Vision Systems in Automation, Robotics, and Other Industrial Applications based on Image Quality Assessment Methods

Abstract: 

The presentation will highlight selected topics related to methods for objective digital image quality assessment in various applications. Reliable image quality assessment allows for assessing both the correlation of image distortions and noise with human perception, as well as the reliability of image analysis results. Example applications of image quality assessment methods discussed during the lecture include assessing the surface quality of 3D prints, detecting cracks in riveted joints in aviation industry, as well as estimating the accuracy of text recognition based on distorted images.

Although typical image quality assessment metrics, highly correlated with human perception of various types of image distortions, belong to the family of full-reference methods, where the knowledge of the reference "pristine" image is assumed, the most attractive solution is the use of no-reference ("blind") metrics that do not require it at all. Particularly in some cases, such as evaluation of 3D printed surfaces or crack detection under riveted joints, it is impossible to collect reference images. Nevertheless, the application of the approach based on the mutual similarity makes it possible to compare some image fragments expected to be nearly identical. Therefore, some full-reference metrics may be applied as pseudo-blind methods as the reference image is no longer needed.

Bio:

Professor Krzysztof Okarma (Senior Member, IEEE) received the M.S. degrees in electronics and telecommunication (Hons.) and in computer science from the Szczecin University of Technology, Poland, in 1999 and 2001, the Ph.D. degree in electrical engineering in 2003, and the Habilitation degree in automation and robotics from the West Pomeranian University of Technology (WPUT), Szczecin, Poland, in 2013. He has been with the Department of Signal Processing and Multimedia Engineering, as an Assistant Professor, since 1999, an Associate Professor, since 2013, and the Head of the Department, since 2016 until 2024. He has also been the Dean of the Faculty of Electrical Engineering, WPUT, since 2016 until 2024. He is the author or co-author of more than 250 journal and conference papers or book chapters and holds two patents. His research interests include computer and machine vision and image processing and analysis, particularly image quality assessment and signal processing methods. He is also the Chairperson of the Board of Control in the Association for Image Processing - Polish Chapter of the International Association for Pattern Recognition (IAPR) and the member of the Council of the National Science Centre in Poland since 2024.