EP1131810B1 - Addressing method for plasma display panel based on separate even-numbered and odd-numbered line addressing - Google Patents

Description

The invention relates to an addressing method and device for plasma panel based on separate addressing of even lines and odd lines.

On plasma screens, the gray level is not achieved in a classic way from an amplitude modulation of the signal but from a temporal modulation of this signal, by exciting the corresponding pixel, more or less long depending on the desired level. This is the phenomenon of integration of the eye which makes it possible to render this gray level. This integration takes place during the frame scanning time.

The eye actually integrates much faster than the frame time and thus risks detecting, in cases of particular bit transition addressing, level variations that do not reflect reality. Defects of contour or "contouring" according to the English name can thus appear on moving images. These faults can be compared to bad temporal restitution of the gray level. More generally, false colors appear on the edges of objects, each of the cells of a color component which may be subject to this phenomenon. This phenomenon is even more annoying when it appears on areas relatively homogeneous.

A simple theoretical solution to limit the appearance of false contours is known from the prior art to be described for example in the application for French patent filed on April 25, 1997 and published under the reference FR 2762704 and which consists in multiplying the number of sub-scans to make minimal disturbance from changes in the video level of a frame on the other. The additional subscans needed are from sweeps saved by simultaneously addressing two adjacent lines. However, this simultaneous addressing results in loss of resolution, the information copied from one line to another being obtained by recoding the grayscale, using the possibilities of code redundancy. He is not however, it is not possible to control the amplitude of these losses in resolution.

Another problem of the prior art concerns the conditions boot.

One of the special features of the plasma cell is that it has a threshold of trigger which is not independent of the state of its immediate neighbors. A cell will be all the more easily excitable when its neighbors are excited, we are actually talking about a priming phenomenon. The barriers separating the different cells are not completely hermetic, a certain number of free electrons from excited neighboring cells come favor the excitation of the addressed cell.

This priming problem is in fact amplified by the non-uniformity of the sign. It is always possible, to promote the excitation of cells to do vary the control voltages, but this becomes impossible when the Glass tiles do not have the same spacing over the entire panel, for example. In this case, the compromise found at the level of the control voltages does not does not optimize the ignition of all cells.

Another problem of the prior art concerns the quantification of low levels.

The plasma panel, unlike the cathode ray tube has a linear response, i.e. the level of luminance emitted is strictly proportional to the video level. Current visualization systems are largely based on the use of cathode ray tubes. So he is performed at the shooting level, an a priori compensation operation of the cathode ray tube response. To be able to correctly view such a signal on a plasma panel, so correction is necessary inverse (gamma correction) to obtain real information in the end.

Figure 1 shows the shape of the compensation curve 1 of the response of a tube to the broadcast, the abscissa axis representing the video level input and the ordinate axis representing the output video level after correction. Curve 2 corresponds to a linear response obtained after application of the correction as shown in 3.

The effect of this correction is to greatly limit the quantification of low levels insofar as at a signal level of output can correspond to several levels of the input signal. this is especially true for low levels, for example in the area materialized in 4 where the entry levels between 0 and 15 correspond to a single level output equal to zero.

To make these low levels perfectly, it would be necessary to have more than 8 quantization bits (10 or 12 for example).

The object of the invention is to solve the cited drawbacks. To this end, the subject of the invention is a method of addressing cells arranged according to a matrix table, each cell being located at the intersection of a line and a column, the table having row entries and column entries for the display of grayscale defined by video words composing a signal digital video and defining an image, the column inputs receiving each a control word in this column corresponding to the video word relative, for this column, to an addressed line, this word being composed of n bits transmitted sequentially, each sequence corresponding to a sub-scan, each bit triggering or not, depending on its state, the ignition of the cell of the row addressed and of the column receiving the command word, for a time proportional to the weight of this bit in the word, characterized by what we do a different coding of the column control words according to that the word is relative to an even or odd line, this difference consisting of that at least m successive bits of determined ranks have different weights from one control word to another, the sum of the weights of these bits remaining identical from one command word to another, to obtain writing moments significantly different from one line to the next.

According to a variant of the method, the writing is simultaneous on two lines successive for at least the first bit of the m successive bits of a word of command relating to one of the two lines.

According to another variant, at least two are simultaneously selected successive lines for at least one of the bits of the column control words whose weight is common from one command word to another.

According to another variant, at least one of the bits of identical weight of a command word to another is used to encode a partial value of luminance common to two successive lines and the writing is simultaneous on these lines for this bit of the control word relating to one of the two lines,

According to another variant, the method is implemented for a number limited lines of the matrix table, these lines corresponding to the areas of the image defined by the video signal having strong vertical transitions, the other zones exploiting sub-scans corresponding to a process address for which the column control words have all the weights identical from one line to another.

According to another variant, the method is implemented for images having strong vertical transitions, the other images exploiting a addressing method for which the column control words have all the identical weights from one line to another.

According to another variant, the switching of the first method with n sub-scans to a second addressing method with a higher number of sub-scans and for which the words of column command have a higher number of bits having identical weights from one line to another is performed by replacing the selection of a line I when writing a bit of different weight on line l, in the first method, by the selection of line I and the immediately preceding line or immediately following for a simultaneous writing on these two lines, in the second process.

The invention also relates to a device for implementing the previous method comprising a video processing circuit for processing received video data, a correspondence memory for the transcoding of this data, a video memory for storing transcoded data, the video memory being connected to circuits column supply to control the column addressing of the panel to plasma from column control words, a circuit for controlling line supply circuits connected to the video processing circuit to select the lines, characterized in that the video processing circuit and the transcoding do different coding of column command words depending on whether the word is relative to an even or odd line, this difference consisting in that at least m successive bits of ranks determined among the bits to be transmitted have different weights from one command word to another, the sum of the weights of these bits remaining identical from one control word to another, to get writing times that are significantly different from line to line next.

According to an alternative embodiment, the device is characterized in that the line supply circuit control circuit selects simultaneously two consecutive lines during transmission by circuits column feed of the first bit of the successive bits of a word command relating to one of the two lines.

According to another variant, the device is characterized in that it also includes a selection circuit receiving the video data for select a coding of the column control words corresponding to a addressing according to n sub-scans or to an addressing corresponding to a higher number of subscans based on variations in luminance from one line to another of an image.

The addressing method according to the invention consists in separating addressing even lines of odd lines using coding different from the column command words. The writing instants from line to line the other, for certain bits of the control words, are significantly different. The initiation of cell excitations is thus favored.

This process allows a partial and variable copying of the video information from one line to another. We can thus play on the compromise number of underscans / loss of vertical resolution. It is then possible, by depending on the content of the video, to modify, for each of the couples of lines, the number of underscans and therefore the difference maximum allowed between two luminance values allowing an error lower than LSB.

Thanks to the invention, the contouring effects are eliminated or less strongly reduced, the quantification of low levels is improved.

• la figure 1, une courbe de compensation de la courbe de réponse d'un tube cathodique,
• la figure 2, un chronogramme montrant des niveaux de codage en fonction du temps,
• la figure 3, un principe de balayage d'un panneau à plasma selon l'art antérieur,
• la figure 4, un principe de balayage d'un panneau à plasma selon l'invention,
• la figure 5, un chronogramme pour l'écriture de deux lignes consécutives selon l'invention pour des bits de mots de commande colonne ayant des poids différents,
• la figure 6, un chronogramme pour l'écriture de deux lignes consécutives selon l'invention pour des bits de mots de commande colonne ayant des poids identiques,
• la figure 7, un exemple d'écriture sur deux lignes consécutives pour des bits de mots de commande colonne ayant des poids identiques,
• la figure 8, un exemple d'écriture sur deux lignes consécutives pour des bits de mots de commande colonne ayant des poids différents,
• la figure 9, un dispositif selon l'invention.

Other features and advantages of the invention will appear clearly in the following description given by way of nonlimiting example and made with reference to the appended figures which represent:

• FIG. 1, a compensation curve for the response curve of a cathode ray tube,
• FIG. 2, a timing diagram showing coding levels as a function of time,
• FIG. 3, a principle for scanning a plasma panel according to the prior art,
• FIG. 4, a scanning principle of a plasma panel according to the invention,
• FIG. 5, a timing diagram for the writing of two consecutive lines according to the invention for bits of column control words having different weights,
• FIG. 6, a timing diagram for writing two consecutive lines according to the invention for bits of column control words having identical weights,
• FIG. 7, an example of writing on two consecutive lines for bits of column control words having identical weights,
• FIG. 8, an example of writing on two consecutive lines for bits of column control words having different weights,
• Figure 9, a device according to the invention.

A plasma panel consists of two separate glass slabs about a hundred microns. This space is filled with a gas mixture containing neon and xenon. When this gas is electrically excited, the electrons gravitating around the nuclei are extracted and become free. The "plasma" means this gas in the excited state. On each of the two slabs of the panel are screen printed line electrodes for a slab and column for the other panel. The number of row and column electrodes corresponds to the panel definition. During manufacturing, a barrier system is implemented space to physically delimit the cells of the panel and limit the phenomena of diffusion from one color to another. Each crossing of a column electrode and a row electrode will correspond to a video cell containing a volume of gas. A cell will be called red, green or blue depending on the phosphor deposit with which it will be covered. A video pixel being composed of a triplet of cells (one red, one green and blue), so there are three times more column electrodes than pixels on a line. On the other hand, the number of line electrodes is equal to the number of lines of the sign. Given this matrix architecture, just come apply to the crossing of a row electrode and a column electrode a potential difference to excite a specific cell and thus obtain punctually a gas in the plasma state. UV generated during excitation of gases are going to bombard the red, green or blue phosphors and give thus a red, green or blue cell lit.

One line of the plasma panel is addressed as many times as there are defined of sub-scans in the gray level information to be transmitted to the pixel, as explained below. The pixel selection is made by the transmission of a voltage called registration pulse, via a supply circuit, along the entire line corresponding to the selected pixel while the information corresponding to the gray level value of the pixel selected is transmitted in parallel on all the electrodes of the column on which the pixel is located. All columns are supplied simultaneously, each of them with a value corresponding to the pixel of this column.

Each bit of gray level information is associated with a time information which therefore corresponds to the ignition time of the bit or more overall at the time between two inscriptions: a bit of weight 4 with the value 1 will thus correspond to an ignition of the pixel for a duration 4 times higher an ignition corresponding to the weight bit 1. This holding time is defined by the time separating the registration top from an erasure top and corresponds to a holding voltage which precisely allows the excitation of the cell after addressing. For a gray level coded on n bits (this is the gray level for each of the components R G B), the panel will be scanned n times to transcribe this level, each of these subscans having a duration proportional to the bit it represents. By integration, the eye converts this "global" duration corresponding to the n bits in an ignition level value. A sequential scan of each of the bits of the binary word is therefore carried out in applying a duration proportional to the weight. The addressing time of a pixel, for a bit, is the same regardless of the weight of this bit, which changes is the ignition hold time for this bit.

Overall, a cell therefore has only two states: excited or not excited. As a result, unlike the CRT, it is not possible to carry out an analog modulation of the level of light emitted. To report on different levels of gray, there must be a temporal modulation of the duration of transmission of the cell in the frame period (called T). This period frame is divided into as many sub-periods (sub-scans) as there are bits coding of the video (number of bits called n). From these n sub periods, we must be able to combine all the gray levels by combination between 0 and 255. The eye of the observer will integrate over a frame period these n sub-periods and thus recreate the desired gray level.

A panel is made up of NI lines and Nc columns supplied by NI line supply circuits and Nc column supply circuits. The generation of the gray levels by time modulation requires addressing the panel n times for each pixel of each line. The matrix aspect of the panel will allow us to simultaneously address all the pixels on the same line by sending an electrical pulse of Vccy level to the line supply circuit. The signals transmitted on the columns are called column control words and relate to the video signal to be displayed, this relationship being for example a transcoding function of the number of bits used. On each of the columns will be present the video information corresponding to the bit of this column control word addressed at this time (corresponding to a sub-scan), it will be materialized by an electrical pulse of "binary" amplitude 0 or Vccx (translating the state of the coded bit). The conjugation of the two voltages Vccx and Vccy at each electrode crossing will cause or not excitation of the cell. This excitation state will then be maintained over a period proportional to the weight of the underscan performed. This operation will be repeated for all the lines (NI) and for all the addressed bits (n). We must therefore address nx NI lines during the duration of the frame, hence the following fundamental relationship: T ≥ not. NOT l .t ad where tad is the time required to address a line.

A sequencing algorithm makes it possible to address all the lines n times, respecting the respective weight of the underscan between each addressing made.

Let us rely on figure 2 to better explain the phenomenon of contouring.

In this figure, the abscissa axis represents time and is divided in frame periods of duration T. Each frame period is divided into sub periods of time whose duration is proportional to the weight of the different sub-scans allowing you to define a video level to be displayed on the screen plasma, (1, 2, 4, 8 ..., 128) for 8-bit quantized video and addressing with 8 sub-scans.

The ordinate axis represents the level 0 or 1 of the address bits during the corresponding frame periods, in other words the off state or on of a cell as a function of time, for a given level of coding.

Curve 5 corresponds to a coding of the value 128, curve 6 to a coding of the value 127 and curve 7 to a coding of the value 128 during the first frame and the value 127 during the second frame and vice versa for the next two frames.

The principle of temporal gray level modulation implies a temporal distribution of the n subscans which transcribe the video over the 20 ms of the frame. If we take an addressing on 8 sub-scans (n = 8) the 127/128 and 128/127 transitions result in all bit switching. The 8 sub-scans being distributed over the 20 ms of the frame, the eye by integrating asynchronously the video, shows black areas, part b of the curve 7 corresponding to a level 0 during the duration of two frames successive, and white areas, part a of curve 7 corresponding to a level 1 for the duration of two successive frames.

The contouring phenomenon manifests itself particularly on moving areas where strong transitions exist (contours of objects) or more generally switching at the most significant level in coding of this video. In the case of a color screen, this takes the form of the appearance on the panel, at these contours, of "false colors" due to an incorrect interpretation of the triplet R G B. This phenomenon is therefore linked the video level timing system and the fact that the eye in his role as an integrator, incorrect contours appear.

One solution to this problem consists in coding the gray level to be transmitted on more bits than is theoretically necessary (8 for coding 256 levels) and thus defining more sub-scanning to better distribute the information in time. In fact, by increasing the number of sub-scans, the respective weights of the sub-scans are reduced, the problems during their switching are limited. Currently, taking into account the characteristics of the panels (number of NI lines) and the time required to address a line (tad), it is possible to perform 10 sub-scans (n = 10) in 20ms. A grayscale transcoding will be for example:
1 2 4 8 16 32 32 32 64 64.

The heaviest weights can therefore be 64 instead of 128.

This solution however applies to the detriment of the quality of the image, the resolution being limited accordingly.

To render a gray level on a plasma panel, it is necessary to perform a time modulation of this level by performing n successive subscans during a frame. The algorithm of sequencing of this addressing leads to performing, in a nested manner, n panel underscans. However, for the sake of simplification of the algorithm and the device implementing this addressing, a line l + 1 is always addressed just after a line I for a given underscan.

A sequencing algorithm according to the prior art is shown in the FIG. 3 and is set out below in order to facilitate understanding of the invention, exposing the differences from this prior art.

This sequencing algorithm is known by the English name Simultaneous Addressing Scanning or SAS, i.e. addressing scanning simultaneous. It makes it possible to address all the lines n times (corresponding to the number n of bits) respecting the duration between each addressing corresponding to the weight of the bit relating to this addressing. Each of the lines is addressed for each of the subscans in an order defined as the shows figure 3 for a system with 4 sub-scans.

The horizontal axis represents time t and the vertical axis the number of line. On the time axis are indicated the periods corresponding to the different subscans SB0 to SB3 for bits 0 to 3 of column control words defining the luminance value to display. The display time, in fact the holding time after registration, depends on the weight of the bits, of this word control. These durations are represented, for each of the bits 0 to 3, by two oblique solid lines framing each of the mentions SB0 to SB3, for example the holding time referenced 8 for the SB3 underscan. The shaded areas 9 and 11 correspond to the scanning of the previous frame and the next frame and the intermediate area 10 corresponds to the scanning of the current frame.

It thus appears that, for a given underscan, the lines are addressed in ascending order. By cons there is overlapping of different underscans, which implies that we successively address a line of the top of the panel for sub-scanning SB1 for example and a bottom line of the panel for the sub-scan SB2 the next instant. In a way practical, four consecutive lines are addressed successively in a addressing cycle which therefore sends four write pulses before the cycle maintenance.

Thus, if we consider for example the vertical strip 12 corresponding to a short instant dt, the intersections with the oblique lines successively represent the beginnings of registration relating to the sub-scans SB3, SB2, SB1 and SB0 of the same frame ( in this example) which reported on the ordinate axis correspond to line numbers l ₃ , I ₃ +1, l ₃ +2, l ₃ +3, for example 100 and the following lines 101, 102 and 103 for SB3 , I ₂ , I ₂ +1, I ₂ +2, I ₂ +3 for SB2, etc. These addressing of the 4 times 4 lines takes place during a time interval dt. The next moment will write lines 104, 105, 106, 107 for SB3 and so on.

The new addressing process, the subject of this request, allows to perform, at different times (and not successive), the writing of the lines I and l + 1. It is actually a question of nesting 2 addressing algorithms, one for even lines and the other for odd lines. Overall, everything is happening as if there were no longer an algorithm of n subscans on NI lines, but rather an algorithm of 2 * n sub-scans on NI / 2 lines. During a cycle addressing, we no longer address the 4 successive lines (I, I + 1, I + 2, I + 3) but the lines of 2 in 2, either (I, I + 2, I + 4, I + 6) or (I + 1, I + 3, I + 5, I + 7) depending on the parity line. This change in addressing mainly concerns the generation of the sequencing of the addresses of the different sub-scans.

Figure 4 shows how, in time, the 2 algorithms are nested. Everything happens as if we had in this case 8 sub-scans, each applying to a line parity only (even or odd).

The solid oblique lines correspond to the sub-scans SB0 to SB3 and the oblique dotted lines in the subscreens SB'0 to SB'3.

For example at an instant t, the line addressed for the subscanning SB3 is an even line l ₃ (in fact the group of four successive even lines I ₃ , I ₃ +2, l ₃ +4, I ₃ +6) , the line addressed for the sub-scan SB'2 is an odd line l ' ₂ (in fact the group of four odd lines I' ₂ , 1 ' ₂ +2, 1' ₂ +4, I ' ₂ , + 6 ) and so on for the other subscans at this time t.

Note that, if the even line l ₃ is written at time t, the next odd line I ' ₃ = I ₃ + 1 is written at a different time t'.

The address separation system for lines I and I + 1 implies therefore that the times of addressing of these lines are different. In consequently, when we address line l, we are in a phase maintenance on line l + 1. It is actually possible to erase the lines I and l + 1 and write the same video information on the 2 lines as explained further. In the same way, it is possible to write the information that on line l, in this case the maintenance phase of line l + 1 will not be disrupted.

The nesting of the subscans SB 'in the subscans SB can be completely arbitrary and it is not necessary that any correlation exists between the instants of underscan of these two types (underscan type SB for even lines and sub-scan type SB 'for lines odd). Similarly, maintenance times can be completely decorated and only depend on the bit weights of the words of column command which will be assigned to each type of underscan. The weight of the column command words can be chosen different for the SB subscanning and for SB 'subscanning.

The diagrams in FIGS. 5 and 6 represent timing diagrams of two successive lines I and I + 1 and the writing instants W for these lines.

Designations of type SB1 mean that it is sub-scan 1 (bit n = 1) for an SB type underscan.

T1 represents the corresponding sub-scan hold time SB1 (bit n = 1).

The arrows appearing on the WI line correspond to the instants writing for line I.

Line l + 1 is controlled by a nested subscanning SB ' as previously stated.

Designations of type SB'1 mean that it is sub-scan 1 (bit n = 1) for an SB 'type underscan.

T'1 represents the corresponding sub-sweep holding time SB'1 (bit n = 1).

The arrows appearing on the line Wl + 1 correspond to the instants writing for line I + 1.

• sur la ligne I :
  • un sous-balayage 2 SB2 durant T2
  • un sous-balayage 3 SB3 durant T3
• sur la ligne I+1
  • un sous-balayage 1 SB'1 durant T'1
  • un sous-balayage 2 SB'2 durant T'2
  • un sous-balayage 3 SB'3 durant T'3.

The diagram of FIG. 5 is to be compared to that of FIG. 4. For FIG. 5, we have:

• on line I:
  • a sub-scan 2 SB2 during T2
  • a sub-scan 3 SB3 during T3
• on line I + 1
  • a sub-scan 1 SB'1 during T'1
  • a sub-scan 2 SB'2 during T'2
  • a sub-scan 3 SB'3 during T'3.

The entry orders are specific to a single line, the durations sub-scans are independent from one line to another.

Referring to Figure 4 and considering for example the instant t, we note that, for a line l3, we start the sub-scanning SB3 which is preceded by the sub-scan SB2. On the next line l3 + 1, we are, at this instant t, during sub-scanning SB'2 which overlaps the sub-scanning SB2 and the SB3 underscan, as shown in Figure 5.

Figure 6 no longer refers to Figure 4 and gives, in a way general, the principle of the invention using a nested scan.

The first timing diagram corresponds to line I and represents 4 sub-scans successive Sb1 to Sb4 of holding time t1 to t4.

The second timing diagram corresponds to line l + 1 and represents 4 successive subscans Sb'1 to Sb'4 of holding time t'1 to t'4.

• sur la ligne I :
  • un sous-balayage 1 Sb1 durant t1
  • un sous-balayage 2 Sb2 durant t2
  • un sous-balayage 3 Sb3 durant t3
  • un sous-balayage 4 Sb4 durant t4
• sur la ligne I+1
  • un sous-balayage 1 Sb'1 durant t'1
  • un sous-balayage 2 Sb'2 durant t'2
  • un sous-balayage 3 Sb'3 durant t'3
  • un sous-balayage 4 Sb'4 durant t'4.

We have:

• on line I:
  • a sub-scan 1 Sb1 during t1
  • a sub-scan 2 Sb2 during t2
  • a sub-scan 3 Sb3 during t3
  • a sub-scan 4 Sb4 during t4
• on line I + 1
  • a sub-scan 1 Sb'1 during t'1
  • a sub-scan 2 Sb'2 during t'2
  • a sub-scan 3 Sb'3 during t'3
  • a sub-scan 4 Sb'4 during t'4.

To go from the first case (fig. 5) to the second case (fig. 6), just pool (for the two lines I and I + 1), the 3 write signals which was specific (to each line) in the first case. Write signals added are circled in Figure 6 and designated with the reference 13.

So, by adding a first and a second write signal on the line I (always preceded by an erase signal for underscan previous), the holding time T2 is divided into two periods t1 and t2 and the holding time T3 in two periods t3 and t4.

For the next line l + 1, the addition of the write signal makes it possible to split the holding time T'2 into two periods t'2 and t'3.

From the nested sub-scans of type SB and SB ', we can therefore reduce to common sub-scans between lines I and l + 1, both in duration in video content (which is either zero or one). It is thus possible perform a line copy. We will describe as "partial" this copy of line as long as it is done on demand. Indeed, the operation which been performed in the example for the 3 writes can be reduced to 0 (this is the first case), with one or two scripts.

We talk about partial and on-demand copying because we introduce a notion variable parameter that can be set depending on the video content.

The big advantage of this method is that you can easily switch from a 16 sub-scan mode to a 13 sub-scan mode (see example given below) from one frame to another and without a transition cycle. The adaptation can therefore be made according to the content of the sequence and even depending on the content of the image. A system for measuring the vertical resolution can be used to make a decision on the number of sub-scans to use. The method even allows for a couple of lines to another, from a mode 13 to 16 sub-scans. Decision information can be calculated for each couple of lines.

In what follows, we will explain the principle of separation of information between a common value and specific values, a process that can be combined with our invention.

The coding of a gray level according to this principle, which results in a column command word, is carried out taking into account not only the luminance value of the selected pixel but also the value of luminance of the pixel on the adjacent row for the same column.

In fact, the column control word, for a given pixel, is separated into two parts, a first command word corresponding to a value common to the two pixels and a second and third word of command corresponding to the specific pixel values.

• une valeur spécifique à la ligne I codée sur n1 bits
• une valeur spécifique à la ligne l+1 codée sur n2 bits
• une valeur commune aux lignes I et l+1 codée sur n3 bits

   avec la relation suivante : n1 + n2 + n3 = 2 × (nombre de sous-balayages par ligne). We wish to obtain the following coding:

• a value specific to line I coded on n1 bits
• a value specific to line l + 1 coded on n2 bits
• a value common to lines I and l + 1 coded on n3 bits

with the following relation: n1 + n2 + n3 = 2 × (number of sub-scans per line).

If we consider a given number of sub-scans, we must effect that the number of subscans relating to the coding bits of the two specific values and the common value, which is n1 + n2 + n3, corresponds to that of the conventional sweeps and relating to the coding bits for line I and to the coding bits for line I + 1.

These different parameters n1, n2, n3 are not fixed. It is possible to modulate the relationship between the definition of specific values and that of common value. Loss of resolution due to coding will be the lower the specific values will be the better defined. Through however, the total number of underscans will be higher as the specific values will be the least well defined. So there is a compromise to find between loss of resolution on the one hand and minimization of defects of viewing each other.

The calculation of specific values is carried out as follows:

The specific values for lines I and I + 1 contain the difference information between these lines I and I + 1. Indeed, if we call NG1 and NG2 the gray levels of the pixels of lines I and l + 1, VS1 and VS2 their specific values and VC the common value, we have the relation: NG1 = VS1 + VC NG2 = VS2 + VC

Therefore, VS1 - VS2 must be equal to NG1 - NG2 (always to have a zero coding error): When we have determined this difference between NG1 and NG2 (called D), we calculate VS1 and VS2 by addition of the term D and of a portion a of the lowest gray level.

We then have: if NG1> NG2 VS1 = D + αNG2 VS2 = αNG2 if NG2> NG1 VS1 = αNG1 VS2 = D + αNG1

The value of α is a parameter to be defined in the same way as n1, n2, n3. This value α is the result of algorithmic tests and is therefore partially determined empirically. The value is chosen in function of induced calculations, for example the value 3/16 facilitating calculations by the digital signal processor DSP (Digital Signal Processing in English).

The common value is calculated by difference between the initial value and the specific value. Taking into account the approximations made on the calculation of the specific values, the common value is obtained by the following calculation: VC = 1/2 × (NG1 + NG2 - VS1 - VS2)

• détermination de la valeur D correspondant à la différence entre les deux valeurs à coder NG1 et NG2.
• calcul des valeurs spécifiques VS1 et VS2 en fonction de D, a et NG1 ou NG2.
• calcul de la valeur commune VC en fonction de NG1, NG2, VS1, VS2.

The calculations therefore boil down to the following steps:

• determination of the value D corresponding to the difference between the two values to be coded NG1 and NG2.
• calculation of specific values VS1 and VS2 as a function of D, a and NG1 or NG2.
• calculation of the common value VC as a function of NG1, NG2, VS1, VS2.

An important point consists in minimizing the recoding error. To be able to minimize this recoding error, we will use a particular coding of the specific value. This is a coding in steps of 5, ie each code is a multiple of 5. The following table shows how the specific and common values are calculated to obtain, in the end, the values VF1 and VF2 as close as possible to NG1 and NG2. In fact the error (E1, E2) is limited to +/- 1. NG1 NG2 D D by 5 VS1 VS2 VC VF1 VF2 E1 E2 60 65 5 5 10 15 50 60 65 0 0 60 66 6 5 10 15 50 60 65 0 -1 60 67 7 5 10 15 51 61 66 1 -1 60 68 8 10 10 20 49 59 69 -1 1 60 69 9 10 10 20 49 59 69 -1 0

The difference D between the gray values is coded from the most near multiple of 5 of this value D. Specific values VS1 and VS2 are multiples of 5 and the proportion of the specific value to the global value (the parameter α) is chosen equal to 3/16. The value of VS1 is thus the modulo 5 value closest to 60 x 3/16.

The specific value, which contains the difference information between the two coded pixels, is defined only on a restricted number of bits. The maximum difference that can be coded will therefore be limited in fact to the value maximum that can be coded as a specific value. So this is going to prohibit coding large differences.

For a strong transition, the difference that can be coded being limited, one of the specific values will be equal to the maximum value and the other will be equal to 0. The common value will be determined so that minimize the error on the final value. In this case, the final error may be greater than 1.

The following table gives an example of a coding between 2 pixels whose difference is greater than the maximum definition of the specific value. The maximum value chosen for the specific value is taken equal to 70: NG1 NG2 D D by 5 limited VS1 VS2 VC VF1 VF2 E1 E2 10 100 90 70 0 70 20 20 90 10 -10

Définition des paramètres:

• n1 = 4 (code 5,10,20,35)
• n2 = 4 (code 5,10,20,35)
• n3 = 12 (code 1,2,4,6,9,12,15,19,23,27,31,36)
• α = 3/16

An application example using the principle of separating information between a common value and specific values is given below for a system allowing 10 sub-scans:

Parameter definition:

• n1 = 4 (code 5,10,20,35)
• n2 = 4 (code 5,10,20,35)
• n3 = 12 (code 1,2,4,6,9,12,15,19,23,27,31,36)
• α = 3/16

This actually allows us to transcribe a gray level into 16 sub-scans, 12 sub-scans being common to 2 lines (therefore equivalent to 6 conventional sub-scans) and 4 sub-scans being specific. In this case, the gain will be 6 sub-scans with an error of recoding less than or equal to 1 (for a difference between lines less than or equal to 70).

FIG. 7 shows such addressing with 16 sub-scans. On line I and the line l + 1 follow each other as a function of time the sub-scans corresponding to the bits of weight 10, 9, 15, 12, 20. The writings referenced 14 are common to lines I and I + 1, for the values 9, 15, 12. The entries referenced 15 are specific to lines I and l + 1 and relate to the values 10, 20.

The 16-bit code thus defined limits the maximum difference between the lines I and l + 1 to 70 (70 = 5 + 10 + 20 + 35). Beyond 70, the coding operation on 16 bits causes the generation of an error greater than LSB.

This problem is solved by combining the principle of nesting underscans to the one previously described.

The 16-bit code above corresponds to the weight of the bits of the column control words calculated from the video information:
1 2 4 5 6 9 10 12 15 19 20 23 27 31 35 36

According to the principle of separation of information between a value common and specific values, each video information is separated into information specific to current line I and common information to the 2 adjacent lines I and I + 1. The specific information is coded on 4 bits whose respective weights are multiples of 5 (5,10,20,35). information common is coded on 12 bits.

The principle of nesting of sub-scans increases the value of this maximum difference from which errors are only more negligible, which is particularly useful when the resolution vertical (difference in luminance) is important.

It allows to dynamically pass from 16 sub-scans (10 common two-line subscans and 4 separate subscans) to 13 subfields.

First of all, the respective order of the different sub-scans is modified as follows:
1 2 4 6 5 10 9 15 12 20 19 23 27 31 36 35

This order defines the rank of the bits of the transmitted control words, represented by their weight.

The first 4 sub-scans (1, 2, 4, 6) are always common to the 2 adjacent lines. The sub-scans 5 and 10 and also 20, 35 are them always specific to lines I and l + 1 (so we always have 2 pieces of information different for these subscans).

For the next 3 sub-scans (9, 15, 12) two cases are possible: either they are common to the 2 lines (and we then return to addressing to 16 sub-scans) or they are partially specific (addressed to 13,14 or 15 sub-scans).

FIG. 8 shows such addressing with 13 sub-scans. On the line I succeeds the sub-scans corresponding to bits of weight 10, 24, 12, 20. On line l + 1 follow each other the sub-scans corresponding to bits of weight 10, 9, 27, 20. The writings referenced 16 are common on lines I and l + 1, for the values 9 and 24. The entries referenced 17 are specific to lines I and I + 1 and relate to the values 10, 20, 12 and 27. In done, it is the inscription relative to the sub-scan 9 which is common but we do not does not erase line l at the end of the maintenance cycle. If there is not erasure, the information entered remains present, which implies that the video information which for weight 9 on line l + 1 has a different weight (24) on line l. On the other hand, the line l + 1 is deleted at the end of the weight cycle 9. A this moment, we just write the following video information (which corresponds to 15 in 16 sub-sweep mode) on line l + 1. Similarly, we do not does not erase at the end of the weight cycle 15 the line l + 1 but the line l. We have so on line l, a sub-scan of duration 24 (9 + 15) whose video content is the same as the duration 9 sub-scan of line I + 1. We then write the video content of the sub-scan 12 on line l. Similarly, when the entry of 12 on line l, there was no erasure on line l + 1. Consequently, the sub-scan 15 of the line l + 1 actually lasts 27 (15 + 12). An erasure signal common to lines I and I + 1 is then made before entering the video information corresponding to the specific values of weight 20.

In conclusion, in the 16 sub-sweep mode, we had 3 sub-sweeps successive commons of respective weight 9,15,12. In mode 13 sub-scans we actually have 2 sub-scans 24 and 12 on line I and 2 sub-scans 9 and 27 on line I + 1. Only constraint, information 24 of the line I and 9 of line l + 1 are common. On the other hand the weights 12 of line I and 27 from line l + 1 are specific. We therefore increase the proportion of specific values compared to common values which allows a higher vertical resolution.

Similarly, the subscans 19, 23, 27, 31, 36 of a addressing 16 sub-scans, can be transformed into 3 sub-scans 42, 58, 36 for line I and 19, 50, 67 for line I + 1. Only constraint, the video information of sub-scan 42 of line I is the same as that of underscan 19 of line I + 1.

For the coding of the values 9, 15, 12, we saved an underscan, for the coding of the values 19, 23, 27, 31, 36, we save two other underscans.

• 4 écritures correspondant à 4 sous-balayages communs (1, 2, 4, 6)
• 4 x 2 écritures correspondant à 4 sous-balayages spécifiques (5, 10, 20, 35)
• 1 écriture correspondant à 1 sous-balayage commun (9 + 15 pour I et 9 pour I+1 se limitant à 1 commande d'écriture commune aux deux lignes pour le sous-balayage 9)
• 1 x 2 écritures correspondant à 2 sous-balayages spécifiques (12 pour I et 15 + 12 pour I+1)
• 1 écriture correspondant à 1 sous-balayage commun (19 + 23 pour I et 19 pour I + 1 se limitant à une commande d'écriture commune aux 2 lignes pour le sous-balayage 19)
• 1 x 2 écritures correspondant à 2 sous-balayages spécifiques (27 + 31 pour I, 23 + 27 pour I+1)
     1 x 2 écritures correspondant à 2 sous-balayages spécifiques (36 pour I, 31 + 36 pour I+1).

Taking into account the specific sub-scans and those common to two lines, let us calculate the number of writes for two successive lines to check the average number of sub-scans per line:

• 4 entries corresponding to 4 common sub-scans (1, 2, 4, 6)
• 4 x 2 entries corresponding to 4 specific sub-scans (5, 10, 20, 35)
• 1 write corresponding to 1 common underscan (9 + 15 for I and 9 for I + 1 limited to 1 write command common to both lines for underscan 9)
• 1 x 2 entries corresponding to 2 specific sub-scans (12 for I and 15 + 12 for I + 1)
• 1 write corresponding to 1 common underscan (19 + 23 for I and 19 for I + 1 limited to a write command common to the 2 lines for underscan 19)
• 1 x 2 entries corresponding to 2 specific sub-scans (27 + 31 for I, 23 + 27 for I + 1)
  1 x 2 entries corresponding to 2 specific sub-scans (36 for I, 31 + 36 for I + 1).

Totaling: 4 + 8 + 1 + 2 + 1 + 2 + 2 = 20 entries.

There is indeed an average of 10 entries for a line.

In another way, we can say that the column command words were coded on 16 bits and, depending on the weight of the bits, the lines were addressed separately or 2 by 2. The scanning times for writing the 2 bits, for which lines were addressed 2 by 2, were therefore divided by 2, reducing the sweep time to that of a control word column of 10 bits (4 + 12/2).

According to the principle of nesting of sub-scans, the words of column command are coded on 13 bits, bits being common to two successive lines.

These column control words have bits of different weight depending on that the line considered is an even or odd line.

• pour une ligne paire (ou impaire selon son choix):
  1, 2, 4, 6, 5, 10, 24, 12, 20, 42, 58, 36, 35
• pour une ligne impaire (respectivement paire):
  1, 2, 4, 6, 5, 10, 9, 27, 20, 19, 50, 67, 35

The weights of the column control words coded on 13 bits (13 sub-scans) are:

• for an even line (or odd according to its choice):
  1, 2, 4, 6, 5, 10, 24, 12, 20, 42, 58, 36, 35
• for an odd line (respectively even):
  1, 2, 4, 6, 5, 10, 9, 27, 20, 19, 50, 67, 35

The weights of bits of rank 7 and 8 have the same sum 36. The weights of bits of rank 10, 11, 12 have the same sum 136.

The lines are addressed 2 by 2, in the example, for the weights:
1, 2, 4, 6, 9 or 24, 19 or 42 (depending on the column command word considered).

The lines are addressed separately for the weights 5, 10, 20, 35.

The lines are addressed separately for the weight (15 + 12), (23 + 27), (31 + 36).

The lines are addressed separately for the weight 12, (27 + 31), 36.

We obtain a scanning time for writing which corresponds to 10 bits: 4/2 + 2/2 + 4 + 6/2 = 10

Overall, thanks to the invention, we go from a maximum difference from 70 for 16 subscans to a difference of 176 (255-42-24-13) for 13 underscans (the values 9/24 and 19/42, like the weights 1, 2, 4, 6 do not can indeed be selected separately). This therefore allows significantly increase the transmitted vertical resolution.

The big advantage of this technique is to be able to perform the switching between 16 sub-scan addressing and 13 addressing sub-scans on demand and for a given couple of lines. It is possible for example to detect upstream areas of the image with strong vertical transitions. All the lines in this zone will then be passed in addressing to 13 sub-scans, the others being able to remain addressed to 16 subfields. This switching, which corresponds to the passage of a addressing according to figure 8 to addressing according to figure 7 is done in a simple way, by replacing the selection of a line I (or a line I + 1) when writing a bit of different weight on line I (or I + 1) by the selection of line I and the immediately next (or previous) line for simultaneous writing on these two lines.

In the same way, it will be interesting to have a detector of "false contours" to judge the need to remain in 16 sub-sweep mode or not. There is a trade-off between vertical resolution and limiting the level of "false contours".

This number of subscans is related to the number of bits having different weights of a column command word corresponding to a row at command word column corresponding to the following line and this number, therefore the column control words used for coding the image, may be chosen according to the images to be processed, this choice can also be performed frame by frame. The weight of the bits concerned can be chosen in depending on image resolution.

The cell priming and quantification problems described previously can be mitigated as follows:

Using the sole principle of separating the addresses of lines I and l + 1, it is possible to improve, in a fairly simple way, the priming of excitations. Indeed, during a conventional addressing, the 4 cells addressed to the during the current cycle are first extinguished by a pulse erasure. The registration which follows immediately after cannot benefit from an effect. proximity of lit cells. The only cells likely to be lit are those located just above or below the pack of 4 lines.

In our case, lines I and l + 1 being addressed at instants different, line l + 1 can benefit from the possible state of excitation of the lines I and I + 2, these have not been extinguished just before. In fact it is possible to to benefit from all the sub-scans of all the lines of this system.

To promote the initiation of all sub-scans, it suffices to have write times on the even and odd lines which are systematically different. A simple way to do this is to shift the 2 addressing systems of a constant time, while keeping in this case the same code on the 2 lines. It is for example possible to use a double addressing system offset from each other by the equivalent of 1/2 LSB.

In the example of figure 8, configuration with 13 sub-scans, certain underscans benefit from this favored priming.

Regarding the quantification of low levels, if we consider 2 separate addresses for odd lines and even lines, it is possible, as indicated previously, to carry out at a given time, a common registration for 2 adjacent lines. This amounts for example to stop the maintenance phase of a sub-scan of line I and come and register on lines I and I + 1 the video information of line I + 1. The duration of the underscan in this case the initial value of line I is reduced. Apply this principle for the least significant sub-scan (duration corresponding to the LSB) amounts to introducing a lower quantization step at LSB. The phase shift between the 2 addresses can be chosen equal to 1/2 LSB. If we apply the principle of addressing common to the 2 adjacent lines, we thus defines 1/2 LSB weight sub-scans. This saves us a quantization levels usable especially for low levels. It is also possible to define an addressing system allowing to increase further this quantification by introducing the weight of 1/4 of LSB.

An exemplary embodiment of the device implementing the method of scanning is described below. The simplified diagram of the control circuits a plasma panel 18 is shown in FIG. 9.

Digital video information arrives at input E of the device which is also the input of a video processing circuit based on microprocessor 19 and the input of a selection circuit 20. The video processing is connected to a correspondence memory 21, to the selection 20, at the input of a video memory 22 and at a scanning generator or control circuit for power supply circuits on line 24. Video memory transmits the stored information to the input of a circuit 23 grouping the column supply circuits.

The scan generator 24 transmits information from synchronization to video memory 22 and controls a circuit 25 grouping the line supply circuits.

The video information coded on 8 bits and received on input E is thus transmitted to the selection circuit 20 which stores the video data on a full picture. This circuit analyzes the content of the video and calculates the number of times there is a difference in luminance in the image between the line I and line l + 1 greater than a preset threshold.

If this number is greater than a predetermined threshold, scanning is carried out by exploiting the principle of nesting of the sub-scans, that is to say from an address with 13 sub-scans. Otherwise, 16 sub-scans are performed. The type of scan information is transmitted to the processing circuit 19 which carries out the coding of the information video accordingly. The processing circuit transmits this information to scanning circuit 24 so that it scans the screen as a function of this coding.

The processing circuit 19 exchanges the video data with the memory or correspondence table 21 which, depending on the values of the video words sent as addresses, will supply as data corresponding words to 13 or 16 bit codes whose weights will have been defined beforehand. This transcoding from the correspondence table 21 is defined as a function of the addressing mode used.

• un premier type de codage fournissant un premier mot de codage correspondant aux lignes paires du panneau à plasma
• un deuxième type de codage fournissant un deuxième mot de codage correspondant aux lignes impaires du panneau à plasma.

When the 13 sub-scanning addressing mode is selected, the 13-bit coded words correspond to two types of coding which are differentiated by the bit weight of the coded words:

• a first type of coding providing a first coding word corresponding to the even lines of the plasma panel
• a second type of coding providing a second coding word corresponding to the odd lines of the plasma panel.

These words are then transmitted to the video memory 22 which makes them stores to supply the column supply circuits, in synchronization with line scanning, the successive bits of the column control words.

The scanning generator 24 performs, for the duration of a frame and via line supply circuits 25, the line sweep of the screen. This circuit 25 supplies the addressing voltage and also the voltage of hold for the duration corresponding to the sub-sweep relative to the weight of the bit sent on the columns for this addressing.

The scan generator 24 performs the sub-scans as a function commands received from the processing circuit.

• un balayage des lignes sélectionnées deux à deux (sélection simultanée des lignes 2I et 2I+1)
• un balayage de chaque ligne successive.

The types of scans used are:

• a scan of the selected lines two by two (simultaneous selection of lines 2I and 2I + 1)
• a scan of each successive line.

Changing from a 13-scan mode to a 16-scan mode is done, in a very simple way by selecting the lines 2I and 2I + 1 instead of the single 2I line or the single 2I + 1 line when writing bits corresponding to the common value VC.

It should be noted that the selection circuit 20 can very well be placed in upstream of the device and in particular of the processing circuit in order to avoid any delay in coding video words.

Obviously, the previous description assumed a selection line of the plasma panel for transmission of video information on the entries in the display columns, but other types could be considered addressing, for example by reversing the function of rows and columns without the process being outside the scope of the invention.

Of course, the invention is not limited by the number of bits quantifying the digital video signal to view, nor the number of sub-scans.

It can also be applied to any type of screen or device to matrix addressing exploiting a temporal type modulation for the display of luminance or gray levels corresponding to each of the three components R V B. The cells of this device or matrix table with row and column entries, the term cell being taken here at broad sense of elements at the intersection of rows and columns, can be plasma panel cells but also micromirrors of circuits to micromirrors. Instead of emitting light directly, these micromirrors reflect, from time to time (a cell corresponding to a micromirror), received light, when selected. Their addressing for the selection is then identical to the addressing of the cells of the panels to plasma as described in the present application.

Claims (13)

1. Process for addressing cells arranged as a matrix array, each cell being situated at the intersection of a line and a column, the array having line inputs and column inputs for displaying grey levels defined by video words making up a digital video signal and defining an image, the column inputs each receiving a control word for this column corresponding to the video word relating, for this column, to an addressed line, this word being composed of n bits transmitted sequentially, each sequence corresponding to a sub-scan, each bit triggering or not, according to its state, the illumination of the cell of the addressed line and of the column receiving the control word, for a time proportional to the weight of this bit in the word, characterized in that a different coding of the column control words is performed depending on whether the word relates to an even or odd line, this difference consisting in the fact that at least m successive bits of specified ranks, m being between 2 and n, have different weights from one control word to the other, the sum of the weights of these bits remaining identical from one control word to the other, so as to obtain writing instants which are substantially different from one line to the next.
2. Process according to Claim 1, characterized in that writing is simultaneous on two successive lines for at least the first bit of the m successive bits of a control word relating to one of the two lines.
3. Process according to Claim 1 or 2, characterized in that at least two successive lines are selected simultaneously for at least one of the bits of a specified rank, which has an identical weight from one control word to the other.
4. Process according to one of the preceding claims, characterized in that at least one of the bits of a specified rank, which has an identical weight from one control word to the other, is used to code a partial value of luminance common to two successive lines and in that writing is simultaneous on these lines for this bit of the control word relating to one of the two lines.
5. Process according to Claim 1, characterized in that it is implemented for a limited number of lines of the matrix array, these lines corresponding to the zones of the image defined according to the luminance variations from one line to the other, the other zones utilizing sub-scans corresponding to an addressing process for which the column control words all have the identical weights from one line to the other.
6. Process according to Claim 1, characterized in that it is implemented for a number of images defined according to the luminance variations from one line to the other, the other images utilizing an addressing process for which the column control words all have the identical weights from one line to the other.
7. Process according to Claim 1, characterized in that the switchover from the first addressing process comprising n sub-scans to a second addressing process comprising a larger number of sub-scans and for which the column control words have a larger number of bits having identical weights from one line to the other is performed by replacing the selection of a line I while writing a bit of different weight on the line I, in the first process, by the selection of the line I and of the immediately preceding or immediately following line for a simultaneous writing on these two lines, in the second process.
8. Process according to Claim 1, characterized in that the value of m or that of the weights corresponding to these m bits is dependent on the vertical resolution of the image.
9. Process according to one of the preceding claims, characterized in that the cells are cells of a plasma panel and in that the selection causes the illumination of the cell.
10. Process according to one of claims 1 to 8, characterized in that the cells are micromirrors of a micromirror circuit.
11. Device for implementing the process according to Claim 1 comprising a video processing circuit (19) for processing the video data received, a correspondence memory (21) for transcoding these data, a video memory (22) for storing the transcoded data, the video memory being linked to column supply circuits (23) for controlling the column addressing of the plasma panel on the basis of column control words, a control circuit (24) for the line supply circuits (25) linked to the video processing circuit so as to select the lines, characterized in that the video processing and transcoding circuits perform a different coding of the column control words depending on whether the word relates to an even or odd line, this difference consisting in the fact that at least m successive bits of specified ranks from among the bits to be transmitted, m being between 2 and n, have different weights from one control word to the other, the sum of the weights of these bits remaining identical from one control word to the other, so as to obtain writing instants which are substantially different from one line to the next.
12. Device according to Claim 11, characterized in that the circuit for controlling the line supply circuits simultaneously selects two consecutive lines during the transmission by the column supply circuits of the first bit of the successive bits of a control word relating to one of the two lines.
13. Device according to Claim 11, characterized in that it also comprises a selection circuit (20) receiving the video data so as to select a coding of the column control words corresponding to an addressing according to n sub-scans or to an addressing corresponding to a larger number of sub-scans, as a function of the variations in luminance from one line to the other in an image or an image part.

Images