Radiographic image segmentation for weld inspection using a robust algorithm

Xin Wang, Brian Stephen Wong

Research output: Contribution to journalArticleResearchpeer-review

13 Citations (Scopus)


The radiographic testing method is often used as a nondestructive testing method for detecting welding defects. Because of the degraded quality and the small size of the defects, X-ray films are sometimes difficult to inspect. The interpretation of such images is often affected by a human operator's subjectivity. Digital image-processing techniques allow the interpretation to be automated. A key step in the automated-interpretation process is the segmentation of indications from the background. In this article, a robust method is presented to segment the radiographic image. In our algorithm, first adaptive wavelet thresholding and adaptive histogram equalization techniques are used to improve the quality of the radiographic image. Then, the radiographic image is divided into three parts, namely black, gray, and white parts using three-level thresholding based on maximum fuzzy entropy. The procedure to find the optimal thresholds is implemented by a genetic algorithm, which can overcome the computational-complexity problem. The experimental results show that our proposed method gives good performance for radiographic images.

Original languageEnglish
Pages (from-to)131-142
Number of pages12
JournalResearch in Nondestructive Evaluation
Issue number3
Publication statusPublished - Jul 2005
Externally publishedYes


  • Genetic algorithm
  • Image segmentation
  • Radiographic inspection
  • Three-level thresholding

Cite this