Abstract: Based on the Canny algorithm, this paper optimizes the weight distribution of operator templates and adopts a separable component strategy of the HSV color space to improve the accuracy and stability of edge detection under noise interference. To mitigate the influence of noise on edge detection, a 5×5 Sobel operator replaces the traditional 3×3 operator to optimize template weight distribution and enhance the gradient response capability of edge detection. Through image color mode conversion, hue, saturation and value components can effectively suppress noise in input images. Generally, image detection accuracy decreases significantly in scenarios with drastic light variations. Combined with morphological operations and the Hough transform, the proposed scheme realizes accurate lane line recognition and fitting, and effectively addresses the low detection accuracy of conventional methods in complex and noisy environments.

Key words: Sobel operator; Canny algorithm; HSV color space; edge detection; ROI extraction; Hough transform

摘要： 文章基于 Canny 算法，通过优化算子模板权重分布、结合可分离 HSV 颜色空间分量策 略，提升了噪声干扰场景下边缘检测的准确性与稳定性。为了克服噪声对边缘检测的影响， 采用 5×5 的 Sobel 算子替代传统的 3×3 算子优化模板权重分布，增强边缘检测的梯度响应 能力。通过图像色彩模式转换，利用色调、饱和度和亮度分量可有效抑制输入图像噪声。通 常情况下，在光线剧烈变化的空间中图像检测精度会大幅度降低，该方案结合形态学操作和 霍夫变换实现了车道线的精确识别与拟合，有效解决了传统方法在复杂环境和噪声环境下的 检测精度问题。

关键词: Sobel 算子；Canny 算法；HSV 颜色空间；边缘检测；ROI 提取；Hough 变换