The video gathers the whole of the Technique S, Technologie, allowing the Enregistrement as well as the restitution of Image S animated, accompanied or not by its, on a support adapted to the electronic and not by photochemical type. The video word comes from the Latin video which means “I see”. It is the Apocope of vidéophonie or videogram. The Substantif video agrees of number, however, the Adjectif remains always invariable.

Theory

A video stream is composed of a succession of images, 25 a second in Europe (30 a second with the USA), composing the illusion of the movement. Each image is broken up into horizontal lines, each line being able to be regarded as a succession of points. The reading and the restitution of an image are thus carried out sequentially line by line like a written text: from left to right then from top to bottom.

Interlacing

The image of a Téléviseur is a succession of line sweeps, from left to right, on the basis of the top, and finishing in bottom of the screen. At the beginning of television, the quality of the phosphorescent elements of the tube is extremely poor. So when the beam sweeps the bottom of the screen, the high one disappeared already, from where a phenomenon of flutter, strongly felt by the human eye for 25  Hz or 30  Hz. The simplest solution had been to accelerate the rate of sweeping, but this also forced to increase the rate of the images, which was useless from a cinematographic point of view (the movement is perceived in the same way), and extremely expensive in material and Band-width. A more astute solution was to double the rate of sweeping, by omitting a line on two, in order to keep a quantity of constant information. Thus, a first master key posts all the odd lines twice less time than for a whole image and a second master key posts the even missing lines: it is what one calls the Entrelacement. One obtains well the same number of lines of sweepings for an image, and one twice sweeps the screen to post only one image. One indicates by the term “screen” (" field" in English) a master key of sweeping. An image thus consists of two screens, since one needs two sweepings to define the image (" frame" in English).

The Caméra S, which function like a “reversed television set”, adopted they also this interlacing of sweeping. In first half of the time of an image, one 1st catch of sight defines all the odd lines, and a half of image later, a second catch of sight defines the even lines. What it is necessary well to include/understand here, it is that the two catches of sights are distant in time (of a half of image). And even if these two catches of sight are complementary to a space point of view (both sweepings are complementary within the framework), these two catches of sight do not post the same contents! If a subject moves in the field, it will have a different position on each of the two screens: there is then an effect of zigzag on each frame .

This problem is partly solved by a device of birefringent crystalline blades which " étalent" details by duplicating the luminous rays. It results from this a loss from Définition which confers on the system STAKE and SECAM a vertical resolution multiplied by 0.7 (Facteur of Kell) and which is more really only of 400 lines approximately.

Because of capture in two screens of 1/50 S each one, it is the exposure time in video (25i).

There exists henceforth more and more of video apparatuses able to post 25,50 or 60 complete images a second, posting is not interlaced any more, one then speaks about progressive Balayage. Among the apparatuses capable of such a posting one finds: the Computer S (their video Chart and them screen), some Vidéoprojecteur S, the Television set S top-of-the-range, some rare platinums DVD and some Caméscope S. It is the mode of capture chosen for films made in HDTV or D-cinema intended to be transferred and projected in 35 mm.

The 25 progressive images (25p) confer then on the cameras an exposure time of 1/25 S what is too long in term of temporal resolution. One then prefers to limit the time of integration of the screens to 1/50 S (electronic Obturateur).

Capture image

The first cameras video, functioning on the same principle that the television sets, analyzed the image formed by the objective using a Cathode tube. Since the end of the year 1980, they are equipped with sensors Load-Coupled Device: CCC or Device with Transfer of charge (DTC) in French.

The transfer of these loads can be done in 3 different ways: transfer interlines (sensor IT: Inteline Transfer), transfer weaves (sensor FT: Frame Transfer) which requires a mechanical obturator and is seldom used or transfer the FIT (Frame Interline Transfer).

At the beginning of XXIe century, the manufacturers of sensors decided to give up this technology and build from now on sensors CMOS (Complementary Metal Oxide Semi-conductor). One finds however still on the market of the video cameras semi-professionals using technology known as " tri-CCD" who allows to notably improve the treatment of the colors.

Resolution of the image and frequency of sweeping

There exist various formats of video image, which depend primarily on the field frequency of the image.

  • 405 lines 50 Hz (standard abandoned English) black and white

  • 525 lines 60 Hz: useful resolution 4/3 = 720 X 480 (American standard) color NTSC and PAL-N
  • 625 lines 50 Hz: useful resolution 4/3 = 720 X 576 (European standard) color STAKE, SECAM and NTSC-4.43
  • 819 lines 50 Hz: useful resolution 4/3 = 1024 X 768 (abandoned French standard) black and white

One can note at this point which there exists a difference between the number of lines composing the image and the number of posted lines. This represents a difference of 49 lines in 50 Hz and 45 lines in 60 Hz. These lost lines are necessary, they represent the time necessary so that the Electron beam sweeping the Cathode tube can go up bottom of the image upwards. These engineering problems do not exist with panels LCD and the flagstones plasma, but it is preserved to ensure compatibility. The free lines are made profitable partially: one places there the signals of the Télétexte, the Sous-titrage and also the Time-code of the professional video equipment.

It is necessary to distinguish two frequencies from sweeping of the image:

  • the field sweep, which is carried out from top to bottom and is used to compose the image. It is carried out 50 or 60 times a second.
  • the line sweep, which is carried out from right to left for each line of the image. The line frequency is thus equal to the vertical frequency multiplied by the number of lines and divided by two because of interlacing.
F_h = \ frac {F_v NR} {2}

What gives the following values:

  • Fh (50Hz) = 50 X 625/2 = 15625 Hz
  • Fh (60Hz) = 60 X 525/2 = 15750 Hz

This result is not randomly. If the horizontal frequencies are almost the same ones in 50Hz and 60 Hz, it is that makes it possible to use the same circuitery of line sweep, therefore to realize savings.

Color

For a few decades one had known the spectral characteristics of the human eye, which posted a very clear preference for certain colors. Moreover one knew that the chromatic spectrum of the eye can break up into three primary colors, which make it possible by mixture to recreate about all the other colors of the spectrum. The cinema color exploited this by using emulsions with several layers, of which each one was sensitive to a given color.

The video engineers chose 3 quite particular colors: Red blue green. These colors are known as primary elections because it is they which, by mixture, will make it possible to recompose a whole spectrum of colors.

The catch of sight color is carried out according to a optical prism which distributes the light on three sensors, in front of which one has a red filter respectively, green and blue. Thus, each sensor records only information of light concerning its color. It is then enough to record then to restore 3 components RVB (English RGB) on a monitor color accepting 3 entries RVB. It should well be understood that one obtains 3 signals in the place of only one. It is not only necessary to triple all the connections cabled between the various equipment, but also to triple the tracks of recording on a video tape recorder, to triple all the equipment of production, to the equipment of hertzian diffusion… the challenge was thus to create a single signal including 3 different information, and which was not to mix before the treatment by the station of reception.

The challenge was also still to preserve total compatibility with the stations Noir and white very present in the hearths. One thus worked with an aim of creating a video signal including: red, green, blue, and N&B in the same pipe, without those mixing.

To start, it was unthinkable to have a camera N&B AND a camera color. It was thus necessary to manufacture N&B starting from 3 components RVB. Basing itself on the sensitivities of the eye to the various colors, one took 59% of green, 30% of red, and 11% of blue which one mixed copiously. One had just invented a new term: the Brightness (Y). TVS N&B could thus see in N&B of the images resulting from camera color. How now to add with this Y information of colors allowing us to find our original RVB? Since one had already the light of our image (Y), it was necessary “to color” this N&B with information of colors which did not contain they, no value of light, but only of the indications of Teinte and saturation.

Once of agreement for this colorized N&B, it was necessary to find the easy way which would make it possible to transmit the light (Y) and chroma (that we will call C to make simple). Electronic processes with the names as alarming as “amplitude modulation in squaring of phase, with removed subcarrier” transfer the day. These solutions were at the same time to mix 2 signals so as to be able to discriminate them with the reception, but also not to have any visible interference in the spectrum of signal N&B.

These solutions were found and applied. Thus were born NTSC (National Television System Committee) in the United States, the SECAM (Sequential Color With Memory) in France, and the STAKE (Phase Alternate Line) in Germany. The technique employed to transform RVB into compatible signal color N&B is called coding. The NTSC, the SECAM and the STAKE are 3 types of different, and of course, incompatible codings between them. To pass from a type of coding to another is called Transcodage.

None of the three solutions is nevertheless transparent, far is necessary some. A coded signal suffers from Artifact S more or less visible according to coding.

A coded video signal of the kind is known as composite Signal, because it contains several sources of different nature. The video standards using the composite go from the U-MATIC/U-MATIC SP to VHS while passing by the 8mm or Video 8, the Betamax, the VCR or the V2000. Within sight of the degradations caused by coding, it became urgent to be exonerated some in production.

At beginning of the year 80, SONY developed a format Vidéo with separate components, made up of several distinct signals, conveyed by distinct cables: the Betacam/Betacam SP. To remain compatible N&B, the RVB carefully was avoided, and one naturally chooses a format comprising famous Y (signal N&B), more of the information of chrominance conveyed by 2 signals: U & V (also called Cr and Cb). For those which would not have taken down yet, U = R - Y, V = B - Y, where Y = 0,30R+0,59V+0,11B (coefficients being different according to coding used). This transformation of RVB into YUV is called Matriçage. Contrary to coding, dieing is a very simple operation, which does not generate degradation, while offering the advantage of compatibility Y.

A few years later, one saw appearing a format general public known as S-Video or Y/C, where brightness Y and the chrominance C (coded in NTSC, STAKE or SECAM) were separate (S-VHS, Hi-8, Super-Betamax). This format is of quality better than a composite format, since the chrominance does not encroach any more on the waveband of brightness, which could bring to artefacts coloured on fine details. The horizontal resolution of these formats could thus be almost doubled (400 points/line instead of 240-250).

Digital video - the 4:2: 2

Introduction

|----- |Many useful samples per line |
720 | |----- | Structure of sampling | colspan=" 3" align=" center" |
Deux interlaced screens |---- |Quantification 8 bits |220 useful levels |225 useful levels |--- |Quantification 10 bits |880 useful levels |900 useful levels |--- |rowspan=" 2" |Signal report/ratio on noise | colspan=" 3" align=" center" |
quality 8 bits: 56 dB
quality 10 bits: 68 dB |----- | |----- |Coding |Binary |Binary shifted |----- |rowspan=" 2" |Rough flow | colspan=" 3" align=" center" |
8 bits: 216 Mb/s
10 bits: 270 Mb/s |----- | |----- |rowspan=" 2" |Flow Net | colspan=" 3" align=" center" |
8 bits: 166 Mb/s
10 bits: 207 Mb/s |} The history of numerical in the video begins truly 1972 with 1982. In the beginning equipment of synchronization, the apparatuses became more sophisticated before entering professional environment. Consequently, the industrialists became aware of the advent of this new phenomenon and presented standards as regards digitalization. A certain numerical anarchy reigned then on the market what forced the hand with CCIR (international Advisory committee of broadcasting) to standardize a video format in numerical components compatible in the whole world: this standard it is the 4:2: 2, or CCIR 601. It specifies the parameters of coding of signals to be digitized (sampling, quantification…) Consequently the innovations did not cease being connected to allow today, with the digital video, to spread in the production centres, chains TV and control of post-production to assist the video Montage.

Video acquisition: analogical/numerical conversion

The process of video acquisition Analogical and its conversion into Numérique can be assimilated to the passage of the oral language to the written language. To take in note the oral speech of a person, the latter should not speak too quickly, in such case it becomes difficult to listen and transcribe simultaneously. Admittedly the person could slow down her flow of word but if one assimilates these words with the analogical video signal, one understands easily that the flow cannot be slowed down. One thus proceeds to the sampling of the speech, i.e. one seizes only of the “pieces” of message to retranscribe them thereafter. The precision of the retranscription thus depends directly on the number of samples of taken speeches. For the video, the phenomenon is identical: it is necessary first of all to know the signal and of knowing which are the signals to be digitized.

Why a coding of the components?

The signal digital video owed, without any doubt, identical being for all the countries: the idea was to digitize of the data common to the systems 625 lines (STAKE, SECAM) and 525 lines (NTSC). The CCIR thus unanimously decided to digitize in a separate way the signals of brightness (Y) and chrominance (Cr; Cb). A system based on the numerical coding of the video components excludes all the problems which could have generated a coding of video signal composite and allows a compatibility global scales. This system should thus seem being the principal accessory of a development of new equipment, but more especially of international exchange of data, constituting the base of the audio-visual one: the communication.

Sampling

The signal sampling, it is cutting in temporal sections of the latter. It is directly followed quantification which consists in punctually taking the value of the signal at regular moments, corresponding to the period of sampling. It is thus necessary that the rate/rhythm of cutting (sampling rate) is high to be able to retranscribe the variation of the signal of the shortest origin. Because if the time interval between two consecutive samples is higher than the time of the fastest variation of the signal of origin, the latter will be lost and will not be taken into account in the numeric signal. Consequently, to sample a signal by preserving its information, it is necessary to know the highest frequency to which it is likely to vary. The mathematical law of Shannon and Nyquist establishes that “a signal whose spectrum is limited to the Fmax frequency is entirely defined thereafter of its samples taken with regular time intervals of value T < 1 (2 Fmax)”.

Consequently, the sampling rate must be ƒ E > 2 Fmax to be the representation of origin. If this condition is not observed, the repetitive spectral components of the sampled signal are not enough spaced and overlap. The zone of folding up, called zone of aliasing, gives rise to a spurious frequency resulting in a moire effect of on the image. To mitigate this problem, a low-pass filter (anti-aliasing filter) is laid out upstream of conversion. This steep slope filter rejects the frequencies of the analogical signal of entry which are higher than 1/2 ƒ E .

The video signal of brightness has a band-width from approximately 6 MHz. To be precisely digitized, the sampling rate of this signal must respect the criteria of Shanon and Nyquist is: ƒ E (Y) > 6 X 2 = 12 MHz

However, to be used on a world level, ƒ E (Y) must be multiple commun run of the frequencies lines of the systems with 525 and 625 lines with knowing 15.625 and 15 734,2 Hz. The CCIR thus retained the following sampling rate: ƒ E (Y) = 13,5 MHz. This frequency is equal to 864 times the frequency line of the systems with 625 lines and 858 times that of the systems with 525 lines (a line activates containing 720 samples).

For the signals of chrominance, the band-width is from approximately 3 MHz. The CCIR decided to sample them at a frequency twice less of that for brightness is ƒ E (Cr) = ƒ E (Cb) = 6.75 MHz. For these signals, there will be thus only 360 samples per active line. This is not really awkward for the human being which is less sensitive to the color than with illumination.

These sampling rates determined by the CCIR are connected with figures 4,2 and 2. For a group of 8 pixels (4 pixels per line and on 2 lines), figure 4 accounts for the number of values indicated by line for brightness (13,5 MHz), 2 the number of pixels having an eigenvalue chrominance (6,75 MHz = 13,5/2) on the even lines, and the last 2 idem for the odd lines. Thus standard CCIR 601, born from these studies, taken the name running of standard 4:2: 2.

The periodicity 2 screens allows three types of structures of sampling: orthogonal, quincunx line and quincunx weave. It is the orthogonal structure which held the attention in the standard 4:2: 2. In this structure, the phase of the clock of sampling is identical for each line and each screen. The samples are thus located at the same sites from one line to another and screen to another.

The quantification

Each sample “is weighed”, just like a food, in order to determine its weight of them. Numerically, this weighing is called quantification. It is carried out, to take again our analogy, using a balance with two plates: in one of the plates is the sample to weigh, in the other the weights necessary to find balance. The precision of weighing thus depends on the value of the smallest weight available. In video, the weight of the sample is the tension of the signal to be digitized and a quantifier balances it. This apparatus converts the tensions into numerical, exploitable values by a virtual station of assembly, for example.

However, the quantification cannot represent perfectly the tension of the sample of the analogical signal of origin. Indeed, an analogical signal can take an infinity of values but it will be converted into a formed signal of a number finished of numerical values “NR” of which each one is coded on “N” bits. There will be thus necessarily, after quantification, a rounding error. The precision of the converted signal will be thus related to the number of liquid assets to translate each sample. The interval located between two values is noted “Q” and names “not quantification”. At every moment “T”, the amplitude of the signal being inside a level is replaced by the value of the level nearest. It is easily understood that the smaller the steps of quantification are, the more they are numerous on a given beach and thus that more the precision of the quantified signal is important (the error rate of quantification being determined by the Terr relation = 1/2n).

The quantification of the video signal is uniform, linear and is carried out in a separate way on Cr and Cb. Initially fixed on 8 bit S, the quantification of the video signal of the standard 4:2: 2 was passed to 10 bits. Indeed, a quantification on 8 bits makes it possible to have 28 = 256 numerical levels (including 220 useful to represent the levels of gray) what is sometimes not sufficient. For a range of gray of the white to the black, for example, a “effect of staircase” appears after digitalization. Moreover, report/ratio S/B of a quantification on 8 bits is of 56 dB whereas the cameras of today reach the 60 dB. The C.C.I.R. thus chose to quantify the video signal on 10 bits, which authorizes a scale of 210 values i.e. 1024 levels (including 880 useful) is 4 times more than one quantification out of 8 bits with for report/ratio S/B = 68 dB.

The signal of brightness is always positive and does not pose problems to be digitized, on the other hand the signals of chrominance are bipolar. One thus had to fix a value for the null signal: values with the top corresponding to a positive signal and those to the lower part with a negative signal. This “zero value” was fixed by the C.C.I.R. at 512.

The coding of channel

Once sampled and quantified, the video signal must be coded in order to optimize its storage or its transmission. Various forms of coding exist and present each one their advantages and disadvantages. The goal of the operation is thus to choose the code more adapted to the use. For that, several codes are at disposal:

  • code NRZ (Not Return to zero): a binary data “1” generates a high level of signal and a data a “0” bottom grade

  • code NRZI (Not Return to zero Reversed): a binary data “1” generates a transition in the middle of the half-period from clock, a data “0” does not have any effect. This type of coding is used in video in the connections series 4:2: 2 because it makes it possible to transmit with the video signal its clock signal.

  • the Two-phase code Mark: used for the signal of LTC of the video tape recorders. One “0” causes a transition and a maintenance of the level for all the period of clock, whereas one “1” involves a transition and a change of level with half of the half-period of clock.

There still exists of other codes (like the code Miller or Miller square codes it) which are used only in certain digital video recorders.

The structure of the numerical line

The analogical lines of the systems with 625 and 525 lines are slightly different durations. Thus, the capacity of an active line must be sufficient to contain a sufficient number of samples in order to cover the lines of the two systems. The CCIR chose 720 samples for the signal of brightness and 360 for the signals of chrominance. This is sufficient because the analogical active lines longest are those of the systems with 525 lines which require more than 710 samples to be completely analyzed. The line activates 4:2: 2 is thus coded on 1440 words (720 X 2). The signals making it possible to position the numerical active line are coded respectively on 264 and 24 words for the systems with 625 lines and out of 244 and 32 for the systems with 525 lines. The face before timing pulses line (SAV) determines the arrival of the first sample and the reference of time for analog-to-digital conversion. The back face (EAV) determines the end of it.

Remarks on the detection and the correction of errors

The support of recording (or the transmission channel) can generate errors in the flow of numerical data. I.e. a binary value can take another value (one “0” becomes “1” and vice versa) or although information can miss at a given time. This error can either affect the visible video image or the other video signals according to the bits which it affects. It can thus have more or less important consequences from where utility to detect them and to correct them.

The difficulty of the systems of correction of error lies in the fact that it is necessary for all to detect the error before being able to correct it. For that, redundant data are added during coding to the useful data, according to a definite and known law of the coder and decoder. With each time this law is not checked with decoding, a process of correction is started. If the lack of information is such as same the redundant data are not enough to find the signal of origin, of the processes of compensation, which consist in calculating the median value between close samples, are carried out. The signal thus corrected can finally be used by the various numerical equipment.

Report/ratio of image: 4/3 and 16/9

Historically, television was developed on screens with format 4/3 (either a ratio of 1,33/1). This format was selected because it was that used by the cinema at the time of the development of television, in the Années 1940. Since, the cinema evolved/moved, with processes such as the Cinémascope and others Panavision based on the use of an objective Anamorphose ur, the standard sizes with the cinema are the 1,85/1 and the 2,35/1. When it was decided to pass television towards a panoramic format, it is the format 16/9 which was selected. It corresponds to a report/ratio of image of 1,77/1, it is rather close to 1,85 and remains a good compromise between the 1,33 (black bars on the left and on the right) and the 2,35 (black bars in top and bottom). The purists preserve the black bars to see the entirety of the image, while those which prefer to benefit from the full screen use the zoom of the television set but lose consequently part of the edges of the image.

Video formats and standards

Analogical

Numerical

Standards of video recording

Video and Data processing

Data-processing postings have of specific resolutions and modes of sweeping quite as specific. The microcomputers 8 bits and first 16 and 32 bits were intended for a connection on a television apparatus, their video out was thus into 625/50 or 525/60. The standards used on PC are different:

  • CGA 320x200x4c or 640x200x 2c to 60 Hz
  • Hercules 640x400 (N/B) to 72 Hz (?)
  • EGA 640x350x16c to 60 Hz
  • VGA 640x480x16c to 60 Hz

The other views are not really standardized. It will be noted that the standard formats of image are declined in a variable number of colors (16, 256,65 ' 536, 16 ' 777 ' 216, 4 ' 294 ' 967 ' 296 and more).

  • 640×480
  • 800×600
  • 1024×768
  • 1152×864
  • 1280×960
  • 1280×1024
  • 1600×1200
  • 2048×1536
  • 2560×2048
The frequency of sweeping is included/understood between 50 Hz and more than 120 Hz. All these postings are with progressive sweeping although in more the high-resolutions, it is possible to find modes interlaced.

It is because of the different frequencies of sweeping that it is not possible to connect a computer directly on a television set, that can even involve the destruction of the television set. In addition, a coder color (STAKE, SECAM or NTSC) is necessary to carry out a video recording of a data-processing image. Therefore certain computers are equipped with a video out independent of the exit intended for the monitor.

Glossary

Technical terms

Apparatuses

The video, through the Video art, knows an artistic practice since the Sixties approximately.

See too

Related articles

Simple: Video

Random links:Robocop (televised series) | Bazooka (group) | Institute of nuclear physics of Lyon | Redshirt | Superliga Española de Hockey Hielo 1998-1999

© 2007-2008 speedlook.com; article text available under the terms of GFDL, from fr.wikipedia.org