Computer vision and biosignal processing

group leader

prof. Mirjana Bonković, PhD

associates

prof. Mojmil Cecić, PhD
prof. Tamara Grujić, PhD
assoc. prof. Ana Kuzmanić Skelin, PhD
assoc. prof. Ivo Stančić, PhD

Research topics

  1. Development and implementation of sensors and measuring systems for measuring anthropometric parameters and bioelectrical signals
  2. Development of advanced algorithms and bio signal classification systems
  3. Development of advanced algorithms for classification of unbalanced data
  4. Development of a system for intelligent monitoring of visual behaviour
  5. Development of wearable sensors for monitoring respiratory system parameters
  6. Development of wearable sensors for monitoring movement parameters with the aim of assessing the success of rehabilitation procedures
  7. Development of algorithms for intelligent vision control (mobile robots and aerial vehicles)
  8. Development and application of data fusion algorithms
  9. Development of advanced algorithms for image quality improvement (superresolution)
  10. Development and application of algorithms for 3D modelling and visualization
  11. Development and application of algorithms for augmented reality
  12. Development of algorithm for steganography
  13. Development and implementation of advanced optoelectronic sensors
  14. Development of customized computer and machine vision algorithms with the aim of execution on end devices
  15. Development of algorithms and application of machine learning in embedded systems

Description of laboratory and equipment

The group carries out its activities by working in two laboratories: the Laboratory for Mobile Robotics and Vision Guidance, and the Laboratory for 3D Modeling and Visualization. The laboratories are equipped with various devices for biosignal measurements (EMG, XSens inertial sensors) and image collection (standard camcorders, 3D cameras and high-speed cameras), as well as other signal collection devices (3D scanner, Leap Motion, accelerometers, Kinect). In addition to the above, the laboratories are also equipped with several drones, mobile robotic platforms, robotics kits (Arduino, NXT Mindstorm, AlphaBot) and several types of embedded computers for use in robotics.

Contacts with academic and other institutions

  • Eastern Macedonia and Thrace Institute of Technology (EMaTTech), Department of Computer & Informatics Engineering, Drama, Greece
  • Bulgarian Academy of Science/Institute of Robotics, Sofia, Bulgaria
  • University of Ljubljana, Faculty of Computer and Information Science, Slovenia
  • University of Dubrovnik, Faculty of Electrical Engineering and Applied Computing, Croatia
  • University of Split, School of Medicine (MEFST), Croatia
  • University of Split, Faculty of Maritime Studies (PFST), Croatia
  • Clinical Hospital Center Split, Croatia
  • Udruga za robotiku
  • Udruga za automatiku i biomehaniku
  • Zajednica za tehničku kulturu
  • Croatian Robotic Association
project title

Intelligent methods for biosignals and environments (IMEOBIO)

Description of research in a 5-year term

The project group research interests are related to various aspects of the functionality of the bio-identification system (persons or activities) and identification/visualization of the environment realized using different types of sensors that include video, but also sensors such as accelerometers, 3D scanners, magnetometers, IR sensors, etc. Given that the interest of the group is partly focused on bioelectrical data obtained, for example, based on EEG, acelerometers, and / or phonopneumograms from which strong features of individuality can be derived, it is expected that the theoretical framework will provide advanced machine learning methods.

A certain set of activities will be focused on the management of mobile robots/vehicles/aircraft with application in agriculture and plant growing. The subject of research is related to the development of mobile robot AI control algorithms, but also to augmented reality visualization techniques that facilitate the task when control is not completely autonomous.