered to be important in theories of abiogenesis (Allamandola, 2011) . There is evidence that PAHs have been detected on two icy Saturnian satellites using the Visual and Infrared Mapping Spectrometer (VIMS) on the Cassini spacecraft (Cruikshank et al., 2007). The hypothesised presence of PAHs in Mar...
Article
Full-text available
Understanding planetary atmosphere-surface exchange and extra-terrestrial-surface formation processes within our Solar System is one of the fundamental goals of planetary science research. There has been a revolution in planetary surface observations over the last 15 years, especially in 3D imaging of surface shape. This has led to the ability to o...
Article
Full-text available
The surface of Mars has been an object of interest for planetary research since the launch of Mariner 4 in 1964. Since then different cameras such as the Viking Visual Imaging Subsystem (VIS), Mars Global Surveyor (MGS) Mars Orbiter Camera (MOC), and Mars Reconnaissance Orbiter (MRO) Context Camera (CTX) and High Resolution Imaging Science Experime...
Article
Full-text available
Four NASA missions over the last forty years with onboard instruments for high-resolution orbital imaging have achieved both global coverage (with 6m CTX, 20m THEMIS-VIS and >8m Viking Orbiter cameras) as well as imaging with very high resolution in specific regions of interest (e.g. 25cm HiRISE and ≈1.5-12m MOC-NA cameras). Overall, this set of ca...
Article
Full-text available
Four NASA missions over the last forty years with onboard instruments for high-resolution orbital imaging have achieved both global coverage (with 6m CTX, 20m THEMIS-VIS and >8m Viking Orbiter cameras) as well as imaging with very high resolution in specific regions of interest (e.g. 25cm HiRISE and ≈1.5-12m MOC-NA cameras). Overall, this set of ca...
Article
Full-text available
Understanding planetary atmosphere-surface exchange and extra-terrestrial-surface formation processes within our Solar System is one of the fundamental goals of planetary science research. There has been a revolution in planetary surface observations over the last 15 years, especially in 3D imaging of surface shape. This has led to the ability to o...
Article
Full-text available
The surface of Mars has been an object of interest for planetary research since the launch of Mariner 4 in 1964. Since then different cameras such as the Viking Visual Imaging Subsystem (VIS), Mars Global Surveyor (MGS) Mars Orbiter Camera (MOC), and Mars Reconnaissance Orbiter (MRO) Context Camera (CTX) and High Resolution Imaging Science Experime...
Article
Full-text available
Within the EU FP-7 iMars project, a fully automated multi-resolution DTM processing chain, called Co-registration ASP-Gotcha Optimised (CASP-GO) has been developed, based on the open source NASA Ames Stereo Pipeline (ASP). CASP-GO includes tiepoint based multi-resolution image co-registration and an adaptive least squares correlation-based sub-pixe...
Article
Full-text available
Within the EU FP-7 iMars project, a fully automated multi-resolution DTM processing chain, called Co-registration ASP-Gotcha Optimised (CASP-GO) has been developed, based on the open source NASA Ames Stereo Pipeline (ASP). CASP-GO includes tiepoint based multi-resolution image co-registration and an adaptive least squares correlation-based sub-pixe...
Conference Paper
Full-text available
Image analysis of the human eye has provided new insight into ocular disorders and has the potential to assist in their automated diagnosis. We herein report results from analysis of the processed fundus (retina) images from a) healthy subjects and patients diagnosed with b) diabetic retinopathy or c) glaucoma, by means of image fractal analysis an...
Article
Full-text available
Understanding the role of different solid surface formation processes within our Solar System is one of the fundamental goals of planetary science research. There has been a revolution in planetary surface observations over the last 8 years, especially in 3D imaging of surface shape (down to resolutions of 10s of cms) and subsequent terrain correct...
Article
Full-text available
This work deals with the meta-data analysis of high-resolution orbital imagery that was acquired over the last four decades of Mars. The objective of this analysis is to provide a starting point for planetary scientists that are interested in examining the martian surface in order to detect changes that are related to not fully understood natural p...
Conference Paper
We assess a processing system for generating multi-resolution digital terrain models (DTM) of Mars within the the iMars project of the European Seventh Framework Programme. This system is based on a non-rigorous sensor model algorithm developed jointly by University of Seoul and University College London for processing high-resolution stereoscopic...
Conference Paper
Since the 1970’s five NASA orbiter missions with high resolution imaging instruments onboard have reached Mars: Viking Orbiter 1 & 2, which were launched in August 1975, and acquired orbital images of the planet between 1976 and 1980 were the first to map the entire surface of the planet with resolution finer than 200 metres per pixel, reaching dow...
Conference Paper
The systematic orbital imaging of the Martian surface started 4 decades ago from NASA's Viking Orbiter 1 & 2 missions, which were launched in August 1975, and acquired orbital images of the planet between 1976 and 1980. The result of this reconnaissance was the first medium-resolution (i.e. ≤ 300m/pixel) global map of Mars, as well as a variety of...
Article
Full-text available
Understanding the role of different planetary surface formation processes within our Solar System is one of the fundamental goals of planetary science research. There has been a revolution in planetary surface observations over the last 8 years, especially in 3D imaging of surface shape (down to resolutions of 10cm) and subsequent terrain correctio...
Conference Paper
The 2018 ESA ExoMars Rover will search for signs of past martian life. We present and describe Aram Dorsum, one of four candidate rover landing sites.
Article
Full-text available
Introduction: The search for life on Mars is a cornerstone of international solar system exploration. In 2018, the European Space agency will launch the ExoMars Rover to further this goal. The key science objectives of the ExoMars Rover are to: 1) search for signs of past and present life on Mars; 2) investigate the water/geochemical environment as...
Conference Paper
Full-text available
Starting from Viking Orbiter 1 launched in August 1975, several robotic orbiters have been sent to Mars to map its surface. Recently, interest has shifted towards high resolution mapping, which allows the identification of previously undiscovered surface changes as well as greater examination of surface composition and geological history. The incre...
Conference Paper
A meta-data analysis has been performed of high-resolution imagery that have been acquired over the last four decades from Mars. More specifically, image footprints downloaded from the PDS Geosciences Node Mars Orbital Data Explorer (ODE) were used to group all Martian imagery based on Martian season that each image was acquired. This clustering al...
Conference Paper
We assess several algorithms for generating digital terrain models (DTM) of Mars using high-resolution stereoscopic images produced by the High Resolution Imaging Science Experiment (HiRISE) camera and Context Camera (CTX) onboard the NASA Mars Reconnaissance Orbiter (MRO) spacecraft. One of the algorithms was developed jointly by University of Seo...
Conference Paper
Understanding the role of different planetary surface formation processes within our Solar System is one of the fundamental goals of planetary science research. There has been a revolution in planetary surface observations over the last 15 years, especially in 3D imaging of surface shape. This has led to the ability to be able to overlay different...
Chapter
Full-text available
Since its introduction, television has been providing to millions of users a non-interactive experience, in which viewers can only participate as passive consumers of audiovisual content. Recently, the extreme proliferation and success of the Internet and the widespread appraisal of the interaction possibilities that it offers gave rise to the idea...
Article
Full-text available
In this work we deal with the problem of video concept detection, for the purpose of using the concept detection results towards more effective concept-based video retrieval. The key novelties of this work are: 1) The use of spatio-temporal video slices (tomographs) in the same way that visual keyframes are typically used in video concept detection...
Article
Full-text available
Mars.eu) have and are developing tools for Mars ex-ploration in preparation for the ESA-Roscosmos Exo-Mars Trace Gas Orbiter 2016 and rover mission in 2018. There are several common themes running throughout these developments including: a) the neces-sity to try to automate as much of the image processing as feasible so that future missions to Mars...
Article
Full-text available
A meta-data analysis has been performed of high-resolution imagery that have been acquired over the last four decades from Mars. More specifically, we are interested in two independent image parameters, the time that each image was acquired and the spatial resolution with which the planetary region is mapped in the image. We are only interested in...
Conference Paper
Full-text available
Understanding the role of different planetary surface formation processes within our Solar System is one of the fundamental goals of planetary science research. There has been a revolution in planetary surface observations over the last 7 years, especially in 3D imaging of surface shape (down to resolutions of 10cm) and subsequent terrain correctio...
Conference Paper
Full-text available
In this work we deal with the problem of video concept detection, for the purpose of using the detection results towards more effective concept-based video retrieval. In order to handle this task, we propose using spatio-temporal video slices, called video tomographs, in the same way that visual keyframes are typically used in traditional keyframe-...
Conference Paper
Full-text available
This paper provides an overview of the tasks submitted to TRECVID 2012 by ITI-CERTH. ITICERTH participated in the Known-item search (KIS), in the Semantic Indexing (SIN), as well as in the Event Detection in Internet Multimedia (MED) and the Multimedia Event Recounting (MER) tasks. In the SIN task, techniques are developed, which combine video repr...
Conference Paper
Full-text available
In this work the problem of how to evaluate video scene segmentation results is examined. The evaluation, which is typically conducted by comparison of the experimental output of scene segmentation algorithms with a ground-truth temporal decomposition, often suffers from ambiguity in the definitionof the ground truth. To alleviate this drawback the...
Chapter
Full-text available
This work deals with the problem of automatic temporal segmentation of a video into elemen-tary semantic units known as scenes. Its novelty lies in the use of high-level audio information, in the form of audio events, for the improvement of scene segmentation performance. More specifically, the proposed technique is built upon a recently proposed a...
Thesis
Full-text available
Video temporal decomposition is an essential element of a variety of video processing applications. The decomposition is traditionally conducted using shots as the video structural units. However, shots lack any semantic meaning. On the other hand, scenes, which are generally defined as the elementary semantic video units, are expected to generate...
Article
Full-text available
In this work a novel approach to evaluating video temporal decomposition algorithms is presented. The evaluation measures typically used to this end are non-linear combinations of Precision-Recall or Coverage-Overflow, which are not metrics and additionally possess undesirable properties, such as non-symmetricity. To alleviate these drawbacks we in...
Conference Paper
Full-text available
This paper provides an overview of the tasks submitted to TRECVID 2011 by ITI-CERTH. ITI-CERTH participated in the Known-item search (KIS) as well as in the Semantic Indexing (SIN) and the Event Detection in Internet Multimedia (MED) tasks. In the SIN task, techniques are developed, which combine motion information with existing well-performing des...
Conference Paper
Full-text available
In this work the contribution of automatically-extracted (thus, imperfect) video structural semantics towards improving interactive video retrieval is examined. First, the automatic extraction of video structural semantics, i.e. the decomposition of the video into scenes that correspond to the different sub-stories or high-level events, is performe...
Conference Paper
Full-text available
This work demonstrates a new approach to video temporal segmentation into scenes. The utilized technique is based on an audio-visual extension of the well-known method of the Scene Transition Graph (STG). This multi-modal extension exploits both low-and high-level audio-visual descriptors to construct distinct STGs. These STGs are employed into a p...
Conference Paper
Full-text available
This work examines the possibility of exploiting, for the purpose of video segmentation to scenes, semantic information coming from the analysis of the visual modality. This information, in contrast to the low-level visual features typically used in previous approaches, is obtained by application of trained visual concept detectors such as those de...
Conference Paper
Full-text available
This paper proposes a novel binary image descriptor, namely the Adaptive Hierarchical Density Histogram, that can be utilized for complex binary image retrieval. This novel descriptor exploits the distribution of the image points on a two-dimensional area. To reflect effectively this distribution, we propose an adaptive pyramidal decomposition of t...
Article
In this article, we discuss the potential benefits, the requirements and the challenges involved in patent image retrieval and subsequently, we propose a framework that encompasses advanced image analysis and indexing techniques to address the need for content-based patent image search and retrieval. The proposed framework involves the application...
Conference Paper
Full-text available
This work deals with the problem of automatic temporal segmentation of a video into elementary semantic units known as scenes. Its novelty lies in the use of high-level audio information in the form of audio events for the improvement of scene segmentation performance. More specifically, the proposed technique is built upon a recently proposed audi...