عنوان مقاله [English]
Statistical process control (SPC) has been playing an important role in many industries over the past several decades especially in complex processes where lack of effective quality control methods could easily lead to the production of defective products and waste of resources. Due to the importance of processes and inefficiency of methods based on human inspection, the use of image data for statistical process control has gained an increasing attention among researchers in recent years. Data of image nature have been considered by industries for many years to prevent defective products to reach customers. In recent years and with the advent of new image technology, researchers in the field of statistical process control have tried to take an advantage of such data to develop methods that could help to detect anomalies in products more effectively. Image data analysis can be divided to two categories of spatial domain and frequency domain. The main concentration of previous researches in the area of process monitoring using image data is in the area of spatial domain analysis. In this paper, three methods based on one-dimensional wavelet decomposition are proposed for monitoring of image data with respect to frequency domain features. In each level of decomposition, wavelet transformation decomposes each signal to two elements including an approximation part, that is similar to the main signal and performs as a low pass filter, and a detail element that performs as a high pass filter. The first method applies approximation coefficient for process monitoring. The second and third methods consider the detail coefficient by using hard thresholding and soft thresholding, respectively. These methods can signal out-of-control conditions and estimate the change point which can help practitioners with diagnostic information. The performance of these methods are evaluated and compared based on average run length and the change point criteria using a generalized likelihood ratio control chart. Simulation studies using a textile image show suitable level of accuracy in detecting out-of-control conditions and change point estimate.