The Research On Check Method Of Aerial Data Quality For UAV

Unmanned aerial vehicle(UAV)photogrammetry is a theory and technique which integrates UAV,remote sensor,remote control,communication,navigation and positioning,image processing and other multidisciplinary technologies and quickly complete aerial photography data processing,measurement mapping,environmental modeling and analysis by getting the geographic spatial information of the target area in real time.Compared with satellite remote sensing and manned aerial vehicle surveying and mapping,this technology is featured with low cost,low risk,flexible operation,short data acquisition cycle and high precision.In recent years,it has developed rapidly in surveying and mapping production and has become an important supplementary form of traditional surveying and mapping methods.Although UAV photogrammetry has many advantages,the quality of data acquired by aerial photography cannot be effectively guaranteed for UAV is highly vulnerable to wind and other natural factors and its flight attitude changes greatly during operation due to its light weight.In order to ensure the smooth processing of UAV surveying and mapping data,the quality of the data acquired by aerial photography should be inspected.The data that does not meet the quality requirements should be rejected;and re-photography by flying the UAV again is required.However,since the existing manual inspection is complicated and it is impossible to complete data quality inspection quickly and accurately on the spot,flying the UAV again will be delayed;and postponing of construction period and increase of production cost will be caused.Therefore,the realization of the fully digitized automatic inspection of the data acquired by aerial photography is of practical significance for improving the operation efficiency of UAV photogrammetry and lowering production cost.This paper conducted a study on the automatic inspection of the data acquired by aerial photograph with UAV from the technical method of inspection and system implementation,specifically as follows:(1)Maintaining a certain degree of overlap among photographic images is the basis of the formation of stereopair for mapping.This paper applied image point feature matching method to image matching.The overlapping areas of the images can be accurately identified by means of monomorphic transformation model,which can effectively solve the difficulty in identifying the overlapping areas among UAV images due to the irregularity.Five feature extraction methods,SIFT,SURF,ORB,BRISK and AKAZE,were used in the experiment to verify the practicability of the inspection method based on image matching.The results showed that the inspection method based on SIFT can effectively calculate the degree of overlap among images at different image scales,but it is slightly slow.The ORB-based inspection method is the fastest,but it has relatively poor reliability in areas with feature points which are difficult to identify.(2)Since the flight attitude of UAV is unstable,the inclination,rotation angle and other indicators of UAV images may easily be caused.This paper proposed to inspect such indicators as the difference in the flight height at camera stations,the bending degree of the flight belt,the rotation angle and inclination of images and realize calculation by programming based on python programming language.A set of relatively perfect automatic inspection system for the data acquired by UAV aerial photography was completed.The data acquired by aerial photography in Xiangtan were used to verify various function modules.The results show that all modules of the system were correctly calculated and that the system runs stably and can inspect the quality of photogrammetry data in practice.