Sonja Grgic

Followers

Following

Public Views

Interests

Uploads

Papers by Sonja Grgic

Classification of image degradation using Riesz transform

This paper presents new method for classification of type of image degradation based on the Riesz... more This paper presents new method for classification of type of image degradation based on the Riesz transform and BRISQUE no-reference quality measure. Riesz transform has great properties and it can be used in many applications. Some of its benefits are: the ability to construct family of steerable wavelets with arbitrary order and any number of dimensions and it can bring the algorithm of filter banks with perfect reconstruction and also go to dimensions higher than two. Statistical properties of MSCN coefficients used by BRISQUE change in presence of distortion and by quantifying this changes with features calculated by using GGD and AGGD model the class of distortion can be determined. We calculated 18 statistical features out of spatial coefficients defined by BRISQUE measure and 19 parameters out of Riesz coefficients to get 37 features in total and then used features as input in SVM regressor in order to identify the type of image degradation. Then, we compared new method with BRISQUE method by using McNemar's statistical test to show statistical significance of our method.

Evaluation of Blur and Gaussian Noise Degradation in Images Using Statistical Model of Natural Scene and Perceptual Image Quality Measure

Radioengineering, Dec 14, 2017

In this paper we present a new method for classification of image degradation type based on Riesz... more In this paper we present a new method for classification of image degradation type based on Riesz transform coefficients and Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE) that employs spatial coefficients. In our method we use additional statistical parameters that give statistically better results for blur and all tested degradations together in comparison with previous methods. A new method to determine level of blur and Gaussian noise degradation in images using statistical model of natural scene is presented. We defined parameters for evaluation of level of Gaussian noise and blur degradation in images. In real world applications reference image is usually not available therefore the proposed method enables classification of image degradation by type and estimation of Gaussian noise and blur levels for any degraded image.

Quality of chest X-ray images compressed using algorhytms JPEG and JPEG2000

Lijec̆nic̆ki vjesnik, 2006

Choice of video coding standard for future DVB-T2 networks in Croatia

This paper presents current and future activities that are an integral part of the decision on vi... more This paper presents current and future activities that are an integral part of the decision on video coding standard for future DVB-T2 networks in Croatia. The aim of this paper is to present the results of two questionnaires about video quality and DVB-T2 products and technologies. The video quality questionnaire was aimed at current free-to-air broadcasters in Croatia, while the DVB-T2 products and technologies questionnaire was aimed at receiver manufacturers and distributors that are present on the Croatian market. H.265/HEVC versus H.264/AVC performance analysis for High-Definition Video Systems and the questionnaire results show that H.265/HEVC is the best choice for future DVB-T2 networks in Croatia. The paper describes future promotion activities and a revision of the Recommendation on minimum receiver technical requirements for the reception of DVB-T and DVB-T2 signal in the Republic of Croatia. Additionally, a status of the current and future preparation activities for the allocation of the second digital dividend band in Croatia are given.

Annotated Lung CT Image Database

Image Quality Images Compressed Using JPEG and JPEG2000 Algorhytms

Download

Real-Time Detection and Recognition of Cards in the Game of Set

2021 International Symposium ELMAR, 2021

This paper aims to present a procedure for real-time detection and recognition of the cards from ... more This paper aims to present a procedure for real-time detection and recognition of the cards from the card game Set using computer vision techniques. Our approach to recognizing cards in images is done in three steps: segmenting cards from the image, extracting features from card images with horizontal and vertical lines, and card feature classification with support vector machine (SVM). Small number of features are extracted from relatively small subset of pixels to achieve real-time processing.

Genre classification of paintings

2016 International Symposium ELMAR, 2016

Extensive digitization efforts in the recent years have led to a large increase of digitized and ... more Extensive digitization efforts in the recent years have led to a large increase of digitized and online available fine-art collections. With digitization of artworks, we aim to preserve all those valuable evidences of various human creative expressions, as well as make them available to a broader audience. The digitalization process of artworks should not constrain only to fulfilling the purpose of preservation, but also serve as a starting point for exploring of this type of data in a novel way, which is made possible with the rise of new achievements in computer vision. In the domain of computer analysis of visual art there are various ongoing research challenges. In this paper, we explore different image feature extraction methods and their applicability in the task of classifying painting by genre. Our dataset includes paintings of various styles grouped in seven genre categories. We achieved an accuracy of 77.57% for the task of genre classification. We concluded that the best performance is achieved when using features derived from a pretrained deep convolutional neural network.

Download

New image quality measure based on wavelets

Proceedings of SPIE, Feb 5, 2009

... Emil Dumic, Sonja Grgic, Mislav Grgic. Abstract. ... However, there is no current standard an... more

Estimation of symbol time offset using cyclic prefix in OFDM system

arXiv (Cornell University), May 4, 2023

In this paper synchronization techniques using cyclic prefix (CP) are analyzed to remove the infl... more In this paper synchronization techniques using cyclic prefix (CP) are analyzed to remove the influence of symbol time offset (STO) for correct synchronization in Orthogonal Frequency Division Multiplex (OFDM) system. For correct detection of STO using CP two techniques are used. The first one is based on finding the maximum correlation between two blocks and the second one on finding the similarity between two blocks which is maximized when the difference between them is minimized. Two cases are observed. The first one uses only additive white Gaussian noise (AWGN), while the second also uses the channel impulse response (CIR) of Rayleigh channel. When channel effect is added both performances of estimation by correlation and difference deteriorate if sufficient level of signal to noise (SNR) and proper length of CP are not provided.

Download

Comparison of Metrics for Colorized Image Quality Evaluation

2022 International Symposium ELMAR

An analysis of the ATM transmission for video signal contribution and distribution

Razvoj digitalne televizije i telekomunikcijskih mreža izaziva velike promjene, ne samo u radiodi... more Razvoj digitalne televizije i telekomunikcijskih mreža izaziva velike promjene, ne samo u radiodifuziji i distribuciji TV signala, vec i u videoprodukciji, mrežama za kontribuciju i mrežama za distribuciju signala unutar TV kuce. Pojava PDH (Plesiochronous Digital Hierarchy) i SDH/SONET (Synchronous Digital Hierarchy/Synchronous Optical Network) telekomunikacijskih mreža je potaknula i razvoj normi za prijenos digitalnog TV signala tim mrežama, a rezultat su dvije norme: CCIR preporuka 721 za prijenos komponentnog digitalnog videosignala za kontribucijske primjene pri brzinama prijenosa oko 140 Mb/s i CCIR preporuka 723 za prijenos komponentnog digitalnog videosignala za kontribucijske primjene pri brzinama prijenosa 34-45 Mb/s, koja je danas u Europi prihvacena kao ETSI norma. Brzine prijenosa propisane ovim normama su uvjetovane propisanim brzinama prijenosa u telekomunikacijskoj mreži. Danas se u kontribucijske svrhe koriste i MPEG-2 signali uz brzine prijenosa od 18-50 Mb/s. Za ...

JPEG image tampering detection using blocking artifacts

2012 19th International Conference on Systems, Signals and Image Processing (IWSSIP), 2012

Image forgery is nowadays widely used as digital images are easy to manipulate due to high availa... more Image forgery is nowadays widely used as digital images are easy to manipulate due to high availability of powerful image processing tools. It is possible to add or remove objects from an image without leaving any visible traces of tampering. This paper describes a method for detection of copy-paste manipulation on JPEG digital images. It is a type of image forgery in which a part of the image is copied to another location in the image with the intent to cover or add an important image object. The detection method was implemented through extracting and analyzing blocking artifact grids (BAGs), introduced by block processing during JPEG compression. Analysis was based on fact that BAGs usually mismatch after performing copy-paste operations. Proposed method was demonstrated on two doctored images.

Download

A Deep Learning Perspective on Beauty, Sentiment, and Remembrance of Art

IEEE Access, 2019

With the emergence of large digitized fine art collections and the successful performance of deep... more With the emergence of large digitized fine art collections and the successful performance of deep learning techniques, new research prospects unfold in the intersection of artificial intelligence and art. In order to explore the applicability of deep learning techniques in understanding art images beyond object recognition and classification, we employ convolutional neural networks (CNN) to predict scores related to three subjective aspects of human perception: aesthetic evaluation of the image, sentiment evoked by the image, and memorability of the image. For each concept, we evaluate several different CNN models trained on various natural image datasets and select the best performing model based on the qualitative results and the comparison with existing subjective ratings of artworks. Furthermore, we employ different decision treebased machine learning models to analyze the relative importance of various image features related to the content, composition, and color in determining image aesthetics, visual sentiment, and memorability scores. Our findings suggest that content and image lighting have significant influence on aesthetics, in which color vividness and harmony strongly influence sentiment prediction, while object emphasis has a high impact on memorability. In addition, we explore the predicted aesthetic, sentiment, and memorability scores in the context of art history by analyzing their distribution in regard to different artistic styles, genres, artists, and centuries. The presented approach enables new ways of exploring fine art collections based on highly subjective aspects of art, as well as represents one step forward toward bridging the gap between traditional formal analysis and the computational analysis of fine art. INDEX TERMS Convolutional neural networks, image aesthetics, image memorability, fine art, visual sentiment.

Download

3D video subjective quality: a new database and grade comparison study

Multimedia Tools and Applications, 2016

This paper presents a research study on the subjective assessment of 3D video quality using a new... more This paper presents a research study on the subjective assessment of 3D video quality using a newly constructed 3D video database (3DVCL@FER). This database consists of 8 original 3D video sequences, each degraded with 22 different degradation types, including degradations specific to stereoscopic systems. The subjective assessment was done with the support of a purpose-built easily customizable grade collection platform and conducted in two research laboratories, in Croatia and Portugal. Subjective scores for quality, depth and comfort were collected and DMOS (Difference Mean Opinion Score) values were calculated. Different objective measures (for image, 3D image, 2D video and 3D video) were separately compared with DMOS values for quality, depth and comfort. The 3D video grade-annotated database described is publicly accessible and can be used in research-related activities like assessment of existing objective measures, using the entire database or parts of it, and construction of new objective measures specific to 3D video degradations. The system presented can also be used to collect and compare subjective quality grades originating from different sites to study the effect of different observation conditions and observer/graders populations on the DMOS quality values for 3D video depth and comfort.

Download

Workshop: The electronic communications regulatory challenges — Experiences from the looking to the future project Croatian approach to digital dividend

Transition from analogue to digital television in Croatia has completed. This makes digital divid... more Transition from analogue to digital television in Croatia has completed. This makes digital dividend spectrum available for use by new services and technologies. Due to propagation characteristics of this part of the radiofrequency spectrum and economies of scale, this calls for a harmonised approach on a regional level. Use of different technologies and different periods of transition to digital television of neighbouring countries, limit the use of this part of the spectrum in Croatia. In this paper, transition from analogue to digital television is explained, possible usages of digital dividend spectrum and regulatory challenges in Croatia are investigated. Auction is the mechanism of assignment of digital dividend and its parameters still have to be defined.

Overall Architecture of Color Management System

Adaptive modulation in microwave networks

ELMAR 2007, 2007

Since different types/levels of modulation have different spectral efficiency, gain of adaptive m... more Since different types/levels of modulation have different spectral efficiency, gain of adaptive modulation has to be considered due to number of bits that are transmitted. We used model that has properties of white noise with probability of level occurrence according to normal Gaussian distribution. In order to perform more exact calculation of the gain of adaptive modulation further research has to be carried out (because of channel properties, properties of logic device that measures BER level and decides between certain modulation levels or types).

Learning the Principles of Art History with convolutional neural networks

Pattern Recognition Letters, 2019

Understanding the historical transformation of artistic styles implies the recognition of differe... more Understanding the historical transformation of artistic styles implies the recognition of different stylistic properties. From a computer vision perspective, stylistic properties represent complex image features. In our work we explore the use of convolutional neural networks for learning features that are relevant for understanding properties of artistic styles. We focus on stylistic properties described by Heinrich Wölfflin in his book Principles of Art History (1915). Wölfflin identified five key visual principles, each defined by two contrasting concepts. We refer to each principle as one high-level image feature that measures how much each of the contrasting concepts is present in an image. We introduce convolutional neural network regression models trained to predict values of the five Wölfflin's features. We provide quantitative and qualitative evaluations of those predictions, as well as analyze how the predicted values relate to different styles and artists. The outcome of our analysis suggests that the models learn to discriminate meaningful features that correspond to the visual characteristics of concepts described by Wölfflin. This indicates that the presented approach can be used to enable new ways of exploring fine art collections based on image features relevant and well-known within art history.

Download

A proposal for characterizing global light pollution using a LEO nanosatellite

2019 42nd International Convention on Information and Communication Technology, Electronics and Microelectronics (MIPRO), 2019

Solid-state lighting (LED lighting) is an energy efficient alternative to conventional sodium, me... more Solid-state lighting (LED lighting) is an energy efficient alternative to conventional sodium, mercury and fluorescent outdoor lighting. The high spectral density of blue light emitted by nominally "white" LEDs has in recent years become an ecological and health concern. Monitoring the global presence of LED lighting as a fraction of total light pollution has hence become a parameter of interest to government organizations dealing with health and environmental protection. This work describes a planned low-earth orbit satellite mission with a low-cost technology and algorithms to provide light pollution images and estimates of the contribution of solid-state lighting relative to conventional illumination globally.

Sonja Grgic

Uploads

Papers by Sonja Grgic

Log In