基于FNLM和Frangi滤波的路面裂缝分割算法

Pavement crack segmentation algorithm based on FNLM and Frangi filters

Frangi滤波是一种由二阶偏导数组成的基于Hessian矩阵的滤波器,对线状物体有较高的响应值,可以突出目标和背景的区别。但其对图像求Hessian矩阵之前需进行高斯预处理,这会导致高频信息丢失、边缘细节模糊及实际路面情况复杂多变。针对此问题,文中提出一种基于FNLM与Frangi滤波的路面裂缝分割算法。该算法对Frangi滤波中原有的高斯核进行改进,将其与快速非局部平均滤波相结合,在平滑噪声的同时保留裂缝边缘细节,从而有效抑制噪声对分割的影响。实验结果表明:所提算法在复杂环境下分割精度较高,其结构相似性为0.9317,相较于传统算法提高27.19%;平均运行时间为10.49 s,可满足实时检测的需求。

Frangi filter is a kind of filter based on Hessian matrix,which is composed of second-order partial derivatives.It has a high response value to linear objects and can highlight the difference between target and background,but it needs Gaussian preprocessing before calculating Hessian matrix of images,which can lead to the loss of high-frequency information,blurry edge details and the complicated pavement situation.On this basis,a pavement crack segmentation algorithm based on FNLM and Frangi filters is proposed.In this algorithm,the original Gaussian kernel in Frangi filter is improved,and combined with fast non-local means filter to smooth the noise while retaining the edge details of the crack,so as to effectively suppress the influence of noise on the segmentation.The experimental results show that the proposed algorithm has high segmentation accuracy in the complex environment,its structure similarity is 0.9317(which is 27.19%higher than the traditional algorithm),and its average running time is 10.49 s,which can meet the needs of real-time detection.