Research On Binocular Attitude Measurement And Multispectral Binocular Ranging Technology

With the development of computer vision and image processing technology,visual stereo measurement has also received increasing attention from researchers.In the field of aerospace,the use of computer vision to estimate the relative attitude of high-speed aircraft is an important part of aircraft aerial refueling,target tracking and positioning.However,there are still some problems in the existing aircraft attitude measurement methods,which are not suitable for the target objects in motion.Moreover,the surface of the aircraft is generally lack of texture information and color information.In some cases,it is not possible to add a mark on the surface.How to perform feature extraction is also a problem worth discussing.In the study of this problem,it is found that when the target is in a complex or similar color environment,it is difficult to perform feature extraction and stereo matching on the target object.When performing binocular image feature extraction,the texture,color,or edge features of the image are generally extracted.However,when the target object is very similar to the background,texture duplication may occur,colors may not be resolved,and edges may be blurred,resulting in extracted features.Inaccurate,there is an error in the final measurement result.To solve this problem,this paper considers the use of multi-spectral detectors to complete binocular stereo vision measurement.For the relative attitude measurement of the needle aircraft,this paper proposes the application of binocular stereo vision measurement technology to achieve the relative attitude measurement of the target object.Firstly,a binocular visual relative attitude measurement experiment platform is established,the consistency and stability of the left and right detectors are ensured on the hardware,and the synchronization of the left and right views is ensured on the software.The two cameras were calibrated using the Zhang calibration method,and the left and right views acquired were corrected to remove the distortion and calibrated in rows.The Speeded UP Robust Features of the image is extracted,and a feature descriptor is generated,and the left and right views are matched according to the descriptor.Through the matching of the real-time pose and the standard pose,the spatial rotation and translation matrix between the two poses are obtained by using the point set registration method,and the final attitude transformation angle and the translation amount are obtained by the method of singular value decomposition..Through experiments,more accurate measurement results were obtained.This method is a sparse point matching method.It has fast running speed and good real-time performance.It has the advantages of simple implementation and high practical value.It can be widely used in visual measurement.For the problem of high similarity between the target object and the background,it is difficult to extract the feature.This paper proposes a new method for binocular stereo measurement using a multispectral detector.The use of mosaic-coated multispectral detectors to form a binocular experimental system,for the problem of differences in the spectral curves extracted at different spatial locations,first spectral analysis of the experimental scene to ensure the feasibility of the program.Preprocessing the mosaic multispectral image and recovering the multispectral data cube.The spectral features such as spectral angles and spectral correlation coefficients were combined with the traditional image feature extraction and stereo matching algorithms to complete the parallax calculation of multi-spectral binocular images and achieve multi-spectral binocular range finding.This article designed a targeted experimental scenario,carried out many experiments,and achieved the expected experimental results.It shows that this method has certain practical value,and it is helpful to solve special features(flat areas,repeated similar structures,etc.),and the feature extraction and matching of binocular images that obscure scenes.