Question

帮帮忙吧，英语翻译，要语句通顺，明天急用！！！

Ferman and Tekalp [5] extract keyframes by clustering, and then go on to prune as well as augment the number of frames identified as keyframes, using a number of criteria. Firstly, since lighting changes can bias the average histogram adversely, outliers with respect to luminance are added to the list of keyframes. Secondly, clusters that contain only a few frames are eliminated. Thirdly, clusters with centers closer in color distance than a threshold TC are merged. Finally, within each cluster, if the cluster radius exceeds TC two representative frames are chosen that are (1) closest to the center, and (2) farthest from the center. The authors use the Y-Cb-Cr color space.
We have found that applying some of the techniques above yield poor results for keyframe extraction in quite common situations. E.g., we considered the sequence shown in Fig. 1 (see http://www.cs.sfu.ca/~mark/ftp/AcmMM00/) and found that lighting changes seen there as the child steps out of the shadows confound the algorithm in [5] because of the change in pixel values under different lighting. The method in [5] applied to such real outdoor lighting change from shadowing produces many keyframes,when in fact we wish to see only one.

Answer 1

Ferman和Tekalp[5]提取关键帧的聚类,然后去修剪以及增强的帧数确定为关键帧,使用数量的标准。首先,由于照明变化可以偏见平均直方图不利,异常值对亮度被添加到列表的关键帧。其次,集群,只包含几帧了。第三,集群与中心距离接近颜色比阈值TC合并。最后,在每个集群,如果集群半径超过TC两帧,选择代表(1)最靠近中心,和(2)从中心最远。作者使用y cb cr颜色空间。
我们发现,应用上面的一些技术为关键帧提取产量差的结果在很常见的情况。如。,我们认为是序列图1所示(见http://www.cs.sfu.ca/ ~马克/ ftp / AcmMM00 /),发现照明变化认为是孩子走出阴影的混淆算法在[5]因为像素值的变化在不同的照明。该方法在[5]应用于这样的真正的户外照明变化从阴影产生许多关键帧,而事实上我们希望看到只有一个。