基于改进Canny算法的受电弓碳滑板边缘检测方法

An edge detection method of pantograph carbon slide based on an improved Canny algorithm

针对传统边缘检测算法对碳滑板边缘进行检测时存在提取不准确、不连续和易受噪声影响等问题,文章提出一种基于改进Canny算法的受电弓碳滑板边缘检测方法。首先,使用各项异性扩散和双边滤波相结合的级联滤波器代替原算法中的高斯滤波器对图像进行去噪,提高算法对多种混合噪声的平滑能力,并采用伽马变换突出图像的边缘特征;采用Scharr算子确定边缘梯度幅值;引入Otsu算法改进原算法中高低阈值需人为设定的不足,使算法在确定双阈值时拥有更好的自适应性;最后利用形态学思想和Zhang-Suen细化算法对边缘轮廓进一步完善,消除边缘中可能出现的细微空洞。试验结果表明,改进后的算法相较于传统算法能更准确地提取碳滑板的边缘信息,并且具备较强的自适应性与抗干扰性,为受电弓碳滑板的后续检测提供了有效的理论支持。

Aiming at the problems of inaccurate extraction,discontinuity and being susceptible to noise when the edge of a carbon slide is detected by using a traditional edge detection algorithm,an edge detection method of pantograph carbon slide based on an improved Canny algorithm was proposed in the paper.Firstly,the cascade filter with a combination of anisotropic diffusion and bilateral filter was used instead of the Gaussian filter in the original algorithm to denoise the image and improved the algorithm's ability to smooth various mixed noises.Gamma transformation was used to highlight the edge features of the image.Scharr operator was used to determine the gradient magnitude of the edges.Otsu algorithm was introduced to overcome the drawback of manually setting the high and low thresholds in the original algorithm,making the algorithm have better adaptability in determining the dual thresholds.Finally,the edge contour was further improved with morphological ideas and the Zhang-Suen refinement algorithm to eliminate possible subtle voids in the edges.The experimental results show that the improved algorithm can extract the edge information of the carbon slide more accurately than the traditional algorithm,and has strong adaptability and anti-interference,which provides effective theoretical support for the subsequent detection of the pantograph carbon slides.