The widespread application of machine vision system promotes the development of image-based statistical process control methodologies to improve the utilization of image data. Previous research has focused on either identification of fault size and/or location or detection of fault occurrence. There is limited research on both fault detection and identification. In this paper, an EWMA and region growing based control charting method is proposed for monitoring images of industrial products of which quality is either characterized by uniformity (e.g. LCD monitors) or a specific pattern (e.g. manufactured tiles). Simulation results show that the proposed method is not only effective in quick detection of the fault but also accurate in estimating the fault size and location. An experimental study is also provided to highlight how practitioners can implement the proposed method in applications.

机器视觉系统的广泛应用促进了基于图像的统计过程控制方法学的发展，以提高图像数据的利用率。先前的研究集中于确定故障的大小和/或位置或检测故障的发生。关于故障检测和识别的研究很少。在本文中，提出了一种基于EWMA和区域增长的控制图表方法，用于监视质量以均匀性（例如LCD监视器）或特定图案（例如人造砖）为特征的工业产品的图像。仿真结果表明，该方法不仅能够快速有效地检测出故障，而且能够准确估计出故障的大小和位置。