日志

[o]深入 BT601 & BT656

已有 28897 次阅读| 2007-8-14 10:18 |个人分类:video tech

天气: 晴朗

心情: 高兴

Before the discuss:
.1 They BT601/656 are used for digitalize the common TV video system:
but you may know if only intent to digitalize ,the ADC can do the works. So , therefor, the following function should adding in .
.2 Beside digitalize, BT601/656 has strong compatible,it is common for 525-line, 625-line system ,and all the system common to them.

.1 what is the ITU-R BT601: BT601 = CCIR601
.11 the ITU-601 is for coding video signals:The YUV color space in,3 components - Luminance Y /Chrominance Cb, Cr
.12 The ITU-601 is for 525-Line /625-Line System

25 frames per second (30 frames per second), that is 50 fields/s (60 fields/s)
625 lines/frame. (525 lines/frame)

.13 The 13.5M sampling or 18M sampling mode limitation

    for luminance: 864 samples/line (858 samples/line), that means for both cases a sampling frequency of 13.5 MHz.
    for chrominance: 432 samples/line (429 samples/line), that means for both cases a sampling frequency of 6.75 MHz.
    so,it limite the input signal up to 576p(720*576) (864Sample/perTotalLine)(Luma BW 12M)input.

.14 Sampling Structure:

    orthogonal sampling coding: uniform. PCM either 8 bits or 10 bits.
    The 8 bit serial protocol (216 Mbit/s) was once used in D1 digital tape recording. Modern standards use an encoding table to expand the data to 9 or 10 bits for improved behavior. over long transmission lines. The 9 bit serial version has a data rate of 243 Mbit/s. By far, the most common version of the interface is the 10-bit Serial Digital Interface (which was later standardized as SMPTE 259M), which is now a ubiquitous interconnect standard for professional video equipment which operates on standard-definition digital video. This format, originally used in D5 digital tape recording, has a data rate of 270 Mbit/s. Additionally, a 360 Mbit/s version of the interface has been defined, which is sometimes used in widescreen applications.
    Color coding system is YUV 4:2:2 OR 4:4:4
    And then,
    The Sampling Structure is defined by Bitband and the Color system
    For instant:
    For a pair of pixels, the data are stored in the order Y1:Y2:Cb:Cr at 4:2:2 8bit mode, with the chrominance samples co-sited with the first luminance sample.

.15 quantization levels (for 8 bits samples) and analogical signal:

    In each 8 bit luminance sample, the value 0 and 255 only for sync.
    In each 8 bit luminance sample, the value from 1 to 254 for video.
    In each 8 bit luminance sample, the value luminance: 16 = black, 235 = white
    In each 8 bit luminance sample, the value chrominance: 128 = no chrominance

.16 The bit rate

    resulting is either 216 Mbps (for 8 bits samples as in D1-DTR), or 270 Mbps (for 10 bits samples as in D5-DTR).

    Active Video
    The active video is the part of the video signal that describes the image, in other words it's the video signal excluding the Vertical Blanking Interval and the Horizontal Blanking Interval.
    576 lines/frame. (480 lines/frame)
    720 samples/line for luminance and 360 samples/line for chrominance.
    The bit rate resulting is 165.9 Mbps for 8 bits samples.

.17 later versions introduced the bit-serial family of serial digital video interfaces that are now commonly used.

.18 601 Data order:

From the picure, we can see, that the total sampling for the Y = 863 and CR=431 CB=491

What is the ITU-R BT 656 = CCIR656
656 that the practical implementation of Recommendation 601 requires definition of details of interfaces and the data streams traversing them,most used for Interfaces for digital component video signals in 525-line and 625-line television systems operating at the 4:2:2 level of Recommendation ITU-R BT.601 (Part A).

The data signal are in the form. of binary information coded in 8-bit words. These signals are:
– video data;
– timing reference codes;
– ancillary data;
– identification codes.

.1 The 656 use the 4:2:2 video coding protocal in 601.
.2 Parallel Interface (Used in mostly TV VIDEO system)

The Y, Cb, Cr signal of ITU-R 601 is multiplexed in the following way: Cb, Y, Cr, Y, Cb, Y, Cr, ..., producing the Parallel Digital Signal, that has 27MHz frequency.The signal is driven to a bus of 11 pairs (10 for the data and 1 for the clock). Lines must be balanced ECL with characteristic impedance 100 ohms. Cables can be 50 m long without equalization (till 150 m otherwise). Connectors are type D 25 pins.

Notes: The D25pin is defined for Line driver and line receiver interconnection . basically 656 use 9 pin ,we called 8 - bit mode not D25 connector.

.3 Serial Interface (Used in most digital camera system)

The Parallel Digital Signal is converted to the Serial Digital Signal, that has 270 MHz frequency. The conversion includes a scrambling process, in order to reduce the DC components and make easier the clock reconstruction, and a Non Return to Zero Inverted coding, that make the signal insensible to polarity.The cable is coax. 75 ohms, the connectors are BNC.
Notes: This mode use only one line to transform. the 656 data, So you see BNC connector here.

.4 Data format:

A BT.656 data stream is a sequence of 8-bit or 10-bit bytes, transmitted at a rate of 27 Mbyte/s. Horizontal scan lines of video pixel data are delimited in the stream by 4-byte long SAV (Start of Active Video) and EAV (End of Active Video) code sequences. SAV codes also contain status bits indicating line position in a video field or frame. Line position in a full frame. can be determined by tracking SAV status bits, allowing receivers to 'synchronize' with an incoming stream.

Individual pixels in a line are coded in YCbCr format. After a SAV code (4 bytes) is sent, the first 8 bits of Y (luma) data are sent then 8 bits of Cb (chroma U), followed by 8 bits of Y for the next pixel and then 8 bits of Cr (chroma V). To reconstruct complete Y,Cb,Cr pixel values, chroma subsampling must be used

The compare from BT601 & BT656 :
.1 The BT601 and BT656 are all digital video signal data format

The video signal means: the input is CVBS,S-VIDEO,and YPBPR for DTV,basic in YUV coding system.

.2 the BT601(ITU-601) and ITU-656 are all 4:2:2 data format support. But 601 is also support 4:4:4.The 656 only support the 4:2:2.

difference:

.1 The 601 sampling rate is 13.5M,The 656 sampling rate is 27M
.2 The data format for 601 include seprated H,Y sync
The 656 has embedded (the digital data include the coding information for Hs Vs and Field)Sync(Called Video Timing Reference Codes SAV,EAV),656 has field informations.

Different from BT.601 and BT.709

BT.601 was established in 1987 for 525-line (NTSC) and 625-line PAL/SECAM. In 1990, BT.709 was introduced for high definition (HDTV) with specifications for 1125 and 1250 lines. In 2000, BT.709-4 added 1080 lines to conform. to the DTV standard. Following are the sampling frequencies used for both standards.

                  Luma        Chroma
   Standard       Sampling    Sampling

   BT.601 SDTV    13.5MHz      6.75MHz  (4:2:2)
   BT.709 HDTV    74.25MHz    37.125MHz (4:2:2)

问题小结：（中国人用中文：）
1 有关于8位，10位，16位，32位的问题：
位数的概念其实是两个方面的：
.1 是标准里面定义的"a uniformly-quantized 8-bit binary encoding,",也就是编码的格式，601里面支持两种格式：8-bit,10-bit.
   如果是8-bit，那么信号的表示范围是0-255，如果同时是4：2：2，记住这个先决条件，那么，就有下面的结论：
   "In the case of the 4:2:2 systems described in this Recommendation, levels 0 and 255 are reserved for synchronization
data, while levels 1 to 254 are available for video."

.2 硬件连接的表示方法：
    因为656一般是用8位数据线的+1根时钟线，那么也叫做8-bit mode,很多人认为这个是串行的，因为相对于16bit的601节省了8根数据线。然而，在656标准里面是却定义为位并行的（BIT-PARALLEL INTERFACE），因为656还有更决的，那就是一根线传送的标准：（BIT-SERIAL INTERFACE）
而601一般是16位数据线的，那么也叫16-bit mode,这个和编码的位的概念是不同的。但是，601似乎也有8-bit的所谓的接口串行的接法。

.3 有关于16-BIT 601，这个是指数据线的位数。
    我们知道，601一般是16位的数据线，因为y/cb/cr的编码定义是同时的，也就是串行的，那么必须是27M的带宽时钟。
    所谓串行不是指数据线的位数，传统认为，数据线位数多的就是并行。
    这里的串行是指在一个基本的编码字（Code word）里面Y/CB/CR是否是交织的，这个在后面我会写详细解释。
    但有时候，也有用16位的数据线，把y 和cb/cr分别同时传，这样只需要13.5M的带宽时钟。

2 有关于支持串行和并行两种模式：
    所谓串行和并行，有时候，定义为如果y/cb/cr同时传送那就是串行的，否则就是并行的。
    所谓串行不是指数据线的位数，传统认为，数据线位数多的就是并行。
    这里的串行可以是指在一个基本的编码字（Code word）里面Y/CB/CR是否是交织的。有是并行的，没有串行的。
    但这些并行，串行的方式总的来说都可称为数字串行的接口（serial digital video interfaces），因为每一个端口的数据都不是固定的，
    根据编码的格式不同，比如：601-8bit , 他的端口（D7-D0）可以是...0(8-bit)，CB(8-bit),Y(8-bit),CR(8-bit),Y,0,...

    601 8-bit模式是Y/CB/CR的串行传送的数字串行接口，所以带宽要求高。
    601 16-bit模式是Y ，CB/CR的并行传送的数字串行接口,但是因为是16bit 所以带宽减半

    656作为"practical implementation of Recommendation 601",因为固定8位的（The data signal are in the form. of binary information coded in 8-bit words.），所以应该归为串行的传送。但是，因为可以656还可以变为一根线传送的，所以，又支持位串行和位并行两种方式，这个和传统的串行和并行不同。

3 有关于601是8-bit 模式的硬件连接，为什么传送带宽反而要求更高。
    因为601是"co-sited"的"sampling structure", 对于，4：2：2结构来说，那么,"co-sited"是指同时的，那么它的带宽就是： Y+CR+CB = 13.5M+6.75*2M = 27M
    也就是在一帧的视阈内，同时在8位的数据线上传送这些数据，所以带宽要求高。当然也有16-bit模式的601.
    656作为601的扩展，支持串行和并行两种模式：并行模式下，16根数据线，可以用一般的带宽分别传送： Y + CRCB =13.5M.

4.16-bit mode 601和656的并行模式区别和选择理由：
    。1 同步传送和非同步
    656除了基本的601信息外，还含有一些同步数字编码信息，比如SAV,EAV等等。所以他总共9位，8-bit data + clock
    601 -16bit 除了16位的数据，还有HS,VS,CLK信号，总共是19Bit.
    CCIR656 是同步传输的，decoder 端不用自己产生 clock。在 CCIR601 中，decoder 要自己去锁定这个同步讯号，再自己产生 clock 来解。所以以 CCIR656 来传递的 video 讯号质量应该更    好控制。

    。2 速度因素：601有更多的Pin脚，而且省去了再解码 656 -〉601 的步骤，所以应该更快

    总结，现在IC pin脚的节省大于速度因素，所以，一般考虑656。

5.抽样频率的选择和两大系统的统一
    对于数字分量编码,CCIR601建议亮度抽样频率为525/60和625/50三大制式行频公倍数2.25MHz的6倍，即13.5MHz。对现行NTSC电视制式而言，亮度信号的带宽是6MHz，13.5MHz>2×6MHz=12MHz，所以它符合奈奎斯特定理。而色差信号的带宽比亮度信号窄得多，所以在分量编码时两个色差信号的抽样频率可以低一些。因同时考虑到抽样的样点结构应满足正交结构要求，两个色差信号的抽样频率均选为亮度信号抽样频率的一半，即6.75MHz，这样亮度信号与两个色差信号的抽样频率之比为 4∶2∶2。

6.数字分量视频信号有效行取样点数的确定和两大系统的统一
每行数字分量信号的取样点数为：
　　对于625行/50场制式：
　　　每行亮度取样点=13.5Mhz/15625Hz=864点/行；每行每个色度取样点=6.75Mhz/15625Hz=432点/行。
　　对于525行/60场制式：
　　　每行亮度取样点=13.5Mhz/15734.266Hz=858点/行；每行每色度取样点=6.75Mhz/15734Hz=429点/行。
　可见，这两种制式选用了相同的抽样频率，但每行取样点数却不相同。所以把两者取样点数之差别放在数字有效行以外的部分，而使每个数字有效行内的取样点数相同。
　　CCIR 601建议两种制式有效行内的取样点数亮度信号取720个，两个色差信号各取360个，即每个数字有效行包括720个亮度数据和720个色度数据(两个色度各360个),这样就统一了数字分量编码标准，使三种不同制式便于转换和统一。所以有效行亮度信号与两个色差信号的取样点数之比也为4:2:2 (720:360:360)。
　　上述两点即为获取高质量的后期制作由CCIR 601建议所确定的数字分量编码标准：

7.同步的方式：
    ITU656接口为8bit数据位宽接口，为内同步方式，即同步字嵌入于ITU656码流当中；
    ITU601接口可以为16bit数据位宽接口，为外同步方式，即同步信号用单独的信号线传输。

    从ITU656接口转换为ITU601接口的方式是通过将嵌入在ITU656码流内部的同步信号提取出来，
    同时将表示亮度和色度的数据流分离开，排成16bit数据位宽的ITU601格式同时输出，从而完成格式的转换。
　
8.656内部使用的校验码基本
    P3，P2，P1，P0是1位误码纠错和2位误码检错的功能
    海明校验码
    这是由Richard Hamming于1950年提出、目前还被广泛采用的一种很有效的校验方法，是只要增加少数几个校验位，就能检测出二位同时出错、亦能检测出一位出错并能自动恢复该出错位的正确值的有效手段，后者被称为自动纠错。它的实现原理，是在k个数据位之外加上r个校验位，从而形成一个k+r位的新的码字，使新的码字的码距比较均匀地拉大。把数据的每一个二进制位分配在几个不同的偶校验位的组合中，当某一位出错后，就会引起相关的几个校验位的值发生变化，这不但可以发现出错，还能指出是哪一位出错,为进一步自动纠错提供了依据。假设为k个数据位设置r个校验位，则校验位能表示2r个状态，可用其中的一个状态指出 "没有发生错误"，用其余的2 r -1个状态指出有错误发生在某一位，包括k个数据位和r个校验位，因此校验位的位数应满足如下关系:
2r ≥ k + r + 1 (2.7)
如要能检出与自动校正一位错，并能同时发现两位错，此时校验位的位数r和数据位的位数k应满足下述关系:
2r-1 ≥ k + r (2.8)

Reference:
.1 http://www.cnblogs.com/wantfei/archive/2006/12/07/62030.html
.2 http://www.dtvbbs.cn/forum/
.3 http://emic.blog.sohu.com/70924040.html
.4 CCIR 601/656
.5 SAA7118

By Makin - 2007-11
所有的总结都建立在标准和实际的例子基础上，如有不对的地方还请和我联系：huang.makin@gmail.com or hsy75@msn.com