|
二维码（扫一下试试看！） |
| CNN-based Image Recognition and Monitoring of States on the Pallet Paper Shortage for Wet Pulp Packaging Machines |
| Received:December 02, 2025 Revised:December 21, 2025 |
| DOI：10.11980/j.issn.0254-508X.2026.05.021 |
| Key Words:image recognition RTSP streaming protocol Compact convolutional neural network feature vector |
| Author Name | Affiliation | Postcode | | XU Feng* | Shandong Sun Paper Co., Ltd., Jining, Shandong Province, 272000 | 272000 | | LI Jianwen* | Shandong Sun Paper Co., Ltd., Jining, Shandong Province, 272000 | 272000 | | ZHANG Ju | Shandong Sun Paper Co., Ltd., Jining, Shandong Province, 272000 | 272000 |
|
| Hits: 923 |
| Download times: 113 |
| Abstract:To address the issue that the wet pulp packaging machine cannot accurately identify paper shortage on wooden pallets, this study proposed a monitoring method based on convolutional neural network (CNN) image recognition, which consists of the following steps: ① Capture front-view images of pallets via the RTSP streaming protocol; ② Perform multiple annotations on the region of interest (ROI), establish an image-to-feature mapping benchmark using a compact network, and apply image enhancement to simulate on-site light intensity fluctuations throughout the process; ③ Conduct convolution operations on real-time images, extract color and edge differences of images as feature vectors, and compare feature vectors following mapping rules to determine the paper shortage status; ④ Output status information to the PLC via the TCP protocol for system-triggered automatic replacement of empty pallets based on classification results. Compared with traditional RGB threshold processing, the lightweight Compact of CNN recognition system effectively mitigated on-site environmental interferences. It enabled automatic replacement of wooden pallets, thereby ensuring continuous and high-efficiency production. |
| View Full Text HTML View/Add Comment Download reader |
|
|
|