Digital Image Quality Assessment Technology for Weld Radiographic Film Based on ROI Multi-Indicator Fusion

• Main Authors: Yang, Bo, Yan, Haobo, Jiang, Hongquan, Zhang, Xiaoming, Zhou, Zhi, Bao, Rui, Tian, Weilong
• Format: Conference Proceeding
• Language: English
• Subjects: Digital images; Entropy; Entropy weighting method; Image quality; Image quality assessment; Noise level; Quality assessment; Radiographic testing; Radiography; Region of interest; Resistance; Welding; X-ray imaging
• ISSN: 2693-3128
• DOI: 10.1109/ITNEC60942.2024.10733054

The digital image quality of weld seam radiographic inspection film is crucial for enabling intelligent defect detection and information management. However, current image quality assessment methods often overlook the characteristics and quality requirements of the region of interest (ROI) in weld seam radiographic films. To address these issues, this paper introduces a digital image quality assessment method based on the multi-indicators fusion of weld radiographic films' ROI. First, four basic image quality indicators were established based on the national standard GB/T 26141.2, which reflect the overall contrast and noise level of digital images. Then, two key quality indicators were established that reflect the digitization quality of ROI regions in images and the level of noise resistance. Finally, an indicator fusion strategy based on the entropy weight method was formulated for image quality assessment. Validated with enterprise data, the results show a correlation coefficient of 0.8197 between the proposed method and expert assessments, meeting the quality assessment requirements for enterprise X-ray images.