Eduardo Peixoto is an Adjunct Professor of the Department of Electrical Engineering in the University of Brasilia.
Eduardo Peixoto received the Engineering and M.Sc. degrees from Universidade de Brasília, Brazil, in 2005 and 2008, respectively, and the Ph.D. degree from Queen Mary, University of London, in 2012, all in Electrical Engineering.
In 2005-2006 he worked at BRISA developing Java applications for Digital Television and Mobile Phones. In 2006, he joined the Digital Signal Processing Group (GPDS) to work with Distributed Video Coding and Transcoding. In 2008, he joined Queen Mary, University of London to pursue a Ph.D., where he worked with Video Coding based on Wavelet Transforms, and with different video transcoders. In 2012, he joined Universidade de Brasília, first as a Substitute Lecturer and then, in 2014, as an Adjunct Professor. Currently, he is with the Department of Electrical Engineering at the University of Brasilia (UnB), where he is a member of the Graduate Program on Electronic Systems and Automation Engineering (PGEA). He is also a member of IEEE and the IEEE Signal Processing Society.
His research interests include image and video processing and coding, as well as the development of applications for embedded systems.
Partners: Edson Mintsu Hung, Eduardo Peixoto and Bruno Macchiavello (UnB), Debargha Mukherjee and Zoe Liu (Google Sponsors)
Funding: Google (Google Research Award)
Improvement of wedge partition techniques used in VP10 Google’s Codec using a boundary based block partition
Video coding is, essentially, an enabling technology. The video data is not compressed for the sake of compression, it is compressed in order to transmit or store more content using the same resources. Most applications would not be feasible without video compression, as the amount of data required would be extremely high. In this project, we aim to improve the performance of motion estimation, an essential tool in all hybrid DPCM/DCT video codecs, by better adapting the prediction step of the codec. The developed algorithms will be implemented in Google's VP10 open source codec.
Partners: UnB, ZTec
Funding: Eletronorte, Intesa
Participants: Francisco Assis de Oliveira Nascimento (Coordinator), Raimundo Saraiva Jr, Anésio de Leles Ferreira Filho, Florêncio Ayres da Silva Filho, Jorge Andrés Cormane Angarita, Eduardo Peixoto, Fernando Guimarães.
This project aims to develop a low cost equipment to measure energy quality parameters. The project comprises the complete development of this equipment, both in hardware and software.
Partners: Eduardo Peixoto (Coordinator), Edson Mintsu Hung and Bruno Macchiavello
Funding: Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) Edital Universal
Development of Fast Algorithms for Video Coding in the HEVC Standard (Desenvolvimento de Métodos Rápidos para Codificação de Vídeo no novo padrão HEVC)
Digital video comsumption has grown exponentially in recent years. Today, we watch video content not only through television, but through the internet as well, whether using personal computers, tablets or mobile phones. At the same time, the quality of the consumed video has improved tremendously in this period - it is common nowadays to watch high resolution (HD) videos through online video streaming. However, transmitting higher quality videos require higher bandwidth. For this reason, the ITU-T and the ISO-IEC released, in 2013, a new video coding standard, so called High Efficiency Video Coding (HEVC). This new standard achieves the same video quality as its predecessor, the H.264/AVC, with only half of the bitrate. However, in order to achieve this efficiency, the new encoder needs much more computational power than the H.264/AVC. Therefore, the development of low complexity, fast encoding algorithms for the HEVC is a highly competitive field, with immediate application. This project focuses on the development of fast video coding algorithms for the HEVC, especially for high definition (1920 x 1080) and ultra-HD (4K, or 3840 x 2160) content.
Participants: Otávio Dias, Maximillian Xavier, Eduardo Peixoto
Heart rate monitoring is an important tool in medicine. The most common methods for measuring the heart rate are the use of chest straps, adhesive electrodes or oximeter sensors. All these techniques require some sort of contact with the patient, which may be undesirable in some cases. The aquisition of this signal without physical contact with the patient is desirable as it simplifies some applications, such as neo-natal ICU, long term monitoring, sleep pattern studies, among others.
In the last decade it has been shown that images of the face of a patient contain sufficient data to measure some important biological signals, such as the heart rate and the arterial pressure. The first studies used thermal images to acquire these signals howeve, recently, some algorithms were shown to be able to gather this information using common video cameras in ambient light.
The main goal of this project is to develop and implement algorithms for heart rate monitoring using just the data from a simple webcam. This algorithm should work in real time, being able to film the face of a person and output an estimative for this person heart rate.
Partners: Ricardo L. de Queiroz, Edson Mintsu Hung, Bruno Macchiavello, Alexandre Zaghetto, Filipe Teixeira
Recently, a new video coding standard has been ratified, so called High Efficiency Video Coding (HEVC). This new standard offers a much better performance compared to its predecessor, the H.264/AVC. The H.264/AVC standard is very successful and it has been widely adopted, both by physical media, internet streaming services and some broadcast and cable television services. Video transcoding is the process that converts from one compressed bitstream (called the source or incoming bitstream) to another compressed bitstream (called the transcoded or outgoing bitstream). The motivation for an H.264/AVC to HEVC transcoder is twofold: (i) to promote inter-operability for the legacy video encoded in H.264/AVC format; and (ii) to be able to take advantage of the superior RD performance of the HEVC, especially for systems that have extensive video libraries. This project focus on developing new algorithms for video transcoding, in particular from H.264/AVC to the current HEVC standard.
Participants: Bruno Pessoa Chaves, Eduardo Peixoto
Conventional Cameras capture only the intensity of light in each point, losing the information about the direction of the incident light rays. Plenoptic cameras, also known as light field cameras, are able to capture not only the intensity, but the direction of the light rays, thus obtaining much more information about the scene than conventional cameras.
An important application for these cameras is the estimation of depth maps, i.e., a map of of the scene that estimates the distance from each object to the camera. These depth maps have broad applications, such as to add or remove blur in the images, as well as to make a 3D scene from the images.
The main goal of this project is to develop algorithms to estimate depth maps from data obtained from a Lytro camera, a low cost first generation plenoptic camera.
Collaboration on project with Queen Mary, University of London
Enabling Technologies for High Resolution Content Exchange and Delivery (Thira)
THIRA is a collaborative project led by the BBC and co-funded by the Technology Strategy Board which addresses usage of larger resolution video formats. Recent advances in video capturing and display technologies provide audiences with higher spatial and temporal resolution content. After High Definition (HD), nowadays Ultra High Definition (UHD) aims to further improve users' quality-of-experience at the cost of increased storage capacity and bandwidth. Ultra high definition targets two formats with 16:9 aspect ratio: the first has twice the width of HD while the second has four times the width of HD. To properly handle the data rates associate to UHD content, the H.265/High Efficiency Video Coding (H.265/HEVC) standard will be mainly used. H.265/HEVC is the newly ratified video compression standard developed by both ITU VCEG and ISO/IEC MPEG in a joint partnership called the Joint Collaborative Team on Video Coding (JCT-VC). HEVC results from a joint VCEG and MPEG call for proposals on compression technology able to achieve the same perceived quality of the H.264/AVC standard by halving the required bitrate. The H.265/HEVC standard provides a base for handling ultra high resolution content. Underpinning the capabilities of HEVC are computationally expensive, extremely complex and resource demanding algorithms. Consequently, in its current state, H.265/HEVC is unfeasible for practical applications involving very high definition content. THIRA aims to facilitate storage and transmission of the huge amounts of data contained in very high resolution content using the H.265/HEVC standard.
Participants: Ricardo L. de Queiroz (Coordinator), Edson Mintsu Hung, Bruno Macchiavello, Rafael Galvão de Oliveira, Tiago Alves da Fonseca, Alexandre Zaghetto, Fernanda Brandi, Eduardo Peixoto, Karen França de Oliveira, Renan Utida Ferreira, Diogo Caetano Garcia, Camilo Chang Dórea, Thacio Scandarolli, Frederico Nogueira Leite
Partners: HP Labs Palo Alto
This long term project was funded by HP Brasil and it had a collaboration with HP Labs in Palo Alto, California.
This project funded six PhD scholarships and more than ten MSc scholarships for many researchers, many of whom are now Adjunct Professors at UnB.
The project itself had several threads, including research in Distributed Video Coding, Video Transcoding, Super-Resolution, 3D-Video, Document Compression, Motion Estimation and Compensation and Directional Transforms for use in Image Coding.
I am currently seeking for motivated students for Final Projects, M.Sc. or Ph.D. Check the research projects and contact me directly.
Download all citations: [bibtex] [plain text]
111813 - Sistemas Digitais
111821 - Laboratório de Sistemas Digitais
167983 - Sistemas Digitais 1
366102 - Codificação Digital de Sinais (Post-Graduate)167983 - Sistemas Digitais 1
167746 - Laboratório de Arquitetura de Processadores Digitais
169935 - Tópicos Especiais em Processamento de Sinais (Microcontroladores MSP 430)
167983 - Sistemas Digitais 1
167746 - Laboratório de Arquitetura de Processadores Digitais
167983 - Laboratório de Sistemas Digitais 1
167746 - Laboratório de Arquitetura de Processadores Digitais
167983 - Laboratório de Sistemas Digitais 1