Key Technology Research On Pose Measurement Based On Monocular Vision

In recent years,with the increasing requirements of the autonomous operation in the space exploration,vision-based navigation technology is increasingly valued and widely used.The position and attitude of the target is an important parameter to reflect its space state.It's an indispensable section for the high-precision and real-time measurement of the target in the operational tasks about autonomous rendezvous and docking and manipulator crawl.The pose measurement system based on monocular vision becomes the primary means of measurement especially at close range;therefore it has inestimable value to do in-depth study about the technology of the measurement based on monocular vision.This article focuses on in-depth study about the relative position and orientation of close-range space target with high-precision and real-time properties.A ground testing system about pose measurement based on monocular vision measurement system was designed ultimately.The main research results of this paper are as follows:(1)A variety of camera imaging model were compared,while introducing the non-linear imaging model including tangential distortion into the Zhang's calibration method in order to make the improved method fully describe the internal characteristics of the camera and make precise geometric distortion correction of the target image.(2)In the preprocessing stage about the extraction of the target image features,an improved means of adaptive median filter and adaptive threshold to obtain each communication area of the characteristic points was obtained.Then we can get the boundary points information of the connected region using contour tracing algorithm.In the stage of sub-pixel positioning about the center of the target features,we focuses on the gravity method and ellipse fitting method,analyzing the criteria about the two methods in different application environments.(3)It makes a systematic comparative analysis of various algorithms based on four coplanar feature points from the operation speed,estimate precision and anti-noise-interference ability.The simulation results show that LUT algorithm combines the features of various aspects.It subtly achieves a balance between the operation speed and estimation accuracy,while effectively eliminating the pose ambiguity problem caused by the planar objects.(4)By building a ground testing system about pose measurement based on monocular vision measurement system which includes some software and hardware components,we get the spatial position and orientation of the target using the LUT algorithm in real-time.Finally,a comprehensive assessment about the translational and rotational measurement accuracy of the system was carried out.