汉英对照:
Chinese-English Translation:
视频抽帧就是在一段视频中,通过间隔一定帧抽取若干帧的方式,模拟每隔一段时间拍摄一张照片并接合起来形成视频的过程(即低速摄像)。相比单纯快进会有不一样的感觉。 视频抽帧效果经常会和我们常说的延时摄影、低速摄像被混为一谈。
Video frame extraction is to extract several frames at a certain interval in a video, to simulate the process of taking a picture at a certain interval and combining them to form a video (i.e. low-speed camera). Compared with simple fast forward, it will have a different feeling. Video frame pulling effect is often confused with what we often call delay photography and low-speed photography.
视频抽帧就是在一段视频中,通过间隔一定帧抽取若干帧的方式,模拟每隔一段时间拍摄一张照片并接合起来形成视频的过程(即低速摄像)。相比单纯快进会有不一样的感觉。 视频抽帧效果经常会和我们常说的延时摄影、低速摄像被混为一谈。
Video frame extraction is to extract several frames at a certain interval in a video, to simulate the process of taking a picture at a certain interval and combining them to form a video (i.e. low-speed camera). Compared with simple fast forward, it will have a different feeling. Video frame pulling effect is often confused with what we often call delay photography and low-speed photography.
抽帧存储的定义是对历史图像进行抽帧存储是可较长时间保留关键帧(I 帧)的存储方式。先全量存储所有录像；当保存时间超过全量存储留存期后，系统将逐步删除非关键帧，减少存储容量，有效延长视频信息的保存时间。抽帧之后画面清晰度不变。
The definition of frame extraction storage is that the key frame (I frame) can be retained for a long time by frame extraction storage of historical images. First, store all videos in full amount; when the storage time exceeds the retention period of full amount storage, the system will gradually delete non key frames, reduce the storage capacity, and effectively extend the storage time of video information. The definition of the picture does not change after the frame is drawn.
视频，严格意义就是逐帧播放单幅画面，利用肉眼的视觉暂留特性，对视觉产生连续动画错觉。抽帧，就是从这些画面中抽出单幅画面。
The strict meaning of video is to play a single picture frame by frame, using the visual persistence of the naked eye to produce the illusion of continuous animation. Frame drawing is to draw a single picture from these pictures.
抽帧存储
Frame extraction storage
对历史图像进行抽帧存储是可较长时间保留关键帧(I 帧)的存储方式。先全量存储所有录像；当保存时间超过全量存储留存期后，系统将逐步删除非关键帧，减少存储容量，有效延长视频信息的保存时间。抽帧之后画面清晰度不变。
The key frame (I frame) can be saved for a long time by frame extraction storage of historical image. First, store all videos in full amount; when the storage time exceeds the retention period of full amount storage, the system will gradually delete non key frames, reduce the storage capacity, and effectively extend the storage time of video information. The definition of the picture does not change after the frame is drawn.
高清监控时代的到来，整个视频监控系统的成本开始大幅度上升。这其中，不仅仅是前端高清摄像机成本的提升，随着图像清晰度从标清向高清的提升，后端存储设备的成本更是上涨惊人。数据显示，在采用传统的 DVR 和 NVR 解决方案时，存储设备在整个视频监控系统的成本大致为 15%～30%；而在数字网络高清视频监控系统中，存储设备在整个系统中的造价则将上升到 50%～70%。根据相关规定，公共场所的监控录像至少需要保存一个月（30 天）以上，有些地方则更久，由此带来的海量数据存储问题可想而知。
With the advent of the era of high-definition monitoring, the cost of the whole video monitoring system began to rise substantially. Among them, not only the cost of the front-end HD camera has increased, but also the cost of the back-end storage device has increased dramatically with the improvement of image definition from SD to HD. The data shows that when traditional DVR and NVR solutions are adopted, the cost of storage equipment in the whole video monitoring system is about 15% – 30%; while in the digital network HD video monitoring system, the cost of storage equipment in the whole system will rise to 50% – 70%. According to the relevant regulations, the monitoring video in public places should be kept for at least one month (30 days), and in some places for a longer time. The massive data storage problems caused by this can be imagined.
定码率抽帧存储技术
Fixed rate frame extraction storage technology
根据现在主流的压缩方式计算，监控录像存储大概所需空间如下：
According to the current mainstream compression method, the monitoring video storage space is roughly as follows:
H.264 压缩方式：HAFLD1 清晰度，实时 25 帧，300M/小时/路；
H. 264 compression mode: hafld1 definition, real-time 25 frames, 300m / hour / way;
H.264 压缩方式：CIF 清晰度，实时 25 帧，200M/小时/路；
H. 264 compression mode: CIF definition, real-time 25 frames, 200m / hour / way;
根据上述数据，我们基本可以估算一路视频监控图像所需的存储空间。举例来说，某银行需要 1 路 D1 视频监控系统，7*24 小时监控，数据保存 3 个月便可以覆盖，那么总计容量需求 1.1TB。
According to the above data, we can basically estimate the storage space needed for a video surveillance image. For example, a bank needs a channel of D1 video monitoring system, which can be monitored 7 * 24 hours and covered after three months of data storage, so the total capacity demand is 1.1tb.
如果我们已存的录像并不是跟新的视频一样重要，就可以采取降低老录像的记录帧率。这样就可以极大地减少老视频对存储空间的需求。同时，如果需要的话，仍然可以调用老录像作为证据。
If the video we have saved is not as important as the new video, we can reduce the frame rate of the old video. This can greatly reduce the storage space requirements of the old video. At the same time, the old video can still be used as evidence if necessary.
采用定码率抽帧存储技术，如一个月内的图像存储 25 帧/S 的录像文件，在不影响录像记录信息的情况下，可以将其降低为 15 帧/S，这样也可以大大节省存储空间。
Using fixed bit rate frame extraction storage technology, such as 25 frames / s video files stored in one month’s image, it can be reduced to 15 frames / s without affecting the recording information, which can also greatly save storage space.
降低帧率以减小存储数据，比如原本 1 秒 25 帧，25M 数据，变为 1 秒 8 帧，8M 数据了。在非压缩视频传输中会造成画面卡顿的视觉感受，在特殊场合，比如带宽极为受限，或者只有静态图像等才会采用。抽帧存储最大的好处是节省存储空间。现在高清化趋势下，带宽的占用和硬盘空间都受到了很大的挑战。虽然硬盘也在不断的变大，但是一旦系统配备的摄像头很多，存储还是一个成本中很大的一部分。智能摄像机能做到在背景没有什么变化的时候，自动降低存储帧率，既不丧失数据，又避免了过多的冗余数据和信息。
Reduce the frame rate to reduce the stored data. For example, the original 25 frames per second, 25m data, has changed to 8 frames per second, 8m data. In the uncompressed video transmission, it will cause the visual feeling of picture jam. In special occasions, for example, the bandwidth is extremely limited, or only static images can be used. The biggest advantage of frame extraction storage is to save storage space. Now, in the trend of HD, the bandwidth and hard disk space are greatly challenged. Although the hard disk is also growing, but once the system is equipped with many cameras, storage is still a large part of the cost. The intelligent camera can automatically reduce the storage frame rate when there is no change in the background. It can not lose data, but also avoid too much redundant data and information.
