Research On Remote Multi-channel Video Acquisition,Transmission And Splicing Technology For Large Scenes

In the industrial and military fields,it is often necessary to obtain remote large-scale scenarios in real time,such as industrial park monitoring and battlefield situation analysis.In order to obtain images of remote high resolution and large field of view,a single camera can no longer meet the actual requirements,and existing panoramic cameras and wide-angle cameras are expensive,and the collected images are severely distorted and the shooting distance is close to meet the requirements.Therefore,it is a feasible solution to obtain a remote,high-quality and large-scenes video by collecting video with multiple common cameras and then splicing at the remote end,which is a research hotspot in the field of large-scenes video acquisition.In order to achieve fast acquisition of multi-channel video and reduce the system cost,this paper designs a multi-channel video acquisition system with FPGA as the control core.The system mainly includes a camera acquisition module,a channel selection module,a video decoding module,a data processing module,a video buffer module and the like.The system uses analog camera to obtain video;uses TW2867 chip that supports four-channel video decoding function for video decoding;uses two DDR2 chips for video data storage;uses Verilog language to complete system function development.In order to realize fast and long-distance transmission of multi-channel video,this paper designs Gigabit Ethernet for remote transmission of video data.Firstly,based on FPGA,the GMII interface with Ethernet PHY chip RTL8211 EG is implemented,and the design of PHY chip controller module is completed.Then,based on this module,the video stream data Ethernet frame transmission based on UDP network protocol is realized by state machine.Finally,by setting the frame start tag for each frame of Ethernet data,the packet loss problem caused by the unreliability of the UDP protocol is effectively overcome,and the differentiation of video streams of different channels is realized.In order to realize the rapid splicing of images,this paper mainly studies the splicing algorithm based on SURF features.Firstly,the cylindrical images of the acquired multi-channel video images are projected onto the same plane,then the SURF feature points are detected,and the detected feature points are coarsely matched.Then,using RANSAC algorithm to accurately register,for the stitching problem caused by the stitching process,the image is merged by the fade-in method,and finally a large-resolution image with high resolution and wide field of view is obtained.In order to test the performance of the system,this paper uses Wireshark software to capture the Ethernet frame,and analyzes the frame data to verify the correctness of the network transmission module.The network transmission rate measured by the Ethernet speed measurement module can be up to 908 Mbps.Finally,the multi-channel video real-time display interface and the image stitching display interface are designed by C++ language and MFC program.The test results show that the multi-channel video picture is clear and jitter-free,and the display of one channel video can reach 40 frames/s,with real-time performance;no stitching seams.The test results show that the system can meet the needs of remote,high quality,large scene image acquisition.