An automatic assessment of histogram mode on X-radiography weld image using non-maxima suppression

摘要

An assessment and understanding of image properties is the key task in the digital image processing field. Among image processing techniques, segmentation of the low illumination image is a more typical and important problem besides the selection of histogram-based segmentation technique also challenging task. Particularly, weld defect X-radiography images are generally in low-contrast, and the defects have appeared on the larger background area of the image. Further, the mode of the histogram performs an important component in the histogram-based segmentation techniques and many techniques are categorized based on histogram mode, such as unimodal, bimodal, and multimodal. Therefore, it is a necessary task to prior assessment of histogram mode of the image, which leads to select the appropriate segmentation technique and without any inappropriate attempt. A new approach is proposed to automatically assess the histogram mode of the image using the non-maxima suppression technique. The proposed method is implemented on the X-radiography weld image dataset and the accuracy of the assessment provides satisfactory.