In order to output the stable video in real time,this paper presented a video stabilization algorithm with mutual combination between improved feature matching algorithm and global motion compensation in line with aerial video jitter under complex background.Firstly,the algorithm used the scale-invariant SURF algorithm to extract the feature points and calculate descriptors,then,it combined with the fast approximate nearest neighbor matching algorithm to obtain the matching points,and then,through two-way matching and K-neighbor algorithm to select the best matching points,so,the matching accuracy was improved.Secondly,this paper proposed a local region matching algorithm to improve the processing speed of the algorithm and avoid the influence of other moving objects in the scene on the video stabilization effect.Finally,by establishing an affine transformation model,and solving the transformation parameters of adjacent frame images,and furthermore,carried out the global motion compensation for the images.The results show that the algorithm has fast speed,high matching precision,and good video image stabilization effect.%针对在复杂背景下航拍视频的抖动情况,为了实时输出稳定的视频,提出了一种改进的特征匹配算法与全局运动补偿相结合的视频稳像算法.首先,利用尺度不变的SURF算法提取特征点并计算描述符,再结合快速近似最邻近匹配算法得到匹配点对,并通过双向匹配以及K近邻算法筛选优秀匹配点,从而提高匹配正确率;其次,提出了一种局部区域匹配法,提高了算法处理速度,并避免场景内运动目标对稳像效果的影响.通过建立仿射变换模型,求解相邻帧图像的变换参数,进而对图像进行全局运动补偿.结果表明,该算法速度快、匹配精度高,有良好的视频稳像效果.
展开▼
