The Pose Measurement For Rocket Nozzle Based On Feature Point Cloud

It is important for controlling the rocket flight attitude to measure threedimensional motion parameters of the rocket nozzle.The technology of computer vision measurement has advantages on large-space and non-contact.Generally,we put markers on the surface of nozzle to complete the measurement.However,when we launch the rocket,markers will deform and have bad effects on the accuracy of the survey.Feature points can reflect geometrical characteristics of targets which are unaffected by external conditions such as illumination and so on.This paper design a binocular visual pose measurement scheme based on feature point cloud.Firstly,this paper introduces the camera imaging model and calculates internal and external parameters by calibrating the camera based on a 2D planar target.During the actual measurement,the lens may distort.We correct coordinate points through using the principle of radial distortion.Then,an improved method of computing spatial point coordinates is derived.We use an approach to sharpen and denoise images.To reduce the environmental interference,the objects are separated from background by using the means of target segmentation.Secondly,aiming at the problem of extracting and matching feature points,the improved SIFT(Scale Invariant Feature Transform)algorithm is proposed.We use the SIFT algorithm to extract feature points and match these points by applying the k-d(kdimensional)tree strategy.Then,the RANSAC(Random Sample Consensus)algorithm is used to optimize the matching results.At the moment,the target feature point cloud data is obtained.Thirdly,the three-dimensional reconstruction method of feature point cloud is studied.We use the Crust algorithm which is based on the bounding box to complete the reconstruction.At last,we use the PCA(Principal Component Analysis)method to obtain the target pose information.Finally,this article build an experimental platform based on binocular vision measurement.We apply the proposed algorithm to measure the nozzle model.Experiments show that the method has higher measurement effects and can obtain the position information successfully.