Publication: Detection of void of single pad in x-ray images using image processing
Loading...
Date
2020-07-01
Authors
Abdullah Norhairi, Ahmad Najib
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
ViTrox Technologies Sdn. Bhd. company uses a three-dimensional (3D) X-ray machine such as the ViTrox AXI to produce X-ray images. The X-ray images obtained are in form of digital image with defects, low contrast and low resolution. Manual inspection methods are performed by inspectors of the company to detect void of single pad in X-ray images using image processing. This manual procedure is too subjective and time-consuming. In addition, the company also faces problem to successfully detect void using their own image processing technique. The purpose of this project is designing a detection system for void on a single pad in X-ray images, which could overcome or reduce the abovementioned problems. There are three stages of image processing applied in this study in order to achieve this project objective, which are image acquisition stage, pre-image processing stage and image processing stage. Image acquisition stage involves two activities namely X-ray images acquisition and image editing activities. During the pre-image processing stage, the contrast of the X-ray images will be enhanced. This study will investigate the capability of four contrast enhancement methods namely Contrast Stretching, Histogram Equalization, Contrast Limited Adaptive Histogram Equalization (CLAHE) and Adaptive Histogram Equalization methods. The Adaptive Histogram Equalization method had been found as the best technique for enhancement stage. Finally, during the image processing stage, segmentation process will be applied to detect the defect regions and then distinguish the detected defect regions from other regions by using thresholding technique. Based on 32 X-ray images, qualitative analysis shows that the proposed void detection system has capability in segmenting void on single pad in X-ray images. Void regions are better segmented, and this could assist company inspectors for better inspection process. With further research, the proposed void detection system could be further enhanced in order to provide an automated void detection system, which could reduce human labor and provide better segmentation performance.