摘要
分析了影响算法精度的主要因素,提出了快速而有效的图像序列预处理方法,以增强投影算法的鲁棒性和提高其估计运动矢量的精度。利用连续多帧子区域的局部和全局运动矢量有效地识别出摄像机的扫描运动和背景存在的运动物体,从而去除这些无效的运动矢量,确保稳像系统的补偿精度。用机载红外摄像设备所摄取的图像序列试验,证明了本方法的有效性。
The main factors to affect the accuracy of gray projection algorithm were analyzed. The fast and efficient preprocessing method of image sequences was proposed to increase robusticity and improve accuracy of projection algorithm. The local motion vectors and global motion vectors of continuous image frames were used to detect the panning conditions and existence of moving object and then these invalid motion vectors can be eliminated to ensure the accuracy of compensation of the electronic image stabilization system. The experiments with aerial infrared image sequences was demonstrated validity of presented method.
出处
《光电子．激光》
EI
CAS
CSCD
北大核心
2003年第2期182-186,共5页
Journal of Optoelectronics·Laser
基金
中国科学院青年创新基金资助项目(Q01R07)
