基于运动自适应的非均匀比率图像扩展算法

Image extension algorithm with non-uniform ratio based on motion adaptation

讨论了传统的4∶3格式的标准清晰度电视信号转换为16∶9格式的高清晰度电视信号的运动自适应的非均匀比率扩展变换算法。算法结合视频图像的运动检测,采用K-均值聚类算法,根据图像的运动重心(motioncenterofgravity-MCOG)选择合适的扩展比率,并针对不同的图像区域进行扩展变换。算法符合人的视觉系统对图像的运动区域具有较强感知度的视觉特性,有效的减小了视频格式转换的图像变形。实验表明了本文提出算法的有效性。

The challenges of video format conversion from SDTV4∶3to HDTV16∶9are reducing the distortion of the con-version.We describe a novel scheme of motion adaptive video format conversion from SDTV4∶3to HDTV16:9by using the K-means algorithm in this paper.A competitive learning algorithm,the K-means algorithm,is used to to estimate the image motion center of gravity(MCOG).According to the value of MCOG,we adopt corresponding image aspect ratios in the differ-ent regions of image.The algorithm exploits the human visual-perception that the motive region is more sensitive than others.And the algorithm effectively reduces the distortion of the conversion.Experimental results of the proposed algorithm shows its effectiveness.[