Image Processing Group (GPI)

The UPC Image and Video Processing Group (GPI) is a research group of the Signal Theory and Communications (TSC) department. Since April 2018, GPI is part of IDEAI, the Intelligent Data Science and Artificial Intelligence Research Center. GPI is located in the Barcelona Knowledge Campus, the first Campus of International Excellence in Spain, a joint initiative of UB and UPC universities.
You can also find us in futur, the Scientific Production Website of UPC.

Welcome to our small window to the world!

About us

The GPI started its activities in the late 80’s and was one of the first groups to apply signal processing techniques to images and latter to video signals in Spain. The group is currently composed of ten faculty members, about the same number of PhD candidates of the TSC Doctorate Program, and usually enrolls from 40 to 50 Final Degree and Master Thesis students. The group teaches undergraduates in Image Processing at the UPC School of Telecommunications Engineering ETSETB (Barcelona) and the School of Industrial, Aerospace and Audiovisual Engineering ESEIAAT (Terrassa), also graduate students in the various masters of ETSETB, including the joint UAB-UPC-UPF-UOC Master in Computer Vision.

The group participates in numerous national and international projects and publishes regularly in journals of international relevance. The GPI has been a Consolidated Research Group of the Catalan Government continuously since the first call 1999 (Calls: 1999-2001, 2002-2005, 2005-2008, 2009-2013, 2014-2016, 2017-).

The group has a long teaching experience, mainly in the Barcelona School of Telecommunications Engineering (Electrical Engineering), TelecomBCN/ETSETB, as well as in the School of Industrial, Aerospace and Audiovisual Engineering in Terrassa, ESEIAAT, a city about 30 km from Barcelona.

Our PhD candidates play an important role within the group. A list of GPI PhD dissertations with links to the actual reports is available from this site. Seminars are regularly held, involving faculty, students, and guest speakers from the Academia and the Industry.

GPI technical staff provides critical support to maintain the research of our group, keeping labs up to date and introducing cutting edge advances in hardware (servers' virtualization, GPGPU, continuous monitoring tools) and software (GPIs own IDE integrating eclipse, svn, trac...)

Regarding facilities, the GPI has two lab rooms for capturing data, experimentation and testbeds: a Smart Room and a recently build Audiovisual Recording Lab. Additionally, a development platform and software libraries were created by the group under the name of ImagePlus. They are constantly updated in a collaborative manner and represent a key element for our research and technology transfer activities.

Main Research Areas

Our main research activities focus on image analysis and on video representation:

• Image Analysis extracts or measures information starting at the level of the sensors and, through an abstraction process, infer or classify conceptual elements and scene semantics.
• Video Representation systems aim to express the visual content of a scene in a compact and efficient support, either at the signal level (coding) or at the semantic level (indexing)

Regarding analysis, the GPI specializes in basic tools for analysis such as nonlinear filtering, mathematical morphology, graph signal processing, segmentation (with Binary Partition Trees), face detection and recognition, multiview analysis for stereo and 3D reconstruction, object tracking, body tracking, gesture analysis, emotion recognition and the modeling of human activity. The Group is also involved in the analysis of remote sensing images in particular hyperspectral and Synthetic Aperture Radar images mainly for denoising, segmentation and classification. Some of the analysis tools developed in the group are also applied to data that are not related to images. This is the case for genomics applications such as gene expression classification or gene regulatory network inference.
 
GPI video compression activities have been the basis of applications related to content-based video coding, video indexing and the creation of tables of content and contributions to international standardization processes such as MPEG-4 and MPEG-7. Most recently GPI is conducting research in multi-view/3D representation and coding.
 
Finally, the GPI also develops biomedical, watermarking applications and multimodal interfaces (Smart Rooms).