三维非刚体图像的特征挖掘定位算法  

Feature mining and localization algorithm of 3D non-rigid image

作　　者：潘燕[1] PAN Yan(School of Information Engineering,Fujian Vocational College of Agriculture,Fuzhou 350007,Fujian,China)

机构地区：[1]福建农业职业技术学院信息工程学院,福建福州350007

出　　处：《上海电机学院学报》2023年第5期305-310,共6页Journal of Shanghai Dianji University

基　　金：教育厅中青年资助项目(JAT191277)。

摘　　要：以实现图像特征定位为目的,设计了一种适用于三维非刚体图像的特征挖掘定位算法。利用摄像机拍摄目标的三维非刚体图像,通过重建非刚体三维结构获取运动目标的三维轨迹信息,再利用直线段检测算法挖掘运动目标三维轨迹的直线特征。将挖掘到的直线特征作为输入信息,利用图像消影点方法处理图像内目标的视觉里程数据,并推算三维非刚体图像内的目标位置,从而实现特征挖掘定位。结果表明:利用该算法构建的三维非刚体目标结构较为精确,提取直线特征信息时不易受图像模糊程度的影响,能够有效定位三维非刚体图像内的目标,定位误差数值较小。该方法具有较好的特征挖掘定位效果。To achieve image feature localization,a feature mining localization algorithm suitable for 3D non-rigid images is designed.The camera is used to capture the 3D non-rigid image of the target,and the 3D trajectory information of the moving target is obtained by reconstructing the non-rigid 3D structure.Then the straight-line segment detection(LSD)algorithm is used to mine the straight-line characteristics of the 3D trajectory of the moving target.The extracted straightline features are used as input information,and the image vanishing point method is applied to process the visual mileage data of the target in the image.The target position in the 3D non-rigid image is calculated to achieve feature mining localization.The results show that the 3D non-rigid target structure constructed using the algorithm is more accurate,and the extraction of straight-line feature information is not easily affected by the degree of image blurring.It can effectively locate the targets in the three-dimensional non-rigid image with a small localization error.This method has better feature mining and localization effects.

关 键 词：三维非刚体 特征挖掘 定位算法 直线检测算法 视觉里程数据 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]