基于图像处理的多种边缘检测算法研究综述

A Review of Various Edge Detection Algorithms Based on Image Processing

摘要: 边缘检测作为图像处理与计算机视觉中的基础问题，在医学影像、自动驾驶和安防监控等领域发挥着重要作用。本文对多种经典边缘检测算法进行了系统性综述，并结合实验对其在噪声环境下的性能进行了对比分析。结果表明，Canny算子凭借多阶段处理机制在抗噪性和定位精度方面表现最佳，LoG算子在抑噪上具有一定优势，而Roberts、Sobel、Prewitt、Laplacian等传统算子则在噪声条件下易产生断裂与伪边。综述进一步指出，各类方法虽在实现复杂度与实时性上各具特点，但普遍存在对弱边响应不足、噪声下误检率高的共性缺陷。最后，本文总结了未来可能的发展方向，包括多尺度特征融合、边缘保持滤波与轻量化学习机制，以期为后续研究提供参考。

Abstract: Edge detection, as a fundamental problem in image processing and computer vision, plays a crucial role in various fields such as medical imaging, autonomous driving, and security monitoring. This paper presents a systematic review of several classical edge detection algorithms and conducts comparative experiments on their performance under noisy environments. The results demonstrate that the Canny operator, owing to its multi-stage processing mechanism, achieves superior noise robustness and edge localization accuracy. The LoG operator shows certain advantages in noise suppression, while traditional operators such as Roberts, Sobel, Prewitt, and Laplacian tend to produce fragmented and spurious edges under noisy conditions. The review further highlights that although these methods differ in terms of implementation complexity and real-time performance, they commonly suffer from weak edge insensitivity and high false detection rates in noisy environments. Finally, this paper summarizes possible future development directions, including multi-scale feature fusion, edge-keeping filtering and lightweight learning mechanisms, in order to provide reference for subsequent research.