<![CDATA[Edge detection is a crucial process in digital image processing, particularly in automated visual inspection systems for packaging quality control. Cardboard packaging used in traditional food products often experiences deformation due to mechanical stress or poor distribution, thus requiring a reliable damage detection method. This study aims to compare the performance of five classical edge detection algorithms, Canny, Sobel, Prewitt, Roberts, and Laplacian of Gaussian (LoG), in identifying contours and structural deformations in product packaging images. Data were obtained through the acquisition of five cardboard images using a high-resolution smartphone camera. The processing steps include image conversion to grayscale, application of the edge detection algorithm, and quantitative evaluation of the results. The evaluation was conducted using three main metrics: Peak Signal-to-Noise Ratio (PSNR), Mean Squared Error (MSE), and processing time. The results show that the Sobel algorithm provides the best performance, with the highest PSNR and lowest MSE values consistently, despite having the longest processing time. In contrast, the Canny algorithm shows the highest efficiency in speed, but with low detection quality. Prewitt and LoG yielded relatively balanced intermediate results between accuracy and efficiency, while Roberts performed moderately across all aspects. These findings indicate that algorithm selection should be tailored to system requirements. Sobel is more appropriate for applications that prioritise accuracy, while Canny is recommended for real-time systems. This study provides an initial basis for the development of lightweight visual inspection systems in the traditional food industry and the MSME sector]]>
Copyrights © 2026
