The Role of Edge Detection in Improving Convolutional Neural Network Accuracy and Robustness: An Overview

Convolutional neural network (CNN) has become the backbone in image pattern recognition. However, traditional CNNs often struggle to accurately capture edge information. Recent research shows that the integration of edge detection with CNN can improve the performance and robustness of the model. This article conducts a comprehensive literature review to evaluate the contribution of edge detection in CNN. The review results show that edge detection can improve classification accuracy, robustness against adversarial attacks, and feature extraction. In addition, edge detection can also help reduce overfitting. This article concludes that the integration of edge detection is a promising approach to improve CNN performance in various computer vision tasks. Furthermore, the integration of edge detection in CNNs needs to be further refined, especially in the face of image variation challenges. Exploration by applying quantum computing in edge detection is also worth considering given its potential in providing more accurate, clean and noise-resistant edge detection results.