Real-time Stitching Technology Based On New Video Processor

The video stitching technique uses video image with the overlap portion from multiple output sensor which is parallel sampled and preprocessed. and then performs matching alignment based on the image feature, forms a stitched image of multiple CCD image by-re-sampling and fusion. The CCD video stitching technology can provide a360-degree protective dome with real-time and situational awareness, which can realize many functions for weapons platform, such as missile detection and tracking, weapons firing point detection, situational awareness and reminder, daytime navigation, long-range detection.The content of this paper is an embedded real-time video image stitching processing system, which can obtain a multi-channel video image through a special imaging device. It corrects lens distortion by the video image stitching processing system, where the space coordinate transformation is perfomied between the adjacent video image sequences. Finally, the video image is smoothly stitched and real-time outputting. The main works of this paper include two aspects, such as software algorithm design and hardware design.In the software algorithm design part, we designed the distortion correction algorithm for effect offish-eye, image coordinate transformation algorithm and image smoothing algorithm. First, we use the corresponding table for relationship between the optical lens aberration curve in the field of view and distortion, and the fish-eye effect distortion correction is achieved through the look-up table method. Then by using eight parameters of perspective transformation model, the fish-eye corrected image is rotationally transformed with a specified angle image, and the adjacent CCD output video image is expanded in a fiat space, so that the stitched image is fit for the effect of actual observation. Finally, the fade-out image smoothing is used for image smoothing. Simulation results show that the effect of three algorithms is very good with lower computational complexity to meet the system requirements.In the hardware design part, we complete the overall design of the system hardware by analyzing components, functions and applications of the entire stitching system. The circuit design of every unit module, the selection of electronic components and the design of electrical schematics are completed based on the layer decomposition of the overall program. It forms a computing processing unit with new video processor core, and the logic control unit is FPGA. The video stitching system is integrated by video codec chip. Following the hybrid circuit design and high-speed analog-to-digital circuit design theory, we complete the printed circuit board design of every unit module, and make the physical circuit board fitting for the relative requirements of signal integrity, power integrity and electromagnetic compatibility.Finally, we achieve the function of the real-time video image stitching by debugging of the physical circuit boards.