期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2017
卷号：5
期号：3
页码：4734
DOI：10.15680/IJIRCCE.2017.0503203
出版社：S&S Publications
摘要：This paper gives a new real-time video stabilization approach for Unmanned Aerial Vehicles (UAV).The main aim of digital video stabilization is to get rid of unwanted movements, undesirable jiggle, blur and poorquality video. Many video stabilization techniques hence are developed with different algorithms & methods. Thepaper presents the review of different methods acquired for the motion estimation and video stabilization. Micro AirVehicles (MAV) at the moment are a days are very popularly and are used for huge range of programs the sort ofsurveillance, exploration and so forth. MAVs are typically equipped with a virtual camera on board it and the videoscaptured via it commonly involve undesired jitter. This paper particularly makes a specialty of stabilizing the volatilevideos; OPENCV is used to process those videos. This video stabilization in particular includes estimation of the digitalcamera motion route and smoothing motion path.In this paper, we stretch out picture sewing to video sewing for recordings that are caught for a similar scene atthe same time by numerous moving cameras. By and by, recordings caught under this situation frequently seem flimsy.Specifically applying picture sewing strategies for shaking recordings frequently experiences solid spatial and worldlyrelics. To take care of this issue, we propose a bound together structure in which video sewing and adjustment areperformed mutually. In particular, our framework takes a few covering recordings as data sources. We evaluate bothentomb movements and intra movements. At that point, we understand an ideal virtual 2D camera way from everysingle unique way. An augmented field of view along the virtual way is at long last acquired by a space-fleetingadvancement that takes both bury and intra movements into thought. Two critical parts of this improvement are that: 1)a network based following technique is intended for an enhanced strength, which produces highlights that are circulateduniformly inside and over various perspectives and 2) a work based movement model is embraced for the taking care ofof the scene parallax. Some test results are given to show the viability of our approach on different purchaser levelrecordings.
关键词：Micro Air Vehicle; Video Stabilization; Motion Estimation; Random Sample Consensus
