PROJECT TITLE:

Image-Based Process Monitoring Using Low-Rank Tensor Decomposition

ABSTRACT:

Image and video sensors are increasingly being deployed in advanced systems because of the rich process data that these sensors can capture. Thence, image data play an necessary role in method monitoring and management in several application domains such as producing processes, food industries, medical decision-creating, and structural Health Monitoring. Existing process monitoring techniques fail to totally utilize the information of color images due to their advanced knowledge characteristics together with the high-dimensionality and correlation structure (i.e., temporal, spatial and spectral correlation). This paper proposes a replacement image-primarily based process monitoring approach that's capable of handling each grayscale and color images. The proposed approach models the high-dimensional structure of the image information with tensors and employs low-rank tensor decomposition techniques to extract vital monitoring features monitored using multivariate control charts. Furthermore, this paper shows the analytical relationships between totally different low-rank tensor decomposition methods. The performance of the proposed technique in quick detection of process changes is evaluated and compared with existing strategies through extensive simulations and a case study in a very steel tube producing method.