The Key Techniques Research Of Pavement Detection Based On Multi-sensor Data Fusion

With the rapid development of China’s economy and the increase of the number of airports,the airport runways are carrying a large number of aircraft taking off and landing every day,and diseases such as uneven and damaged runways will bring about major hidden dangers to the safety of aircraft.On the other hand,the mileage of highway is also increasing,and road pavement diseases pose a threat to the safety of vehicles and personnel.For a long time,the means of pavement detection in china is based on manual inspection,with disadvantage of long detection time,low detection accuracy and single detection equipment,which can no longer meet the needs of present highway development.Therefore,it is very significant to study the key technique involved in the current automatic detection of various pavement diseases and design a method suitable for the detection and evaluation of various pavement diseases.According to the needs of various types of pavements,combined with aircraft navigation technology and image processing technology,this paper proposes a multi-sensor data fusion pavement detection method.Integrating the data from GPS,attitude sensor,accelerometer,laser sensor,line array camera,area array camera and other sensors.The data fusion,for the reasonable matching of different road surfaces,solves the key technique in the road surface detection and realizes the automatic detection of road surface diseases.The research contents involved are as follows:1.A method of road surface detection based on sensor data fusion is designed.In the aspect of road surface undulation height measurement,the data fusion algorithm of road undulation elevation is proposed.The complementary data is used to improve the data accuracy of the attitude sensor.The Kalman filter algorithm is used to fuse the data of the sensor and GPS road undulation elevation,and the road undulation is completed.The example verification and error analysis of elevation detection;in the aspect of pavement profile detection,the principle and implementation method of data extraction of longitudinal and cross-section of pavement are given.The calculation method and flow of profile shape are analyzed in detail,and the details are derived.The calculation steps are carried out;the image of the roadbed bristles is reconstructed by the center of gravity method and the complement method,and the errors of the algorithm and the acquisition equipment are calculated and verified.2.Construct a 3-D road surface model based on data fusion.The three-dimensional coordinate system of the road is constructed,the collected discrete point data are processed functionally,the road trajectory is reconstructed by integrating GPS data,the control nodes of the road section are refined by Hermite interpolation algorithm,the three-dimensional reconstruction of the road surface is performed by NURBS masking method,and the road surface display system is developed by Open GL technology,which realizes the display of the road surface and texture mapping.3.A feature recognition and detection scheme for pavement diseases is designed.According to the disease type,different image acquisition equipment and detection methods are designed.The detection of pavement cracks adopts the preprocessing of original image noise reduction,enhancement and threshold segmentation,the maximum entropy method completes the image binarization,and the adaptive lifting method realizes the identification and classification of cracks’-D median filter is used to reduce noise in the original image,Sobel operator is used to extract the edge of the sign line,and area comparison method is used to distinguish the sign line from damage.4.A fast vehicle-mounted pavement detection platform is constructed.A road surface detection system based on sensor fusion is developed.In terms of hardware,a sensor signal acquisition circuit and a sensor signal processing method are designed,and a temperature and humidity compensation method is developed for the accelerometer to improve the sampling accuracy of acceleration.In terms of software,a set of engineering application software system for pavement detection and evaluation has been developed.The system can synchronously collect the data of various sensors in the system at high speed,calculate the undulating elevation and profile of the road surface according to the parameters selected by the user and generate a three-dimensional road surface,calculate and evaluate various indexes of different roads according to national standards,and evaluate the disease grade.