Robot Vision

Subject description

Visual perception: light, human vision, cameras, illumination, image quality parameters, sampling and quantization, visualization, image formats and standards.

Digital image processing and restoration: smoothing and sharpening, statistical and morphological filtering, image resampling, geometrical transformations and registrations.

Robust recognition of 2D objects: keypoints, lines, circles, and template detectors, 3D model alignment to 2D images, basics of unsupervised and supervised object detection. Solution in case of partial object occlusion in the image.

Calibration of imaging systems: distortions of real optical systems, accuracy and precision, spatial homogeneity, temporal stability, self-calibration.

3D object reconstruction: review of techniques for depth perception from 2D images, concepts of systems and methods for stereo vision, structured light and photometric stereo.

Visual navigation: concepts of methods for image-based object tracking and motion analysis, pose estimation, and object localization and environment mapping from 2D images.

Applications of robot vision: visual quality control, product sorting, object and obstacle detection, modelling of object shape and appearance, motion trajectory planning.

The subject is taught in programs

Objectives and competences

The objective of this course is: to introduce the principal building blocks of a robot vision system and the associated fundamental problems; to introduce the main concepts and techniques used to solve those problems; to enable students to create robot vision systems and implement problem solutions; to enable students to understand the basic methodology discussed in the robot vision literature.

Teaching and learning methods

Theoretical foundations, concepts and exemplar use cases are given during lectures, while practical skills are gained through lab works, weekly labwork assignements and an individual seminar work.

Expected study results

After successful completion of the course, students should be able to:

  • list the technologies and main building blocks of robot vision systems, and understand and be able to critically assess and compare their key features,
  • develop and use method for basic image processing and analysis,
  • detect objects and their pose in the image using automated methods,
  • perform system calibration and conduct object measurements,
  • compute depth and shape of an object from 2D images,
  • track object motion in the image, analyse its trajectory and use this data for selflocalization and environment mapping, and
  • design a robot vision system by defining, assembling and using all vital hardware and software components.

Basic sources and literature

  1. Wilhelm Burger in Mark J. Burge. Principles of Digital Image Processing: Fundamental Techniques, Springer, 2009.
  2. Wilhelm Burger in Mark J. Burge. Principles of Digital Image Processing: Core Algorithms, Springer; 1st Edition. 2nd Printing, 2011.
  3. Reinhard Klette. Concise Computer Vision: An Introduction to Theory and Algorithms, Springer, 2014.
  4. Richard Szeliski.  Computer Vision: Algorithms and Applications, Springer; 2011 edition, 2010.
  5. Predstavitev in informacije na spletni strani http://lit.fe.uni-lj.si/RV, navodila za vaje in ostala gradiva v spletni učilnici FE: https://e.fe.uni-lj.si

Stay up to date

University of Ljubljana, Faculty of Electrical Engineering Tržaška cesta 25, 1000 Ljubljana

E:  dekanat@fe.uni-lj.si T:  01 4768 411