图像编码 综述

MPEG and ITU-T VCEG

In the beginning there was COST 211 (don’tworry, they don’t exist anymore). They created the H.120 standard, whichbasically flopped. But it paved the way for what we have today.

The ITU-T VCEG (Video Coding Expert Group) formedto improve upon H.120, and that’s all we have to know about that.

MPEG (Moving Pictures Experts Group) formed tofind a way to incorporate codecs for broadcast work. To thisday, the two committees – MPEG and VCEG, work side-by-side. MPEGspecializes in broadcast (television), while the ITU (InternationalTelecommunications Union) focuses on telecommunications (phone, internet).

In today’s world, their goals mostly overlap(because everything is going the way of the Internet), and this association is likelyto continue.

JPEG (Joint Photographic Experts Group) becamea popular codec of choice for images right about that time, and in the same‘vein’, MPEG was formed to take care of the broadcast industry’s needs.

MPEG-1

MPEG adopted H.261 and JPEG together to formwhat is called a ‘Suite’. The specific name for the first suite is MPEG-1.

Parts and Layers

MPEG suites have sub-divisions, called Parts.Traditionally, Part 1 is always for the ‘System’ (file format). Part 2 is forvideo, and Part 3 is for audio. Look at the image and you’llsee this clearly.

MPEG-1 Part 2 is also H.261, for ourpurposes.

Parts are further sub-divided into Layers.Audio has three or more layers, called Layer I, Layer II and Layer III and soon.

MPEG-2

In 1999, technology had advanced enough towarrant an upgrade of the suite.

Part 1 of the MPEG-2 suite had two majorclassifications: Program Streamand Transport Stream.

Part 2, video is also called H.262. It hadadditional support for interlacing and 4:2:2. The bigchange for audio was the addition of 5.1 channels, and MP3was revised to incorporate this specification.

Most of broadcast television adopted MPEG-2,and it is still the most widely used codec today for broadcast. Additionally,DVD incorporated this technology, and was able to include surround sound as aresult.

MPEG-2 also had more than three parts. Themost important of these (not shown in the image) is Part 7, called the AdvancedAudio Coding (AAC) audio format.

Professional camera codecs that use MPEG-2include HDV and XDCAM.

MPEG-4

By 2004 the world had suddenly discovered theInternet. MPEG was ready with the third suite, MPEG-4 (MPEG-3 was not used). .

Part 2, or technically MPEG-4 Part 2, is alsocalled H.263. It included a new concept called ‘Profiles’ 。 HDCAM SRuses this specific Part.

Part 3 is still audio. AAC has beenincorporated into Part 3.

MPEG-4 has about 30 parts, one for eachtechnology. The two parts we are most concerned with are Parts 10 and 14.

H.264 or MPEG-4 Part 10/AVC

Part 10, or technically MPEG-4 Part 10,describes the AVC (Advanced Video Coding) format. This is H.264. Rather thanstart a new suite, they decided to add this as a Part.

H.264 isthe most widely used codec on earth, even surpassing broadcast MPEG-2, simplydue to the power of the internet. It is used by Youtube, and every other videoprovider of note.

It is also the codec that drives Blu-ray.Many broadcast pipelines and distribution channels have adopted H.264 (and MPEG-4 Part 2) as well.

Sony has added the XAVC codec (see the ‘AVC’in the name?) to its line-up because XDCAM couldn’t deal with 4K and above.

MP4

Part 14 describes a container format for MPEG-4 codecs, as in *.mp4. That’sall this is.

AVCHD

Sony and Panasonic took the parts they likedabout H.264 and called it AVCHD(MPEG-4 Part 10 AVC+HD). The key differences between AVCHDand H.264 is the former supports Dolby AC-3 encoding in addition to LPCM, and islimited by a total bit rate of 24 Mbps.

Whereas H.264 is designed as a distribution format(all MPEG suites are distribution formats, for the end user), AVCHD is designedfor cameras as an acquisition format as well.

Quality-wise, there’s no difference.

AVCHD 2.0

AVCHD 2.0 keeps up with a newer version of H.264 with better compression and sampling.But it’s still 4:2:0 and is limited to 28 Mbps. Make no mistake, AVCHD is aconsumer codec.

HEVC or H.265

As of 2013, the new kid on the block is MPEG-H.Part 2 (video, of course) is also called H.265 or HEVC (High Efficiency VideoCoding). There are many vendors already claiming compatibility with thisnew format. Here are its key benefits:

·        Up to 8K UHDTV (8192×4320 maximum)

·        12-bit color bit depth

·        4:4:4 and 4:2:2 chromasub-sampling

·        Supports up to 300 fps (earlierversions only supported up to 59.94 fps)

·        Data rates of several GB/s

·        File size ‘subjectively’ halfthe size of H.264 with better quality!

This is clearly the future. However, don’tbet on H.264 going away anytime soon. Who’s going totranscode the billions of H.264 videos already online?

Profiles

The new reality of the MPEG suite, beginningwith MPEG-4, is the concept of Profiles and Levels – in addition to Parts andLayers. Profiles are like presets, specific variants of the specification to beused for specific applications.从MPEG-4开始有了Profilesand Levels的概念，profile就像一个预设，当然用户可以根据需要改变其中的参数

After all, the MPEG-4 specification with itsthirty parts is quite large. Why would an encoder or software need to complywith all of it, if its chosen function is only for a specific need?

Profiles are unfortunately a necessary evil.Technology is moving too fast for committees to keep up, and many vendors don’thave to comply to the broadcast model anymore. profiles不是特别好，限制比较多，对于互联网上来说，需要更加自由的类型。

————————————————————————————————

There are two broad groups of compressiontechnologies currently used in video surveillance: frame-by-frame encoding （orintra-frame）andtemporal encoding （inter-frameor ‘time-based’ encoding）在视频监控领域有两种编码技术、

frame-by-frame encoding：

相比temporalencoding，可以减少延时；搜索比较快，逐帧。用的比较多是JPEG和JPEG2000.JPEG2000比JPEG节省30%的文件大小和带宽；有很好的动态范围(awide dynamic range)和更高的分辨率分级（scaleto higher resolutions）。关于动态范围：很多压缩技术只有8bit的深度，256级的动态，但是摄像头往往能捕捉更宽的动态范围。JPEG2000能保留这些动态范围。

temporal encoding：MPEG4 和H.264

解释IBPA GOP is typically composed of three different frame types: I, P, and B frames.Intra-Frames (I-Frames) are complete encoded images similar to the imagesgenerated using JPEG or JPEG2000 compression.Predicted-Frames(P-Frames) are coded with reference to the previous image, which can beeither another P-Frame or the previous I-Frame. Bidirectional-Frames(B-Frames)are sandwiched between I-Frames and P-Frames, andcontain information on the changes calculated between the previous andsubsequent frames.

MPEG4 part 2 受限于VGA的分辨率。H.264凭借运动矢量进行运动补偿，获得更高的压缩效率；还有他有de-blockfilter技术，能够在高压缩率的情况下保留更多的图像细节。H264可以使用更高分辨率的摄像头。

CBR和VBR ，CBR恒定码率，运动画面多的时候，就会提高压缩率，使得画面有些糊，VBR保证统一的画面质量。

逐帧编码对网络、存储等设计较为简单，保持恒定码率

Mathematically lossless compression refers toany compression applied to an image or video that can be completely reversed.Visually lossless compression refers to any compression that looks identical tothe original image but is digitally compressed in a way that can not be seen bythe human eye.无损压缩的意思

H.264和JPEG2000 的不同应用As resolution increases and framerate decreases, the benefits of temporal compression technologies like H.264diminishes. But, for smaller scenes that require a higher frame rate and use acamera with 2MP resolution or less, a temporal compression may be the mosteffective option.

 

 

H264 VS MPEG2

1、DCT

2、motion compensation

3、intra prediction 有8种预测模式

4、CABAC mpeg2 是VLC