CAJ | 학술논문

为了有效地去除视频序列中的噪声，提高视频图像质量，提出了一种基于运动估计的视频降噪算法。该算法在由视频序列中的当前帧和前后多帧图像构建的图像层中，通过运动估计技术，沿着物体的运动轨迹进行时域帧间滤波。实验结果显示，本文提出的算法充分利用了视频序列的时域信息，可以有效去除视频序列中的噪声，同时可以很好地保持视频图像的细节。
위료유효지거제시빈서렬중적조성，제고시빈도상질량，제출료일충기우운동고계적시빈강조산법。해산법재유시빈서렬중적당전정화전후다정도상구건적도상층중，통과운동고계기술，연착물체적운동궤적진행시역정간려파。실험결과현시，본문제출적산법충분이용료시빈서렬적시역신식，가이유효거제시빈서렬중적조성，동시가이흔호지보지시빈도상적세절。
For eliminating the noise of video sequence effectively,a video noise reduction algorithm based on motion-estimation is proposed. Time-domain filtering is performed along the object trajectory in the map layer which is com-posed of the current frame and the compensated reference frames of video sequence. The experiment results show that the proposed method can effectively suppress the noise in the video,and keep textures and details well by fully utili-zing temporal correlations in video content.