Goran Zajić - Academia.edu (original) (raw)

Papers by Goran Zajić

Physiological Measurement, Aug 11, 2016

Healthy versus unhealthy heart sound computer-aided classification tools are very popular for sup... more Healthy versus unhealthy heart sound computer-aided classification tools are very popular for supporting clinical decisions. In this paper a new method is proposed for the classification of heart sound recordings from a statistical standpoint without detection and localization of fundamental heart sounds (S1, S2). This study analyzes the possibility of detecting healthy heart sound signal from a large set of measurements, corresponding to different pathologies, such as aortic regurgitation, mitral regurgitation, aortic stenosis and ventricular septal defects. The proposed method employs singularity spectra analysis and long-term dependency of irregular structures. Healthy signals are firstly separated from the rest of the recordings. In the second step, the signals with a click syndrome, used here as a reference, are detected in the unhealthy group. Innocent murmurs have not been considered in this paper. Each auscultatory recording is classified into one of the following classes: healthy; click syndrome; and other heart dysfunctions. The results of the proposed method provided high recall and precision values for each of the three classes. Since the presence of additive noise may affect the classification, we also analyzed the possibility of classifying signals in such circumstances. The method was tested, verified and showed high accuracy.

Computational and Mathematical Methods in Medicine, 2013

Phonocardiography has shown a great potential for developing low-cost computer-aided diagnosis sy... more Phonocardiography has shown a great potential for developing low-cost computer-aided diagnosis systems for cardiovascular monitoring. So far, most of the work reported regarding cardiosignal analysis using multifractals is oriented towards heartbeat dynamics. This paper represents a step towards automatic detection of one of the most common pathological syndromes, socalled mitral valve prolapse (MVP), using phonocardiograms and multifractal analysis. Subtle features characteristic for MVP in phonocardiograms may be difficult to detect. The approach for revealing such features should be locally based rather than globally based. Nevertheless, if their appearances are specific and frequent, they can affect a multifractal spectrum. This has been the case in our experiment with the click syndrome. Totally, 117 pediatric phonocardiographic recordings (PCGs), 8 seconds long each, obtained from 117 patients were used for PMV automatic detection. We propose a two-step algorithm to distinguish PCGs that belong to children with healthy hearts and children with prolapsed mitral valves (PMVs). Obtained results show high accuracy of the method. We achieved 96.91% accuracy on the dataset (97 recordings). Additionally, 90% accuracy is achieved for the evaluation dataset (20 recordings). Content of the datasets is confirmed by the echocardiographic screening.

Medicinski časopis, 2022

Svetska zdravstvena organizacija (WHO) i Međunarodna federacija ginekologa i opstetičara (FIGO) t... more Svetska zdravstvena organizacija (WHO) i Međunarodna federacija ginekologa i opstetičara (FIGO) tumore jajnika klasifikuju na benigne tumore, tumore niskog malignog potencijala i maligne tumore (2). U našoj zemlji, prema podacima Registra za rak, karcinom jajnika zauzima peto mesto po učestalosti. patient, on the adequate treatment of ovarian tumors.

Timočki medicinski glasnik, 2021

Prema definiciji Svetske zdrastvene organizacije (WHO) o neplodonosti para govorimo onda kada se ... more Prema definiciji Svetske zdrastvene organizacije (WHO) o neplodonosti para govorimo onda kada se i pored redovnog nezaštićenog odnosa u toku jedne godine ne dogodi trudnoća. Jedan od razloga steriliteta su urođene i stečene anomalije materice. Najbolja vizuelizacija unutrašnjosti materice se radi histeroskopski. Histeroskopija je minimalno invazivna hirurška procedura koja omogućava dijagnostiku i otklanjanje urođenih, ali i stečenih patoloških promena u šupljini materice. Kod infertilnih pacijentkinja učestalost anomalija je veća i kreće se od 3-6% , a kod habitaulnih pobačaja 5-10%. Cilj rada je bio da se proceni uspeh operativne video-histeroskopije u tretmanu patoloških promena kavuma uterusa usled pojave mioma uterusa kod infertilnih pacijentkinja, na osnovu broja recidiva u prvih šest meseci posle operacije i broja intraoperativnih i postoperativnih komplikacija. MATERIJAL I METODE: Ispitivanjem je obuhvaćeno 200 infertilnih pacijentkinja do 40 godina starosti kod kojih je urađena operativna histeroskopija zbog dijagnostikovanih mioma u kavumu uterusa. Pacijentkinje su operisane na odeljenju steriliteta u Ginekološko-akušerskoj klinici "Narodni Front" u Beogradu. Od patoloških promena kavuma uterusa histeroskopski su odstranjeni submukozni miomi materice tip 0 i tip I. REZULTATI: Histeroskopski su najčešće resecirani submukozni miomi tip 0 i tip 1 veličine od 21-25 mm, koji su bili lokalizovani na zadnjem zidu korpusa uterusa. Ne postoji statistički značajna razlika u veličini submukoznih mioma između pacijentkinja sa primarnim sterilitetom i pacijentkinja sa sekundarnim sterilitetom (U=76,000; p>0.05). Nađena je visoko statistički značajna korelacija starosnih grupa i tipa steriliteta uz umerenu jačinu povezanosti varijabli (r=0,408; p<0,001), što nam ukazuje na značajno ćešću povezanost sekundarnog steriliteta i starije životne dobi pacijentkinja. Dužina trajanja steriliteta kod pacijentkinja sa primarnim sterilitetom je statistički značajno veća u odnosu na pacijentkinje sa sekundarnim sterilitetom, (U=3907,5; p<0.05). Ne postoji značajna razlika u odnosu na pojedine lokalizacije submukoznih mioma između pacijentkinja sa primarnim sterilitetom i pacijentkinja sa sekundarnim sterilitetom (U=76,500; p>0.05). Komplikacije histeroskopske miomektomije se dele na intraoperativne i postoperativne. Ukupno je bilo komplikacija kod 3 pacijentkinje. Desile su se 2 intraoperativne komplikacije(1%) tokom histeroskopske opercije: perforacija uterusa i krvarenje sa mesta resekcije Postoperativno je samo kod jedne pacijentkinje došlo do stvaranje adhezija posle operacije (0,5%). Ukupno je bilo 1,5% komplikacija kod operisanih pacijentkinja. Recidiv submukoznog mioma se javiio kod samo jedne (0,5%) pacijentkinja tokom šestomesečnog postoperativnog toka. ZAKLJUČAK Video-histeroskopija predstavlja sigurnu i efikasnu endoskopsku proceduru za dijagnozu i hiruško uklanjanje submukoznih mioma kao jednog od faktora patoloških stanja šupljine materice. Submukozni miomi koji deformišu šupljinu materice smanjuju fertilnu sposobnost žene. Procenat intraoperativnih komplikacija bio je mali samo 1% kao i postoperativnih komplikacija 0,5%. Procenat recidiva šest meseci od operacije je bio minimalan 0,5%. Prednosti histeroskopskog pristupa su kraće trajanje zahvata, bolja preglednost operativnog polja, veća preciznost zahvata, manja bolnost, manji morbiditet, izostanak rezova, brži oporavak i dostizanje radne sposobnosti.

Abstract The modern age is characterized by great professional and private multi-media production... more Abstract The modern age is characterized by great professional and private multi-media production of which the largest percentage are images. Search a large number of images from the user&amp;amp;amp;amp;amp;amp;amp;#x27;s perspective is almost always based on content. The aim of this work is to create a database search algorithm for images that is content based. Starting from the features matrix, algorithm is based on working with color, texture and shape. Using the five characteristics of color, four characteristics of texture, and one feature for shape, the initial ...

Abstract The influence of the feature vector (FV) content on the CBIR (content-based image retrie... more Abstract The influence of the feature vector (FV) content on the CBIR (content-based image retrieval) system efficiency was considered. By using two different FVs and applying three different learning methods, it was shown that the efficiency of retrieving depends on both the FV content and the learning method, independently.

HAL (Le Centre pour la Communication Scientifique Directe), Nov 1, 2006

In this paper, we give an overview of the four tasks submitted to TRECVID 2008 by COST292. The hi... more In this paper, we give an overview of the four tasks submitted to TRECVID 2008 by COST292. The high-level feature extraction framework comprises four systems. The first system transforms a set of low-level descriptors into the semantic space using Latent Semantic Analysis and utilises neural networks for feature detection. The second system uses a multi-modal classifier based on SVMs and several descriptors. The third system uses three image classifiers based on ant colony optimisation, particle swarm optimisation and a multi-objective learning algorithm. The fourth system uses a Gaussian model for singing detection and a person detection algorithm. The search task is based on an interactive retrieval application combining retrieval functionalities in various modalities with a user interface supporting automatic and interactive search over all queries submitted. The rushes task submission is based on a spectral clustering approach for removing similar scenes based on eigenvalues of frame similarity matrix and and a redundancy removal strategy which depends on semantic features extraction such as camera motion and faces. Finally, the submission to the copy detection task is conducted by two different systems. The first system consists of a video module and an audio module. The second system is based on mid-level features that are related to the temporal structure of videos.

A content-based image retrieval system with query image classification prior to retrieving proced... more A content-based image retrieval system with query image classification prior to retrieving procedure is proposed. Query image is compared to representative patterns of image classes, not to all images from database, accelerating thus initial retrieving step. Such procedure is possible when images from database are grouped into classes with similar content. Classification is performed using minor component (MC) analysis. Since it is expectable that MCs mainly depend on image details, not on an image background, this approach seems to be more efficient than classic CBIR. Minor components may be calculated by using single-layer neural network. The efficiency of proposed system is tested over images from Corel dataset. Index terms-Content-based image retrieval, relevance feedback, principal/minor component analysis, image classification, neural networks.

Page 1. Global Image Search vs. Regional Search in CBIR Systems Stevan Rudinac Marija Ušćumlić Fa... more Page 1. Global Image Search vs. Regional Search in CBIR Systems Stevan Rudinac Marija Ušćumlić Faculty of Electrical Enginering Univ. Belgrade, Serbia stevanrudinac@gmail.com muscumlic@yahoo.com Maja Rudinac ...

Medicinski casopis

Objective. We present a patient aged 55 years, with ovarian fibroma measuring 134x72x112 mm and t... more Objective. We present a patient aged 55 years, with ovarian fibroma measuring 134x72x112 mm and the associated fallopian tube measuring 55x9mm (previously analyzed ex tempore method). A fibroma is a benign ovarian tumor consisting of fibroblasts, collagen fibers and arising from the ovarian stroma. It accounts for 1% to 5% of all ovarian tumors and occurs in the population aged 20 to 80, most often after the age of 50. Methods. An ultrasonographic examination showed that the uterus swims in about 100 ml of anechoic free fluid. Anteuterine, a larger soft-tissue tumefaction, without a clear connection with the uterus, with clear borders and seemingly mobile, measures 106x80 mm. Following that, MSCT and MR of the abdomen and small pelvis were performed. After adequate preparation, the patient underwent a surgery, and the obtained material was sent for histopathological diagnosis. Results. The analyzed sample consisted of tumor proliferation of predominantly spindle cells organized by v...

2020 Zooming Innovation in Consumer Technologies Conference (ZINC)

Computer vision techniques implemented in modern vehicles should be designed to distinguish diffe... more Computer vision techniques implemented in modern vehicles should be designed to distinguish different changes in a video sequence, captured by RGB and RGBD cameras mounted in or out a vehicle. Autonomous driving process could improve safety of all passengers by introducing additional sensing. In this paper, we used input data from mentioned cameras acquired with inertial sensor for road roughness as a limiter of velocity. Abrupt changes of the velocity and driver comfort affects the driver’s head position. The head position is monitored using 3D skeleton model and depth information. The results show possibility of detection of the potential risk found for unusual driver behavior. Then, the human control could be taken by safety application and system.

2020 28th Telecommunications Forum (TELFOR), 2020

There has been considerable effort in designing new video coding and compression standards. The A... more There has been considerable effort in designing new video coding and compression standards. The AV1 is one of the most recent formats valuable for OTT video services and internet delivery. In this paper SVT-AV1 is analyzed according to standard statistical techniques. Moreover, it is compared to VP9 and HEVC video formats in the case of 4k video delivery. The comparison is made for three video sequences from PSNR and time coding standpoint. The results show advantages of SVT-AV1 format.

2018 14th Symposium on Neural Networks and Applications (NEUREL), 2018

This paper analyzes possibilities to integrate scanning devices and methods of diagnostics road c... more This paper analyzes possibilities to integrate scanning devices and methods of diagnostics road conditions, into the more global network such is the Internet of Vehicles, connecting vehicles and/or stationary access points. Moreover, initial estimation of real time processing effectivity, based on the interaction of diagnostic equipment with the appropriate road surface material, is considered.

Multimedia Tools and Applications, 2018

Efficient management of video sequences is based on adequate video content description. This desc... more Efficient management of video sequences is based on adequate video content description. This description can be used for various purposes in different applications, telecommunication services, video and multimedia systems. Video hard cut detection represents the foundation of temporal video segmentation. In this paper, a new video hard cut detection methodology is proposed using multifractal features. Transition between two shots can be described as color and texture differences within a decoded video sequence. In the proposed methodology we formed specific structures by measuring color differences between frames. The formed structures are used for hard cut candidate detection. This is followed by multifractal representation of texture changes by Hölder exponents. The proposed methodology achieves high performance using more than 750,000 frames, extracted from forty different video sequences, classified by four well known genre groups. Moreover, the proposed hard cut detection achieves high performance regardless of high level video production or complex non-linear editing for different genre groups. This is confirmed by comparison between the proposed methodology and other recent work on hard cut detection.

2016 13th Symposium on Neural Networks and Applications (NEUREL), 2016

In this paper Content Base Image Retrieval (CBIR) system with relevance feedback is presented, wh... more In this paper Content Base Image Retrieval (CBIR) system with relevance feedback is presented, where image database search is performed using singularity strength (Holder exponent). Images in database are described with low-level features for color and texture, which are concatenated in feature vectors (FV). Relevance feedback is implemented in CBIR system employing the artificial intelligence based on Radial Bases Functions. Feature vectors are refined using local singularities. Search efficiency of the CBIR system is compared for direct and refined feature vector. CBIR system is tested on Corel 1000 image database. The test results showed that CBIR system using local singularities has high search efficiency, and the search results are comparable with the results for direct feature vectors.

2016 13th Symposium on Neural Networks and Applications (NEUREL), 2016

In this paper we analyze video traffic variability taking into account H.265/HEVC video encoding.... more In this paper we analyze video traffic variability taking into account H.265/HEVC video encoding. In particular, we examine the video trace variability considering high definition and ultra high definition (4k) formats. The preliminary results show the frame based differentiation between the new and common formats. The obtained results seem promising and can be considered useful for video traffic monitoring and next-generation network support.

2015 23rd Telecommunications Forum Telfor (TELFOR), 2015

This paper presents a new algorithm for detection of abrupt shot changes in video sequence based ... more This paper presents a new algorithm for detection of abrupt shot changes in video sequence based on Multifractal Analysis (MA). Video sequence is decoded into series of frames, and low level feature for texture and color are extracted from each of them. Content difference between adjacent frames is calculated based on correlation of frame low level features. MA is used for description of content difference arrays in multifractal singularity domain (α - domain) with emphasized local singularities. The detection of abrupt changes is realized using fixed thresholds technique and singularity shape analysis. Proposed algorithm is tested on more than 600 000 frames, with high efficiency of detection.

2015 23rd Telecommunications Forum Telfor (TELFOR), 2015

In this paper we perform animation frame analysis. The animation frame/video content is usually c... more In this paper we perform animation frame analysis. The animation frame/video content is usually characterized by specific color distribution and texture within frames. The existence of animation within frame can be described by carefully selected features. Furthermore, frames can be partially animated. Here, we present our initial results related to automatic animation interpretation using different types of video sequences.

2008 9th Symposium on Neural Network Applications in Electrical Engineering, 2008

Abstract This paper describes a content-based image retrieval (CBIR) system which makes use of bo... more Abstract This paper describes a content-based image retrieval (CBIR) system which makes use of both labeled images, annotated by the user, and unlabeled images available in the database. The system initially retrieves images objectively closest to the query image. The user then subjectively labels retrieved images as relevant or irrelevant. Although such relevance feedback from the user is an effective way of bridging the semantic gap between objective and subjective similarity, it is also very time consuming, requiring huge human ...

Physiological Measurement, Aug 11, 2016

Healthy versus unhealthy heart sound computer-aided classification tools are very popular for sup... more Healthy versus unhealthy heart sound computer-aided classification tools are very popular for supporting clinical decisions. In this paper a new method is proposed for the classification of heart sound recordings from a statistical standpoint without detection and localization of fundamental heart sounds (S1, S2). This study analyzes the possibility of detecting healthy heart sound signal from a large set of measurements, corresponding to different pathologies, such as aortic regurgitation, mitral regurgitation, aortic stenosis and ventricular septal defects. The proposed method employs singularity spectra analysis and long-term dependency of irregular structures. Healthy signals are firstly separated from the rest of the recordings. In the second step, the signals with a click syndrome, used here as a reference, are detected in the unhealthy group. Innocent murmurs have not been considered in this paper. Each auscultatory recording is classified into one of the following classes: healthy; click syndrome; and other heart dysfunctions. The results of the proposed method provided high recall and precision values for each of the three classes. Since the presence of additive noise may affect the classification, we also analyzed the possibility of classifying signals in such circumstances. The method was tested, verified and showed high accuracy.

Computational and Mathematical Methods in Medicine, 2013

Phonocardiography has shown a great potential for developing low-cost computer-aided diagnosis sy... more Phonocardiography has shown a great potential for developing low-cost computer-aided diagnosis systems for cardiovascular monitoring. So far, most of the work reported regarding cardiosignal analysis using multifractals is oriented towards heartbeat dynamics. This paper represents a step towards automatic detection of one of the most common pathological syndromes, socalled mitral valve prolapse (MVP), using phonocardiograms and multifractal analysis. Subtle features characteristic for MVP in phonocardiograms may be difficult to detect. The approach for revealing such features should be locally based rather than globally based. Nevertheless, if their appearances are specific and frequent, they can affect a multifractal spectrum. This has been the case in our experiment with the click syndrome. Totally, 117 pediatric phonocardiographic recordings (PCGs), 8 seconds long each, obtained from 117 patients were used for PMV automatic detection. We propose a two-step algorithm to distinguish PCGs that belong to children with healthy hearts and children with prolapsed mitral valves (PMVs). Obtained results show high accuracy of the method. We achieved 96.91% accuracy on the dataset (97 recordings). Additionally, 90% accuracy is achieved for the evaluation dataset (20 recordings). Content of the datasets is confirmed by the echocardiographic screening.

Medicinski časopis, 2022

Svetska zdravstvena organizacija (WHO) i Međunarodna federacija ginekologa i opstetičara (FIGO) t... more Svetska zdravstvena organizacija (WHO) i Međunarodna federacija ginekologa i opstetičara (FIGO) tumore jajnika klasifikuju na benigne tumore, tumore niskog malignog potencijala i maligne tumore (2). U našoj zemlji, prema podacima Registra za rak, karcinom jajnika zauzima peto mesto po učestalosti. patient, on the adequate treatment of ovarian tumors.

Timočki medicinski glasnik, 2021

Prema definiciji Svetske zdrastvene organizacije (WHO) o neplodonosti para govorimo onda kada se ... more Prema definiciji Svetske zdrastvene organizacije (WHO) o neplodonosti para govorimo onda kada se i pored redovnog nezaštićenog odnosa u toku jedne godine ne dogodi trudnoća. Jedan od razloga steriliteta su urođene i stečene anomalije materice. Najbolja vizuelizacija unutrašnjosti materice se radi histeroskopski. Histeroskopija je minimalno invazivna hirurška procedura koja omogućava dijagnostiku i otklanjanje urođenih, ali i stečenih patoloških promena u šupljini materice. Kod infertilnih pacijentkinja učestalost anomalija je veća i kreće se od 3-6% , a kod habitaulnih pobačaja 5-10%. Cilj rada je bio da se proceni uspeh operativne video-histeroskopije u tretmanu patoloških promena kavuma uterusa usled pojave mioma uterusa kod infertilnih pacijentkinja, na osnovu broja recidiva u prvih šest meseci posle operacije i broja intraoperativnih i postoperativnih komplikacija. MATERIJAL I METODE: Ispitivanjem je obuhvaćeno 200 infertilnih pacijentkinja do 40 godina starosti kod kojih je urađena operativna histeroskopija zbog dijagnostikovanih mioma u kavumu uterusa. Pacijentkinje su operisane na odeljenju steriliteta u Ginekološko-akušerskoj klinici "Narodni Front" u Beogradu. Od patoloških promena kavuma uterusa histeroskopski su odstranjeni submukozni miomi materice tip 0 i tip I. REZULTATI: Histeroskopski su najčešće resecirani submukozni miomi tip 0 i tip 1 veličine od 21-25 mm, koji su bili lokalizovani na zadnjem zidu korpusa uterusa. Ne postoji statistički značajna razlika u veličini submukoznih mioma između pacijentkinja sa primarnim sterilitetom i pacijentkinja sa sekundarnim sterilitetom (U=76,000; p>0.05). Nađena je visoko statistički značajna korelacija starosnih grupa i tipa steriliteta uz umerenu jačinu povezanosti varijabli (r=0,408; p<0,001), što nam ukazuje na značajno ćešću povezanost sekundarnog steriliteta i starije životne dobi pacijentkinja. Dužina trajanja steriliteta kod pacijentkinja sa primarnim sterilitetom je statistički značajno veća u odnosu na pacijentkinje sa sekundarnim sterilitetom, (U=3907,5; p<0.05). Ne postoji značajna razlika u odnosu na pojedine lokalizacije submukoznih mioma između pacijentkinja sa primarnim sterilitetom i pacijentkinja sa sekundarnim sterilitetom (U=76,500; p>0.05). Komplikacije histeroskopske miomektomije se dele na intraoperativne i postoperativne. Ukupno je bilo komplikacija kod 3 pacijentkinje. Desile su se 2 intraoperativne komplikacije(1%) tokom histeroskopske opercije: perforacija uterusa i krvarenje sa mesta resekcije Postoperativno je samo kod jedne pacijentkinje došlo do stvaranje adhezija posle operacije (0,5%). Ukupno je bilo 1,5% komplikacija kod operisanih pacijentkinja. Recidiv submukoznog mioma se javiio kod samo jedne (0,5%) pacijentkinja tokom šestomesečnog postoperativnog toka. ZAKLJUČAK Video-histeroskopija predstavlja sigurnu i efikasnu endoskopsku proceduru za dijagnozu i hiruško uklanjanje submukoznih mioma kao jednog od faktora patoloških stanja šupljine materice. Submukozni miomi koji deformišu šupljinu materice smanjuju fertilnu sposobnost žene. Procenat intraoperativnih komplikacija bio je mali samo 1% kao i postoperativnih komplikacija 0,5%. Procenat recidiva šest meseci od operacije je bio minimalan 0,5%. Prednosti histeroskopskog pristupa su kraće trajanje zahvata, bolja preglednost operativnog polja, veća preciznost zahvata, manja bolnost, manji morbiditet, izostanak rezova, brži oporavak i dostizanje radne sposobnosti.

Abstract The modern age is characterized by great professional and private multi-media production... more Abstract The modern age is characterized by great professional and private multi-media production of which the largest percentage are images. Search a large number of images from the user&amp;amp;amp;amp;amp;amp;amp;#x27;s perspective is almost always based on content. The aim of this work is to create a database search algorithm for images that is content based. Starting from the features matrix, algorithm is based on working with color, texture and shape. Using the five characteristics of color, four characteristics of texture, and one feature for shape, the initial ...

Abstract The influence of the feature vector (FV) content on the CBIR (content-based image retrie... more Abstract The influence of the feature vector (FV) content on the CBIR (content-based image retrieval) system efficiency was considered. By using two different FVs and applying three different learning methods, it was shown that the efficiency of retrieving depends on both the FV content and the learning method, independently.

HAL (Le Centre pour la Communication Scientifique Directe), Nov 1, 2006

In this paper, we give an overview of the four tasks submitted to TRECVID 2008 by COST292. The hi... more In this paper, we give an overview of the four tasks submitted to TRECVID 2008 by COST292. The high-level feature extraction framework comprises four systems. The first system transforms a set of low-level descriptors into the semantic space using Latent Semantic Analysis and utilises neural networks for feature detection. The second system uses a multi-modal classifier based on SVMs and several descriptors. The third system uses three image classifiers based on ant colony optimisation, particle swarm optimisation and a multi-objective learning algorithm. The fourth system uses a Gaussian model for singing detection and a person detection algorithm. The search task is based on an interactive retrieval application combining retrieval functionalities in various modalities with a user interface supporting automatic and interactive search over all queries submitted. The rushes task submission is based on a spectral clustering approach for removing similar scenes based on eigenvalues of frame similarity matrix and and a redundancy removal strategy which depends on semantic features extraction such as camera motion and faces. Finally, the submission to the copy detection task is conducted by two different systems. The first system consists of a video module and an audio module. The second system is based on mid-level features that are related to the temporal structure of videos.

A content-based image retrieval system with query image classification prior to retrieving proced... more A content-based image retrieval system with query image classification prior to retrieving procedure is proposed. Query image is compared to representative patterns of image classes, not to all images from database, accelerating thus initial retrieving step. Such procedure is possible when images from database are grouped into classes with similar content. Classification is performed using minor component (MC) analysis. Since it is expectable that MCs mainly depend on image details, not on an image background, this approach seems to be more efficient than classic CBIR. Minor components may be calculated by using single-layer neural network. The efficiency of proposed system is tested over images from Corel dataset. Index terms-Content-based image retrieval, relevance feedback, principal/minor component analysis, image classification, neural networks.

Page 1. Global Image Search vs. Regional Search in CBIR Systems Stevan Rudinac Marija Ušćumlić Fa... more Page 1. Global Image Search vs. Regional Search in CBIR Systems Stevan Rudinac Marija Ušćumlić Faculty of Electrical Enginering Univ. Belgrade, Serbia stevanrudinac@gmail.com muscumlic@yahoo.com Maja Rudinac ...

Medicinski casopis

Objective. We present a patient aged 55 years, with ovarian fibroma measuring 134x72x112 mm and t... more Objective. We present a patient aged 55 years, with ovarian fibroma measuring 134x72x112 mm and the associated fallopian tube measuring 55x9mm (previously analyzed ex tempore method). A fibroma is a benign ovarian tumor consisting of fibroblasts, collagen fibers and arising from the ovarian stroma. It accounts for 1% to 5% of all ovarian tumors and occurs in the population aged 20 to 80, most often after the age of 50. Methods. An ultrasonographic examination showed that the uterus swims in about 100 ml of anechoic free fluid. Anteuterine, a larger soft-tissue tumefaction, without a clear connection with the uterus, with clear borders and seemingly mobile, measures 106x80 mm. Following that, MSCT and MR of the abdomen and small pelvis were performed. After adequate preparation, the patient underwent a surgery, and the obtained material was sent for histopathological diagnosis. Results. The analyzed sample consisted of tumor proliferation of predominantly spindle cells organized by v...

2020 Zooming Innovation in Consumer Technologies Conference (ZINC)

Computer vision techniques implemented in modern vehicles should be designed to distinguish diffe... more Computer vision techniques implemented in modern vehicles should be designed to distinguish different changes in a video sequence, captured by RGB and RGBD cameras mounted in or out a vehicle. Autonomous driving process could improve safety of all passengers by introducing additional sensing. In this paper, we used input data from mentioned cameras acquired with inertial sensor for road roughness as a limiter of velocity. Abrupt changes of the velocity and driver comfort affects the driver’s head position. The head position is monitored using 3D skeleton model and depth information. The results show possibility of detection of the potential risk found for unusual driver behavior. Then, the human control could be taken by safety application and system.

2020 28th Telecommunications Forum (TELFOR), 2020

There has been considerable effort in designing new video coding and compression standards. The A... more There has been considerable effort in designing new video coding and compression standards. The AV1 is one of the most recent formats valuable for OTT video services and internet delivery. In this paper SVT-AV1 is analyzed according to standard statistical techniques. Moreover, it is compared to VP9 and HEVC video formats in the case of 4k video delivery. The comparison is made for three video sequences from PSNR and time coding standpoint. The results show advantages of SVT-AV1 format.

2018 14th Symposium on Neural Networks and Applications (NEUREL), 2018

This paper analyzes possibilities to integrate scanning devices and methods of diagnostics road c... more This paper analyzes possibilities to integrate scanning devices and methods of diagnostics road conditions, into the more global network such is the Internet of Vehicles, connecting vehicles and/or stationary access points. Moreover, initial estimation of real time processing effectivity, based on the interaction of diagnostic equipment with the appropriate road surface material, is considered.

Multimedia Tools and Applications, 2018

Efficient management of video sequences is based on adequate video content description. This desc... more Efficient management of video sequences is based on adequate video content description. This description can be used for various purposes in different applications, telecommunication services, video and multimedia systems. Video hard cut detection represents the foundation of temporal video segmentation. In this paper, a new video hard cut detection methodology is proposed using multifractal features. Transition between two shots can be described as color and texture differences within a decoded video sequence. In the proposed methodology we formed specific structures by measuring color differences between frames. The formed structures are used for hard cut candidate detection. This is followed by multifractal representation of texture changes by Hölder exponents. The proposed methodology achieves high performance using more than 750,000 frames, extracted from forty different video sequences, classified by four well known genre groups. Moreover, the proposed hard cut detection achieves high performance regardless of high level video production or complex non-linear editing for different genre groups. This is confirmed by comparison between the proposed methodology and other recent work on hard cut detection.

2016 13th Symposium on Neural Networks and Applications (NEUREL), 2016

In this paper Content Base Image Retrieval (CBIR) system with relevance feedback is presented, wh... more In this paper Content Base Image Retrieval (CBIR) system with relevance feedback is presented, where image database search is performed using singularity strength (Holder exponent). Images in database are described with low-level features for color and texture, which are concatenated in feature vectors (FV). Relevance feedback is implemented in CBIR system employing the artificial intelligence based on Radial Bases Functions. Feature vectors are refined using local singularities. Search efficiency of the CBIR system is compared for direct and refined feature vector. CBIR system is tested on Corel 1000 image database. The test results showed that CBIR system using local singularities has high search efficiency, and the search results are comparable with the results for direct feature vectors.

2016 13th Symposium on Neural Networks and Applications (NEUREL), 2016

In this paper we analyze video traffic variability taking into account H.265/HEVC video encoding.... more In this paper we analyze video traffic variability taking into account H.265/HEVC video encoding. In particular, we examine the video trace variability considering high definition and ultra high definition (4k) formats. The preliminary results show the frame based differentiation between the new and common formats. The obtained results seem promising and can be considered useful for video traffic monitoring and next-generation network support.

2015 23rd Telecommunications Forum Telfor (TELFOR), 2015

This paper presents a new algorithm for detection of abrupt shot changes in video sequence based ... more This paper presents a new algorithm for detection of abrupt shot changes in video sequence based on Multifractal Analysis (MA). Video sequence is decoded into series of frames, and low level feature for texture and color are extracted from each of them. Content difference between adjacent frames is calculated based on correlation of frame low level features. MA is used for description of content difference arrays in multifractal singularity domain (α - domain) with emphasized local singularities. The detection of abrupt changes is realized using fixed thresholds technique and singularity shape analysis. Proposed algorithm is tested on more than 600 000 frames, with high efficiency of detection.

2015 23rd Telecommunications Forum Telfor (TELFOR), 2015

In this paper we perform animation frame analysis. The animation frame/video content is usually c... more In this paper we perform animation frame analysis. The animation frame/video content is usually characterized by specific color distribution and texture within frames. The existence of animation within frame can be described by carefully selected features. Furthermore, frames can be partially animated. Here, we present our initial results related to automatic animation interpretation using different types of video sequences.

2008 9th Symposium on Neural Network Applications in Electrical Engineering, 2008

Abstract This paper describes a content-based image retrieval (CBIR) system which makes use of bo... more Abstract This paper describes a content-based image retrieval (CBIR) system which makes use of both labeled images, annotated by the user, and unlabeled images available in the database. The system initially retrieves images objectively closest to the query image. The user then subjectively labels retrieved images as relevant or irrelevant. Although such relevance feedback from the user is an effective way of bridging the semantic gap between objective and subjective similarity, it is also very time consuming, requiring huge human ...