The three-dimensional (3D) reconstruction of moving image sequences in video surveillance is always the re-search importance of 3D reconstruction. But the current 3D reconstruction method adopts the volume rendering method to realize the 3D reconstruction of two-dimensional motion image sequences in video surveillance by means of acquiring the moving image sequences in two-dimensional video surveillance,and adopting Java Applet and Java Application programming,which can′t guarantee the reconstructed image quality,and has low image resolution. Therefore,an image feature points extraction and matching based 3D reconstruction method of moving image sequences in video surveillance is proposed. The feature points of mo-tion image sequences in video surveillance are detected,and their image texture in a certain neighborhood,structure and other characteristics are calculated. The characteristics of the feature points of the motion image sequences are compared to realize the feature points extraction and matching of feature points of the motion image sequence. The structure and motion of the moving image sequence in video surveillance are initialized,and the self-calibration of the video surveillance camera is performed to realize the 3D reconstruction of the moving image sequence in video surveillance. The experimental results show that the method can in-crease the definition of the reconstructed moving image sequences in video surveillance effectively,reduce the 3D reconstruction process,improve the reconstruction efficiency of moving image sequence,and has high application value.%视频监控中运动图像序列三维重建一直是三维重建研究的重点,其重建效果会对运动图像的清晰度产生影响.但当前三维重建方法都是通过获取二维视频监控中的运动图像序列,通过基于Java Applet与Java Application编程,采用体绘制法完成视频监控中二维运动图像序列的三维重建,该方法无法保证重建后的图像质量,导致图像清晰度不高.为此,提出基于图像特征点提取与匹配的视频监控中运动图像序列三维重建方法.首先,对视频监控中运动图像序列特征点进行检测,并对特征点一定邻域内图像的纹理、结构以及其他特征进行统计,通过对运动图像序列特征点的特征比较,完成运动图像序列特征点提取与匹配;然后,对视频监控中运动图像序列的结构和运动初始化,并进行视频监控相机的自标定,实现对视频监控中运动图像序列的三维重建.实验结果表明,所提方法能够有效提高三维重建后视频监控中运动图像序列的清晰度,减少三维重建流程,提高运动图像序列重建效率,具有良好的使用价值.
展开▼
