Institute of Information Theory and Automation

You are here


Dept.: ZOI Duration: 2024 - 2026
Current convolutional networks work with inefficient pixel-wise image representation, which does not provide almost any invariance. This leads to using very large training sets and massive augmentation. We propose to decompose intra-class variances into two degradation operators where one of them can be mathematically modelled by a superposition integral with a transformation of the...
Dept.: ZOI Duration: 2021 - 2024
The project aims at developing a conceptually novel approach to inverse problems by proposing a unified theory of invariants to integral transformations using projection operators, and applying it to image restoration and classification problems in biomedical imaging and robotics.
Dept.: ZOI Duration: 2018 - 2021
The proposal falls into the area of computer image analysis and pattern recognition. It is focused on special type of data - multidimensional vector and tensor fields. Vector fields may describe particle velocity, optical/motion flow, image gradient, deformation/condutivity/diffusion tensors, and other phenomena. Each pixel value is a vector/tensor whose components cannot be treated...
Dept.: ZOI Duration: 2018 - 2023
Dept.: ZOI Duration: 2015 - 2017
The project falls into the area of image analysis and object recognition. We mainly focus on two important connected areas: object description by invariant features and image fusion. The theoretical concept is based on the formulation of all tasks as optimization problems and looking for proper cost functions and algorithms for their maximization. We propose a representation of objects and images...
Dept.: ZOI Duration: 2014 - 2017
The goal of the project is to develop a sophisticated software for videokymography (VKG) which will enable automatic evaluation of medical videokymographic recordings of vibrating vocal folds and arrive at correct medical diagnosis. Further goal is to develop a certified method of VKG evaluation to be used in clinical practice.
Dept.: ZOI Duration: 2012 - 2014
The project deals with interdisciplinary research of digital image processing methods applied to analysis and conservation of artworks. The aim is development of image processing methods for multimodal data acquired during the analysis of historical artworks exploiting the knowledge of multispectral properties of dyes, used for artwork creation. Part of the project is an experimental study of...
Dept.: ZOI Duration: 2011 - 2013
The project deals with generalization of existing theory of moment invariants, with derivation of new invariants for object description and recognition, with proposal of stable and efficient numerical algorithms for their computation, and with their software implementation in a form of MATLAB toolbox. Theoretical part covers a design of invariants to image blurring as well as to geometric...