摘要
针对视频特技中手工消除钢丝绳的问题 ,提出了一种基于静止图像的检测和消除方法 ,并在此基础上提出了一种基于运动估计模型的改进方案 ,改进方案充分考虑了特技演员和摄像机的运动对钢丝绳位置的影响。通过实验分析了 2种检测和消除方案的准确性和适应性。实验结果表明 ,对于复杂背景下的图像 ,改进方案虽计算量较大 ,处理时间较长 ,但检测效果较好。
A kind of detection and elimination method based on still image was presented for the problem produced by manual elimination of the acrobat video rope. And then,an ameliorative method based on motion estimation model was put forward, in which the impact modes coming from the movement of both acrobat and camera were taken into account. The accuracy and the adaptability of the two detection and elimination methods were analyzed through the experiments. The results of the experients under complicted condition,the inproved scheme of image has a better defection effect,though the calculation and processing take a longer time .The work in this paper is significant for the research on automated processing of acrobat video by computer.
出处
《重庆邮电学院学报(自然科学版)》
2003年第4期5-10,共6页
Journal of Chongqing University of Posts and Telecommunications(Natural Sciences Edition)
基金
国家自然科学基金资助项目 ( 6 0 2 0 2 0 0 5 )
关键词
视频特技
钢丝绳
目标检测
运动估计
video acrobat
rope
object detection
motion estimation