Research On Pothole Repair Detection System Based On Kinect

At present,the various methods of repairing the pothole of asphalt pavement are in low intelligence;whereas the spray technology has huge develop space in intelligence area.Therefore,an integrated research test bench is established by combining the spray technology.The research scope of this test bench includes the 3d modeling and repair process detection system based on the vision,the trajectory repair system based on automation control and the repair process and parameter optimization system.The research of this paper mainly focuses on the 3d modeling and repair process detection system.First of all,the ranging principal of Kinect camera and the conversion relationship between these coordinate systems is introduced;the calibration of camera parameters was completed by using zhang zhengyou calibration method and Matlab calibration toolbox.Secondly,using the Kinect to collect pothole depth image,the noise and black hole problems was deal with filtering and zero pixel replacing method.Then,multiple images was used to test the inpainting effect.The verification results show that the effect of image restoration method is good.Then,in order to complete the modeling more accurately,the previously restored images were segmented by a series of steps such as enhanced contrast,and the binarization mask images that separated the pit and the background were obtained.In order to complete the image modeling of the actual coordinates,the calculation method of the 3d coordinates of the pit groove is established.Then the 3d visualization model of pit groove was established by using the three-dimensional point coordinates of the pit groove in Matlab.In order to control the parameter information of the pit groove,the parameters such as pit volume and surface area are calculated.The parameter accuracy of the model is analyzed by comparing the actual measurement with the parameters obtained by the software.Finally,the pothole repair test was carried out on the laboratory stage.The visual interface of image and parameter display was designed in the Matlab graphical user interface.This interface reads the image sequence and its volume and other was calculated according to above method.The calculation and display speed was about three pictures per second.