视觉传达下多帧影视图像特征实时跟踪仿真

摘要

对多帧影视图像目标进行实时跟踪,可以增强影视后期处理的稳定性,提升影视画面的整体质感.但采用当前方法进行特征实时跟踪过程中,难以克服多帧影视图像复杂背景的影响,存在影视图像特征提取精度较低的问题.为此,提出一种基于粒子群优化的视觉传达下多帧影视图像特征实时跟踪方法.上述方法首先利用小波变换对多帧影视图像中的灰度和细节特征进行分解,结合灰度投影法对影视图像目标进行特征提取,建立相应的特征图,将多帧影视图像特征跟踪问题转变为图像目标前景与背景的分类问题,以目标前景与背景区域图像特征分布方差间的比值函数为判定图像目标区域与背景区域的分离系数,利用粒子群算法对多帧影视图像特征分量进行加权处理实时获取图像目标跟踪特征,由此实现多帧影视图像特征实时跟踪.实验结果表明,所提方法实现了对目标的快速提取,具有较快的运算速度,足以实现目标实时跟踪.%Focuses on real-time tracking method for feature of multi-frame film and television image under visual communication,the research used wavelet transform to decompose gray level and detail feature in the image,and carried out feature extraction for target of the image integrated with gray projection method.Then it built corresponding feature graph and converted feature tracking problem of the image to classification problem of foreground and background of image target.The research used function of specific value among distribution variance of image feature in area of foreground and background of image as separation coefficient judging the area of foreground and background.The research also used PSO to carry out weighted processing for component of the image feature and acquired tracking feature of the image target in real time.Experimental results show that the method can achieve rapid extraction of target and has faster arithmetic speed.It can achieve real-time tracking of target enough.