Machine Vision and Canny Edge Detector–Based Printing Image Detection and Recognition Model.
Saved in:
| Title: | Machine Vision and Canny Edge Detector–Based Printing Image Detection and Recognition Model. |
|---|---|
| Authors: | Liu, Bei1 (AUTHOR) liubei@mjc-edu.cn, Calì, Michele (AUTHOR) michele.cali@dieei.unict.it |
| Source: | Modelling & Simulation in Engineering. 3/6/2026, Vol. 2026, p1-12. 12p. |
| Subjects: | Edge detection (Image processing), Image recognition (Computer vision), Thresholding algorithms, Image converters, Image quality in imaging systems, Computer vision |
| Abstract: | Traditional image detection and recognition models face issues such as high miss detection rates, low classification accuracy, and poor real‐time performance when dealing with complex printing images, including defects and flaws. To address these problems, this study proposes a printing image detection and recognition model based on machine vision technology and the Canny edge detector. The model effectively detects printing image quality issues that traditional methods cannot identify, such as subtle defects, image noise, and deformations, particularly in images with complex backgrounds. By incorporating the Canny operator for edge detection, optimizing thresholds with the Otsu method, and using a camera linear model for image acquisition and preprocessing, this study enhances the accuracy and efficiency of image detection and recognition. The outcomes indicated that the average accuracy and precision of the research‐proposed model after training were 97.48% and 97.85%, respectively, in the simulation and simulation running experiments. The average recall and F1‐score of the model were 0.93 and 0.94, respectively. Furthermore, the average ratio of the intersection and concatenation of the model′s detection region and the real region was 93.55%. In the actual model performance experiments, the average training time of the model was 7.75 s, and the average inference time was 2.75 s. Furthermore, the average interclass spacing of the model′s extracted features was 3.71, and the average intraclass distance was 0.49, and the ratio was as high as 7.88. In addition, the model has the highest efficiency in printing image segmentation in different scenes with rich detail information. In summary, the proposed model can improve the stability and robustness of printing image detection and recognition and realize the application of computer vision technology in a wider range of image processing fields. [ABSTRACT FROM AUTHOR] |
| Copyright of Modelling & Simulation in Engineering is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | Traditional image detection and recognition models face issues such as high miss detection rates, low classification accuracy, and poor real‐time performance when dealing with complex printing images, including defects and flaws. To address these problems, this study proposes a printing image detection and recognition model based on machine vision technology and the Canny edge detector. The model effectively detects printing image quality issues that traditional methods cannot identify, such as subtle defects, image noise, and deformations, particularly in images with complex backgrounds. By incorporating the Canny operator for edge detection, optimizing thresholds with the Otsu method, and using a camera linear model for image acquisition and preprocessing, this study enhances the accuracy and efficiency of image detection and recognition. The outcomes indicated that the average accuracy and precision of the research‐proposed model after training were 97.48% and 97.85%, respectively, in the simulation and simulation running experiments. The average recall and F1‐score of the model were 0.93 and 0.94, respectively. Furthermore, the average ratio of the intersection and concatenation of the model′s detection region and the real region was 93.55%. In the actual model performance experiments, the average training time of the model was 7.75 s, and the average inference time was 2.75 s. Furthermore, the average interclass spacing of the model′s extracted features was 3.71, and the average intraclass distance was 0.49, and the ratio was as high as 7.88. In addition, the model has the highest efficiency in printing image segmentation in different scenes with rich detail information. In summary, the proposed model can improve the stability and robustness of printing image detection and recognition and realize the application of computer vision technology in a wider range of image processing fields. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 16875591 |
| DOI: | 10.1155/mse/7182616 |
