Serial Number Recognition of Ceramic Membrane Based on End-to-end Deep Learning

for manufacturers, serial number (SN) is not only beneficial to the centralized assembly of products, but also brings great convenience to the traceability of production process. For workers in wastewater treatment process, the SN of ceramic membrane is also the basis for them to install ceramic membrane correctly. Image resolution, as a key factor to assess the quality of digital images, is the basis for the subsequent processing of ceramic membrane SN recognition. In this article, an image super-resolution (SR) algorithm based on end-to-end DL and computer aided design (CAD) model is proposed. A deep learning (DL) model suitable for mixed views as input signals is selected, and a hierarchical learning structure is constructed by using deep neural network. Combined with the extracted CAD model views, the SN of ceramic membranes is identified. It’s not difficult to seen from the test results that the ceramic membrane SN image recognition model in this article has obvious advantages over the recurrent neural network (RNN), with an accuracy of over 96% and an error of over 25% lower than that of the comparative RNN model. This algorithm improves the reconstruction effect of detailed information in the image, optimize the reconstruction image, improve the automation of Ceramic membrane production and sewage treatment, and promote the development of industrial software. © 2024, CAD Solutions, LLC. All rights reserved.