基于视觉传达的多帧图像特征目标跟踪仿真  

Multi-Frame Image Feature Tracking Simulation Based on Visual Communication

作　　者：韦超现 WEI Chao-xian(Nanning Normal University,Nanning Guangxi 530000,China)

机构地区：[1]南宁师范大学,广西南宁530000

出　　处：《计算机仿真》2021年第1期404-407,420,共5页Computer Simulation

基　　金：2018年度广西高等教育本科教学改革工程项目(2018JGZ127)。

摘　　要：针对当前方法提取的多帧图像目标特征精度较差,导致多帧图像特征目标跟踪准确率较低、跟踪时间较长的问题,提出了基于视觉传达的多帧图像特征目标跟踪方法。采用稀疏表示方法采集多帧图像目标特征,利用高斯分布构建图像运动模型,小波分析多帧图像灰度及细节特征,根据灰度投影法提取多帧图像目标特征,并匹配多帧图像特征点,获取多帧图像轮廓轨迹追踪目标,运用图像运动目标状态模型,求解前景轮廓的目标函数权重,实现多帧图像特征目标跟踪。实验结果表明,所提方法提取的多帧图像目标特征精度较好,能够有效降低多帧图像特征目标跟踪时间,提高多帧图像特征目标跟踪准确率。This paper proposes a multi-frame image feature target tracking method based on visual communication for improving the accuracy of multi-frame image feature target tracking and shortening the tracking time.The sparse representation method was adopted to collect the target features of multi-frame images.According to Gaussian distribution,the image motion model was constructed.The gray level and detail features of multi-frame images are analyzed by wavelet.The feature extraction of multi-frame image depended on gray projection method.The feature points of multi-frame images were matched to obtain the tracking target of contour trajectory of multi-frame images.The objective function weight of foreground contour was obtained by the utilization of image moving object state model,thus the multi-frame image feature target tracking was achieved.The simulation results show that the proposed method has high extraction accuracy and short tracking time.

关 键 词：视觉传达 多帧图像 灰度投影法 稀疏表示 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]