基于改进Canny算法的图像边缘检测
Image Edge Detection Based on Improved Canny Algorithm

鉴于传统Canny边缘检测方法对处理椒盐噪声的高敏感性、边缘梯度变化细微，以及难以精确捕捉目标轮廓边缘信息等问题，本文提出了一种将自适应高斯滤波、曲度算子、灰度梯度映射和最大类间方差法(Otsu’s method)综合运用的方法，旨在提高图像处理的效果和准确性。首先，通过精确调整高斯滤波器和扩展梯度计算邻域，平衡了去噪和保留边缘信息的需求，提高易用性和鲁棒性；其次，引入自动参数调整算法并结合其他边缘检测技术，利用并行计算技术加速处理，以提高计算效率和实时性能。与传统的边缘检测算子相比，去噪图像质量提升10%~22%，边缘评价指标提高15%~25%。该算法不仅有效去除噪声，而且在边缘提取方面表现更优秀。
In this paper, the traditional Canny edge detection algorithm is improved to deal with the problems of sensitive pepper and salt noise, small edge gradient change, and difficulty in effectively extracting the target contour edge information. The algorithm in this paper integrates adaptive Gaussian filter, curve operator, gray gradient mapping and Otsu’s Method to improve the effect and accuracy of image processing. Firstly, by precisely adjusting the Gaussian filter and extending the gradient calculation neighborhood, the need for denoising and preserving edge information is balanced, and the ease of use and robustness are improved. Secondly, the automatic parameter adjustment algorithm is introduced, and other edge detection techniques are combined to accelerate processing using parallel computing technology to improve computational efficiency and real-time performance. Compared with the traditional edge detection operator, the denoised image quality is improved by 12%~22%, and the edge evaluation index is improved by 15%~25%. This algorithm not only effectively removes noise, but also performs better in edge extraction.