A Spatiotemporal Method for the Monitoring of Image Data

Abstract

Machine vision systems are increasingly being used in industrial applications because of their ability to quickly provide information on product geometry, surface defects, surface finish, and other product and process characteristics. Previous research for monitoring these visual characteristics using image data has focused on either detecting changes within an image or between images. Extending these methods to include both the spatial and the temporal aspects of image data would provide more detailed diagnostic information, which would be of great value to industrial practitioners. Therefore, in this article, we show how image data can be monitored using a spatiotemporal framework that is based on an extension of a generalized likelihood ratio control chart. The performance of the proposed method is evaluated through computer simulations and experimental studies. The results show that our proposed spatiotemporal method is capable of quickly detecting the emergence of a fault. The computer simulations also show that our proposed generalized likelihood ratio control charting method provides a good estimate of the change point and the size/location of the fault, which are important fault diagnostic metrics that are not typically provided in the image monitoring literature. Finally, we highlight some research opportunities and provide some advice to practitioners.

Control Phase

“…Most often, images have been used in industry for product inspection, where non-conforming items are separated from the conforming items, but we see opportunities for detecting more subtle quality changes before non-conforming items are produced. Megahed et al [29] reviewed work on the use of control charts with image data.…”

Spatiotemporal Surveillance Using Biaxially Weighted Scan Statistic

“…The provided evidence and the timely judgment empower decision makers to manage the course of events. For example, monitoring the spread of a disease (Woodall [22], Tsui et al [19]), detecting the emergence of a production fault (Hansen and Thyregod [3], Tong et al [18], Megahed et al [10]), and investigating abnormal weather conditions (Lalaurette [7]) are a few critical problems in public health, industrial, and environmental applications, respectively. Studying the clustering would even reveal the contributing factors of anomaly to investigators.…”

A novel framework for spatiotemporal monitoring and post‐signal diagnosis of processes with image data

“…For example, for 8‐bit grayscale images, the intensity value will be between 0 (black) and 255 (white); for 16‐bit images, the intensity value will be between 0 (black) and 65535 (white). In industry, 8‐bit images are usually used for monitoring due to easy preparation and analysis and shorter processing time (Megahed et al). The images obtained from MVS are preprocessed before they are used to make interpretation easier.…”

“…This includes studies in which not only are processes monitored over time and the change point estimated but also the location(s)/dimension(s) of the occurrence of defect(s) is determined. Megahed et al presented a spatiotemporal generalized likelihood ratio (GLR) control chart to monitor gray‐scale images of products. In the proposed method, they divided each image into predefined subsets, referred to as regions of interest (ROIs), and used the GLR control chart to monitor the process.…”

“…Koosha et al transformed images to wavelet profiles using wavelet transformation and utilized profile monitoring techniques for estimating the real time of change and detecting the location/dimension of a single defect. Despite the good performance of the methods proposed by Megahed et al and Koosha et al, these two approaches are focused on identifying a single defect, and if there is more than one defect in the product, these methods cannot identify them properly. To address this issue, He et al used a multivariate GLR (MGLR) control chart for image data monitoring.…”