Concrete Surface Pitting Detection Algorithm Based On Mathematical Morphology

Because of many reasons such as raw materials,construction technology,external environment and so on,the surface of concrete will more or less have pitting surface,honeycombs,holes,cracks,stains,chtomsyiv sberration,which will affect the appearance quality of hardened concrete.In order to improve the appearance quality of concrete,it is necessary to carry out many technological tests,by observing and analyzing the defect changes after each test,to improve the technological parameters of construction.At present,in the process of construction and test,the artificial observation is the most popular method to detect the defects such as pitting surface,honeycombs,holes and so on.In this method pits are marked with color markers on the surface of concrete,followed by observaitons,measuring and statistical analysis of the pits.This method has some problems,such as low efficiency,strong subjectivity,inaccurate results,and it will leave stains on the surface of concrete.In order to solve the problems,this paper proposes a method based on mathematical morphology to detect the pits on concrete surface,which can effectively eliminate the problems of artificial observation.The method consists of three modules: image acquisition,scale recognition and pits recognition.The image acquisition module is to place a scale vertically on the concrete surface and take pictures with a high-definition digital camera to obtain high-definition images.The scale recognition module has two functions: one is to recognize the coordinates of the scale in the image,extract the scale from the image,and then divide the whole image into two separate parts: the scale and the concrete surface;the other is to extract the scale lines in the image by mathematical morphology method,and detect the interval between each two scale lines.The number of pixels is calculated by combining the frequency distribution histogram function and the standard deviation function to get the average value after eliminating the offset value,and the number of pixels in unit length is obtained.Therefore the actual area of a single pixel is calculated,which provides a basis for determining the size of the whole image and the area of the pits points.Pits recognition module identifies,counts and analyses the pits on the concrete surface in the image.The area with scale shadows in the image is separated from the area not affected by shadows.The two parts are processed by different image inhancement methods.After binarization,the two parts are merged into a complete image.Mathematical morphology is used to identify the pits.The total number of pits and the number of pixels of a single pit are counted by the method of extracting connected components,and the actual area of the pits is calculated.Finally,the number of pits whose area bigger than 1 square millimeter and an area bigger than 2 square millimeters is calculated.In this paper,the testing program is programmed by using MATLAB,and the method is tested.A total of 20 actual images were selected as input in the test,and the results ofsimulation were obtained.In the test,morphological gradient method was used to draw the edge contour of the pits in the original images,which is convenient for manual verification of the test results.Through manual verification,it is proved that the algorithm has high accuracy for scale detcting and high rate and accuracy for pit detecting,which can meet the needs of statistical analysis for pit in construction and experiment.