Flaw detection improvement of digitised radiographs by morphological transformations
| Autor: | F. Ghasemi, N. Rastkhah, Mohammad H. Kargarnovin, B. Rokrok, Hamid Soltanian-Zadeh, Mehran Taheri, Kaveh Edalati, Amir Movafeghi |
|---|---|
| Rok vydání: | 2005 |
| Předmět: |
Computer science
business.industry Image quality Mechanical Engineering Fingerprint (computing) ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Metals and Alloys Radiographic testing Wavelet Mechanics of Materials Frequency domain Digital image processing Materials Chemistry Computer vision Artificial intelligence business MATLAB computer Graphical user interface computer.programming_language |
| Zdroj: | Insight - Non-Destructive Testing and Condition Monitoring. 47:625-630 |
| ISSN: | 1354-2575 |
| Popis: | Radiographic inspection is one of the most appreciated techniques among non-destructive testing methods due to the production of a film which acts as a unique fingerprint record. Converting radiographs to a digital format and further digital image processing is the best method of enhancing the image quality and assisting the interpreter on his evaluation. In this research, different algorithms were used for image enhancement and radiograph interpretation in MATLAB environment. A graphical user interface (GUI) was created for implementation of different algorithms. Both spatial and frequency domains techniques were used for image enhancement. The main enhancement technique was a spatial technique based on morphological transformation (top-hat and bottom-hat transforms). The wavelet filtering was implemented as the frequency domain technique. A third technique, based on pseudo-colouring, was also used for a better comparison between different methods. Wavelet-based techniques are used very often nowadays. However, we show that wavelet applications are more complicated and a supervised and intelligent usage has to be carried out. A wisely established spatial technique can result in reasonably proper results. The diagnostic test reliability was carried out for the quantitative evaluation of the results with the help of industrial radiography interpreters. The obtained results indicated that the defect recognition probability can be increased effectively in both domains. However, it was concluded that better results can be achieved using a morphological method. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|