Research On The Posture Measurement Method Of Projectile Based On ZYNQ Binocular Vision

The era of Industry 4.0 has arrived,and visual measurement technology has also played an important role in the era characterized by intelligence.As an important branch of image vision technology,stereo vision has been used in various fields due to its unique non-contact,low cost and high precision.The research of this subject is to break through the limitations of a single FPGA or ARM,complete the parallel acquisition,storage and display of the binocular camera with high performance,and apply the binocular stereo measurement technology to the missile attitude measurement.The project uses ZYNQ-7000 as the development platform,calls chip resources,through the collaborative design of software and hardware,according to the four functions of image acquisition,data transmission,image storage and image display,complete the construction of binocular vision system.FPGA-based PL aspect:writing and designing corresponding functional modules;completing image acquisition of two cameras in parallel through important synchronization signals;writing FPGA logic,using AXI bus to complete efficient transmission of image data between modules;Achieving binocular display by logic writing.ARM-based PS and PS-PL interconnection:Writing a C-based control language,completing the initial configuration,interrupt configuration,and drive of the entire system through hardware and software interconnection resources,and implement image three-cache mode to improve transmission and display quality.For the aspect of image processing,the principle of binocular stereo vision is studied.Through the corresponding feature point detection,target extraction,stereo matching and 3D reconstruction image processing,the central axis method is used to complete the attitude detection based on binocular vision.Finally,through debugging and experiment,the measurement error of the attitude of the missile is within 0.6~?,which verifies the feasibility of the attitude measurement method based on ZYNQ binocular vision.