Research On Video Mosaic Technology Based On Improved SURF And RANSAC

Video mosaic is the process that splices a series of video frames with overlapping regions collected by multiple cameras,and finally synthesized into a video frame with a wider viewing angle.Compared with the video frames obtained by the monocular camera,the video obtained by splicing has larger video frame and more information.Video mosaic has a wide range of application prospects in the field of intelligent monitoring and videoconferencing.Therefore,research on video mosaic technology is of great significance.From the collection of video images to the final stitching,it mainly involves feature point detection and extraction algorithm,image registration algorithm and image fusion algorithm.There are mainly two problems including real-time and visual effects in the process of its implementation.To meet the application needs of modern intelligent monitoring system,a video mosaic algorithm based on improved SURF and RANSAC is proposed.Firstly,the improved SURF algorithm is used to detect and extract feature points.Then,the improved RANSAC algorithm is used to eliminate the mismatched pair of points.Finally,the gradual weighted average fusion method is used for fusion processing to generate wide-scene video.The main research contents of this article include:(1)Introduce the main technical principles in the field of image mosaic in detail.Mainly discuss and analyze the basic process of image splicing,camera imaging geometry,image mosaic basic transformation model and image registration technology,and determine the algorithm idea of registering the to be spliced video image with feature point registration technology.(2)Analyze the classic image feature detection algorithm SURF algorithm in detail,and propose improved SURF based on the original algorithm.The improved SURF algorithm firstly reduces the dimensionality of the generated feature descriptors.Then,the ROI is set to define the detection area of feature points.Finally,the feasibility and superiority of the algorithm are verified through experiments.(3)After the SURF feature extraction,image registration is required.For this reason,the traditional coarse matching method and purification method are analyzed,and the traditional RANSAC algorithm is improved.The improved RANSAC algorithm first performs pre-filtering on the matched feature points.Then,by partitioning the area where the feature points are located,the problem of poor accuracy and instability of algorithm caused by the closeness of the feature points are avoided.Experimental results show that the improved RANSAC algorithm has higher operating efficiency and stronger robustness.Then,the classical image fusion algorithm is analyzed.By comparing the fusion results,the gradual weighted average fusion method is used for video fusion to achieve dynamic video splicing.(4)Set up the video splicing experimental simulation environment.Complete the video splicing system program design based on the algorithm proposed in this paper by using Visio Studio 2008 and OPENCV4.1,and the dynamic splicing of two video frames is realized.The simulation results show that the proposed algorithm can effectively achieve the video mosaic,and the generated video has better visual effects and higher real-time performance.