The Algorithm Of Pavement Crack Detection Based On Image Processing

With the rapid development of highway construction,the requirements for highway classification have been continuously improved,and the inspection and maintenance of highway pavements has become the primary task of highway construction in recent years.Pavement cracking is an early form of most pavement diseases.Detecting cracks and repairing them in time can effectively reduce the damage caused by road surface damage.Therefore,pavement crack detection has important and practical significance.Pavement crack detection generally includes four parts: image preprocessing,segmentation,feature extraction and classification & identification.Based on digital image processing technology,the detection and recognition system of pavement crack image is studied in this paper.The main work and conclusions are as follows.In the image preprocessing part,the improved median filter and unsharp mask algorithm are combined to obtain the details of the image by using the difference between the improved median filter and the original image,and the detail is sharpened by using adaptive adjustment parameters.Experimental results show that the algorithm can highlight the details of the image,improve the overall contrast,and can effectively suppress the noise,and lay the foundation for the subsequent crack segmentation.In the image segmentation part,the specific effect of the edge detection operator in the crack image segmentation is firstly studied.Aiming at the problem that the edge blur extracted is not easy,an optimized fuzzy clustering segmentation algorithm is proposed.The shared k-Nearest Neighbor algorithm is used to obtain the best initial clustering center to avoid local optimization of clustering,so that the segmentation result is optimal.Finally,a series of morphological processing operations,such as expansion,erosion,opening operation and closing operation,are carried on the obtained binary image.In the image feature extraction part,the four types of road surface cracks are first studied.The shape features are extracted by projection features and Hu invariant moments.The texture features are extracted by the gray level co-occurrence matrix and Gabor wavelet transform.The weights of each feature are calculated separately.The four features are weighted and fused to obtain the final fusion feature.In image recognition part,the support vector machine is used as the final classifier for the features of the road surface cracks extracted in this paper,and the method of One-Against-All is uesd to train the extracted features.The experimental results show that theimage recognition based on a single feature can not accurately identify the crack image.However,the algorithm in this paper improves the recognition rate of multi-feature fusion and reduces the computational complexity of the traditional fusion algorithm.