| In China’s highway construction projects,bridges are an important part of the transportation system.During its whole service term,concrete bridge is continuously bearing environmental action from climate,geological conditions,traffic loads as well as other,load,etc.,causing varying extent of damage.Therefore,regular inspection and maintenance of highway bridges are meaningful.In the periodic inspection and evaluation of bridges,cracking is one of the most critical indicators.In traditional cracking detection,usually only while very little work is done current status and characteristics of crack are judged,on the change of cracks during the detection interval as well as prediction of its future development trends.In this paper a method of bridge crack extraction and matching based on digital image processing technology is proposed.The main steps of this method include collect the cracks of the concrete bridge through the digital camera,comparatively the crack binary image obtained by crack image pre-processing,crack information extraction and crack feature matching with the cracks in the crack library collected in previous inspections.Finding the same crack via the highest matching value,it is convenient to carry out comparative analysis to identity the change of the bridge during different inspection intervals and to predict the future development trends of the cracks,so that inspectors are able to assess the health of bridge more accurately.The following work of are completed in this thesis : combining with the principle of multi-scale features of images,multi-scale segmentation of crack images,and then fusion at a considerable proportion,so that it can highlight the crack features,after traditional image pre-processing,that is,grayscale Transform,histogram equalization,and median filtering fuzzy denoising methods are used to pre-process the acquired crack images.The seepage model is used to extract the crack characteristics by seepage treatment,and some parameters are adjusted on the basis of the experiment,which makes it more suitable for the extraction of cracks in concrete bridges.The extracted crack information is calculated by Hu invariant moments to describe the characteristics of the crack,and the crack is compared with the crack in the crack library.Finally,the similarity measure of the crack is calculated according to the similarity measure method.,to get the best matching crack picture,to complete the analysis and comparison.At the end of the paper,an engineering example experiment was carried out to collect the cracks of the bridge deck and the superstructure of Kaiping Bridge in Kaiping City,Guangdong Province.According to the method proposed in this paper,the collected crack images were processed and analyzed,can verify high effectiveness,engineering adaptability.The experimental results were good,and considerable efficiency of the proposed method. |