Research On Binocular Vision Measurement Of Rotor UAV Based On Airborne Single Camera

With the advent of the intelligent era,visual measurement as a branch of artificial intelligence has gradually become a hot research direction.The advantages of visual measurement with noncontact passive measurement are widely used in industrial parts measurement,mobile robots,drones and other fields.In this paper,the visual measurement of the mobile monocular camera rotor unmanned aerial vehicle is taken as the research background.The two images captured by different cameras at different observation points and different angles are used to complete the binocular stereo vision measurement according to the parallax principle,and the mathematical model is established.At the current observation point position,the three-dimensional data information of a feature point of the real-time frame space object area relative to the camera coordinate system and the world coordinate system.In this thesis,the ideal perspective projection model and nonlinear imaging model of the camera are described in detail.Based on the perspective projection model,the conversion relationship between a point in the world coordinate system and its projection point in the image coordinate system is derived.The calibration method is used to calibrate a single camera and use the calibrated data to eliminate image distortion.For the common salt and salt and Gaussian noise in the image,adaptive median filtering is used for noise reduction.Then,using the HSV color model,the image segmentation of the target region,the morphological opening operation and the "filling hole processing" are completed to complete the final target region;for the selection of the target region characteristics,the matching based on the Harris feature points is selected before the matching.In order to make the distribution of key points uniform and reduce the number of key points to reduce the operation time,the ANMS algorithm is used to screen some key points.The obtained feature points are based on the nearest neighbor and Lowe distance ratio method to complete the rough matching,and then based on the homography matrix H.The RANSAC algorithm eliminates mismatching;since the external parameters of any two images acquired are unknown,the correction method of weakly calibrated image pairs is adopted,and the method has good adaptability.Finally,the binocular method is derived by trigonometric method,which shows that the monocular camera constructs a binocular stereo vision measurement system at different positions and angles;defines the relevant coordinate system in the visual measurement of the mobile monocular camera rotorcraft drone And deducing the transformation relationship between the camera coordinate system and the inertial coordinate system;deriving the model formula of the baseline distance B according to the rotary wing unmanned aerial vehicle GPS system;for the convenience of description,setting the fuselage coordinate system and camera coordinates The attitude change strategy of the system;in order to obtain the three-dimensional data information of a certain feature point of the target object area in the real-time frame at the current observation frame position in the camera coordinate system and the inertial coordinate system,the relevant mathematical model is established,and the experiment is built using the laboratory environment.The platform verifies the rationality of the model.