基于SURF的老电影抖动修复技术研究  

Old filmjitter repair technology based on SURF

作　　者：吴兵[1,2] 丁友东[1] 宫若彬 

机构地区：[1]上海大学上海电影学院,上海200072 [2]泰山学院文学与传媒学院,山东泰安271000

出　　处：《电视技术》2017年第6期110-116,共7页Video Engineering

基　　金：国家自然科学基金项目(61402278);上海市科委科技攻关项目(14511108400);教育部人文社会科学项目(16YJC760056);上海大学电影学高峰学科项目(2016GF001);山东省社会科学规划研究项目(15DWYJ06)

摘　　要：针对老电影的帧间抖动现象,研究一种基于SURF的全局抖动修复方法。该方法首先依据老电影抖动的特点建立对应的数学模型,然后运用加速健壮特征(SURF)算法进行特征点匹配,并针对摄像机是否存在主观运动提出相应的修复方案,同时运用运动阈值策略去除误匹配,最后计算位移均值作为全局运动参数,通过反向平移运动参数完成抖动修复。在VS2013中使用Open CV实现了相关算法,并详细介绍了系统的功能和框架。实验给出的电影去抖动效果证明了该方法的有效性。This study proposes a framework to solve the jitter problem in old films. Firstly, according to the characteristics of the old films jitter, a corresponding mathematical model is established in this paper; Secondly, the speed-up robust features（SURF） algorithm is used to match the feature point, and the corresponding repair scheme is put forward according to the subjective motion of the camera. Meanwhile, a motion threshold strategy is added to remove error matching; Finally, mean moving value of the points which selected in previous step is calculated as the global motion parameter, then complete the restoration by reverse translation of the motion parameter. The relevant algorithm was designed by VS2013 and OpenCV, function and framework of the system is given as well. Results show that the method can effectively remove global jitter in old films.

关 键 词：视频稳像 SURF算法 运动估计 电影修复 抖动修复 

分 类 号：TN911.72[电子电信—通信与信息系统]