Wavelt-based approach to the detection of defects in texture images: theoretical and practical aspects

Friday, May 19, 2017 - 10:00
Rectorate Hall (K. Donelaicio St. 73-402 room)

Author, Institution: Gintarė Vaidelienė, Kaunas University of Technology

Science Area, Field of Science: Physical Sciences, Informatics – 09P

Summary of the Doctoral Thesis: Summary

Scientific Supervisor:  Prof. Dr. Jonas VALANTINAS (Kaunas University of Technology, Physical Sciences, Informatics, 09P).

Dissertation Defence Board of Informatics Science Field:

Prof. habil. dr. Rimantas BARAUSKAS (Kaunas University of Technology, Physical Sciences, Informatics, 09P) – chairman;
Prof. habil. dr. Juozas AUGUTIS (Vytautas Magnus University, Physical Sciences, Informatics, 09P);
Prof. habil. dr. Genadijus KULVIETIS (Vilnius Gediminas Technical University, Physical Sciences, Informatics, 09P);
Prof. habil. dr. Minvydas Kazys RAGULSKIS (Kaunas University of Technology, Physical Sciences, Informatics, 09P);
Prof. dr. Miguel A. F. SANJUAN (Rey Juan Carlos University, Spain, Physical Sciences, Informatics, 09P).

The Doctoral Dissertation is available on the internet and at the libraries of Kaunas University of Technology (K. Donelaičio St. 20, Kaunas) and Vytautas Magnus University (K. Donelaičio g. 52, Kaunas)

Annotation:

This dissertation proposes a novel defect detection technique for texture (surface) images. The developed technique is implemented in the spectral discrete wavelet domain. The proposed texture defect detection technique is based on multiple image scanning, which allows forming a multi-valued texture defect detection criterion, when the final decision on the quality of the texture image under processing is made depending on the percentage of unfavourable criterion values. The task-oriented statistical analysis of specific subsets of Haar wavelet coefficients of a texture image enhances the implementation of the texture defect detection system (methodology), which ensuring the flexibility of the system. The proposed technique for detecting texture defects is successfully applied for the analysis of actual texture surfaces and their quality control. Defect detection is relatively high (83–98 %) and competitive image testing accuracy is guaranteed.