CN104637455A - Scheme for adjusting image data of LCD (liquid crystal display) by aid of local darkened backlight - Google Patents

Abstract

The invention relates to a scheme for adjusting image data of an LCD (liquid crystal display) by the aid of local darkened backlight, and particularly discloses a method for accurately and effectively compensating pixels in local darkening procedures. The scheme has the advantages that a method for inhibiting visible tailored artifacts by means of applying local tailoring inhibiting factors to reduced images is introduced for conserving a large quantity of energy, high-quality images with image details and image brightness are ultimately fundamentally kept, and accordingly energy conservation and the image details can be compromised with one another at a low processing cost.

Description

Utilize the dimmed backlight in local to the adjustment of the view data of LCD

Technical field

The application relates to energy-conservation LCD, particularly by utilizing the dimmed backlight in local to realize to the adjustment of the view data of LCD the original brightness reaching image at lower cost.And cut out suppression retain image detail by application local, thus make LCD achieve between energy-conservation and image detail compromise.

Background technology

Display be modern man-machine interface a core technology and for multiple use.The example comprising the equipment of a display be widely used is computing machine, mobile phone, and traditional TV.The common type of of display technique is a LCD display (LCD).But, the power consumption of LCD along with LCD be increasingly used for movement, battery supply set and the trend of the these two aspects of larger display that formed and by growing interest.Backlight for LCD is usually the assembly of the LCD with the highest power consumption.Light emitting diode (LED) backlight a kind ofly in current used LCD, allows best display performance and the backlight type of sharpness.

LED can by PWM(pulse-length modulation) control, make the intensity of light be almost continually varying.LED light concerning the response of LED current for the mankind eyes be very short or postpone be non.For most LCD, back light unit comprises multiple LED.They can be arranged in a structure of specifying (as network or bar shaped).Described structure is after the pixel panel of LCD or in the frame of the pixel panel of LCD.

Time after the back light unit comprising multiple LED is positioned at the pixel panel of LCD, this back light unit is just called as direct LED backlight.Described direct LED backlight is used as the rear that multiple LED is configured in pixel panel equably as light emitting source, makes backlight can evenly be communicated to whole screen.And when the back light unit comprising multiple LED is placed in the frame place of the pixel panel of LCD, this back light unit is called as side light type LED backlight source, it is configured in one or more frames of pixel screen multiple LED, then light guide plate of arranging in pairs or groups.Like this, when LED backlight unit is luminous, the region that just the light transmission light guide plate of launching from screen frame can be transported to center Screen is gone.

Several LED component can be grouped into a string, and is controlled by a LED driver passage (usually in PWM mode).In text below, a LED strip with several LED component that are that controlled by a LED driver passage or that always receive substantially identical electric current is called as a LED.LED light is distributed by photoconduction.For a good backlight design, the backlight illumination distribution of LCD can be regarded as substantially uniform.Uniform backlight value maximum like this can be standardized as 100%.The gray-scale value of each pixel of image can be fed to the electronic control unit of the driving element of timing controller and/or such as TFT and so on, and without the need to any process and/or operation.

Such as, Fig. 1 shows a dimmed processing procedure in local disclosed in US8421741B2.Image is reduced to the gray-value image/matrix (frame 11) of low resolution.LED optimizer (frame 12) by substantially meet the image of reduction demand and by considering that the physical attribute of LCD determines the value of LED.In pixel compensation step (frame 13), to be used to as shown in Equation 1 such calculates backlight illumination for LED value:

$B_{\mathrm{ij}}=\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{\mathrm{Ak}}_{\mathrm{ij}}\·L_k$

Formula 1

B _ijthe backlight illumination at pixel ij place, Ak _ijthe impact of a kth LED at pixel ij place, and L _kbe the value (brightness) of a kth LED, m is the total quantity of LED.Based on original pixels gray-scale value G _ij, the transmission value t of the pixel of this LCD _ijbe adjusted to or be operating as formula 2:

$t_{\mathrm{ij}}=\frac{G_{\mathrm{ij}}}{B_{\mathrm{ij}}}$

Formula 2

In order to the picture quality obtained, as retained the details of image, retain brightness, high uniformity etc., the transmission value t of each LCD pixel _ijmust accurately be calculated.This and then the backlight illumination B requiring again after each pixel _ijalso must be determined precisely.The existing method to pixel compensation point spread function is used for pixel after the calculating of backlight illumination, the calculating of described backlight illumination only consider LED tribute adjacent on a small quantity to light.This scheme can have the LCD(of direct LED backlight after this referred to as directly-down type LCD) center in work.For there is the LCD(of side light type LED backlight source after this referred to as side-light type LCD), this scheme will cause a large amount of mistakes, because point spread function can not meet the actual performance of LCD, especially can not meet the actual performance of side-light type LCD.

Fig. 2 shows the function of a LED on the impact of the whole panel of side-light type LCD.X and Y-axis represent the coordinate (resolution 1280 × 720) of pixel.Z axis is linear, and the impact in arbitrary unit is shown.This influence matrix can be called as light spread function (being called for short LSF).Because the light sent by LED can be spread widely and cover whole panel, the backlight illumination after each pixel can control based on the LED of all LED or majority.Therefore, the calculating of backlight illumination can consider the contribution of all LED or most LED.Their total contribution will be the summation according to formula 1.In order to calculate backlight illumination, the light spread function of LED must be used.Institute's facing challenges is the labour intensity calculating the whole computation process of the backlight illumination after each pixel according to formula 1 is huge, because the quantity of the pixel of a modern LCD display is very high.In addition, because light spread function is not ground in ideal distribution, real light spread function value may need to be stored in memory.Therefore, for high display resolution, the size storing the storer of light spread function value is also huge.This can cause the high cost of computational logic and storer.Processing speed may reduce simultaneously.The power consumption of such calculating may be very high, and be especially irrational for the such as movable electrical appliances such as smart mobile phone or dull and stereotyped PC.

Summary of the invention

Content of the present invention is provided to be to introduce some concepts that will further describe in the following specific embodiments in simplified form.Content of the present invention is not intended to the key feature or the essential characteristic that identify theme required for protection, is not intended to the scope for limiting theme required for protection yet.

In order to solve the various problems existed in the dimmed processing procedure in existing local, the invention provides a kind of method for being adjusted by the view data of the dimmed backlight in local to LCD, described LCD has the back light unit comprising multiple LED controlled separately, and described method comprises: the light spread function value pre-determining a LCD model; The resolution of the light spread function of described LCD model is tapered to the resolution of the sampled point of Two dimensional Distribution, and the light spread function value of the resolution of sampled point is stored in a lookup table; The LED value of former frame or former frame is used for the calculating of the backlight illumination of sample point; The contribution of this LED to the backlight of this sample point is calculated by the light spread function value and described LED value that are applied in a LED of a sample point; By all or most LED to be calculated the backlight illumination of this sample point in the contribution summation of this sample point to backlight; By carrying out interpolation to calculate the backlight illumination after other pixel between this multiple sampled point to the backlight illumination of multiple sample point; The transmission value of each pixel is calculated based on the gray-scale value of pixel of image of input and the backlight illumination after this pixel; And the transmission value of described pixel is supplied to the electronic control unit for LCD.

According to another aspect of the present invention, provide a kind of method that not only image detail can have been retained but also substantially can keep the brightness of image, described image is displayed to the LCD of the back light unit with the LED comprising multiple independent control, and described method comprises: based on the pixel of the surrounding of sampled point signature grey scale value and as calculated or the backlight illumination of this sampled point estimated calculate a sampled point cut out inhibiting factor; By by as calculated or the backlight illumination estimated be multiplied insert a virtual backlight illumination with described inhibiting factor of cutting out; From two-dimentional sampled point benchmark to virtual backlight illumination described in pixel benchmark interpolation; The transmission value of each pixel is calculated based on the gray-scale value of pixel of image of input and the described virtual backlight illumination after this pixel; The transmission value of described pixel is supplied to the electronic control unit for LCD.

Describe in detail by reading and check associated drawings below, these and other feature and advantage will be apparent.Be appreciated that aforementioned general description and following detailed description are all only illustrative, and do not limit each side required for protection.

Accompanying drawing explanation

Below with reference to the accompanying drawings the present invention is described in detail, in accompanying drawing:

Fig. 1 is the dimmed processing procedure in the local of prior art.

Fig. 2 is the light spread function of an example.

Fig. 3 shows the example of a sampled point.

Fig. 4 shows an example of the LCD with the LED being placed on a frame place.

Fig. 5 shows the sampled point of corner regions.

Fig. 6 shows the other method that a kind of resolution for the treatment of display according to the present invention is not the situation of the multiple of a rational bigit.

Fig. 7 shows the processing procedure that a local is dimmed according to an embodiment of the invention.

Fig. 8 shows the processing procedure comprising the pixel compensation cutting out suppression according to an embodiment of the invention.

Fig. 9 a shows the stingy figure of original image.

Fig. 9 b shows according to an embodiment of the invention by cutting out the stingy figure of the image after handled by inhibiting factor.

Fig. 9 c shows according to an embodiment of the invention with by the stingy figure cutting out the image after handled by inhibiting factor of spatial filtering.

Figure 10 a illustrates the example distribution (CSF1) of cutting out inhibiting factor at the resolution place at sampled point according to formula 9.

Figure 10 b illustrates the distribution of cutting out inhibiting factor after filtering of example.

Embodiment

In order to solve the problems of the prior art, in the present invention, the dimmed imagination in a kind of local is proposed.The dimmed imagination in described local is based on the following fact: i) image may comprise a large amount of dark parts; Ii) the LED component of back light unit can be individually controlled.If carry out control LED by this way, such as, the backlight after the dark portion of image is dark, and is bright in the backlight of the light of image, and this just means, the brightness of each LED can be different, and can depend on the content of image.Brightness (this is provided by view data) needed for a pixel is that the transmission value of backlight illumination and the LCD pixel generated by LED realizes.The dimmed advantage in local is to reduce power consumption.

Brightness in order to the image by display remains on as required, and the transmission value of each LCD pixel must be adapted to backlight, and this backlight may be no longer uniform in the pixel panel of LCD, because the value of LED can be different.When dimmed backlight, this means that this backlight can lower than standardized 100%, and the transmission value of a LCD pixel can correspondingly be brought up to a suitable value, therefore, the brightness of this pixel is identical with what do not produce through the process that backlight is dimmed substantially, and this is called as the compensation of pixel.By pixel compensation, the brightness of the image of display is retained substantially; Add static contrast, make the image of display seem more lively.Therefore, be by the dimmed next energy-conservation key in local the transmission value increasing many LCD pixels.

Due to only modeling in linear domain in the present invention, the gray-scale value of image pixel is also in linear domain.This means, the view data (this is normally encoded by a gamma function) of input will be processed by a Gamma correction, described correction generation one linear grey value.Output image data of the present invention is the grey scale pixel value operated in linear domain.In order to meet the view data of gamma coding, inverse gamma function is applied to the grey scale pixel value of operation.These two operations can be performed by two look-up tables (LUT) and before the input of process disclosed in this invention and performing afterwards of exporting.After this, can not describe specially again.

In the present invention, we also contemplate such fact: the spatial frequency of light spread function will be significantly less than the spatial frequency of pixel resolution.This is especially remarkable when side-light type LCD.The quantity of the LED in side-light type LCD can be more much lower than directly-down type LCD.The light spread function of an example as shown in Figure 2.This means that an interpolating function can be similar to the real light spread function value of the pel array between a small amount of sampled point.Fig. 3 shows the example of sampled point, and they are arranged in a rectangular node.Can measure and/or simulate or be extracted by similar mode the S of four sampled points ₁₁, S ₁₂, S ₂₁and S ₂₂light spread function value Ak_S ₁₁, Ak_S ₁₂, Ak_S ₂₁, Ak_S ₂₂.K represents a kth LED.Light spread function that is measured by photometer (video camera) or simulation can have the resolution more much higher than the resolution of sampled point.High-resolution smooth spread function to be declined/tapers to the low resolution of sampled point by using mean value, maximal value, histogram value or other filter function.The light spread function value of predetermined sampled point resolution like this can be stored in advance in a look-up table.

For each sampled point distributes a cell (31), this cell can be called as pixel compensation cell, for describing the compensation process of the pixel in scope of the present invention.These cells can be identical with the cell of image reduction disclosed in US8421741B2, this means identical size and position.But operation and the image reduction of pixel compensation are contrary, namely from low sampled point resolution extension to high pixel resolution.At these four sampled point (S ₁₁, S ₁₂, S ₂₁, S ₂₂) between the light spread function value of pixel of rectangular array can be calculated by interpolating function.A simple function can be a bilinear function.Bicubic, polynomial function, splines or other known interpolating functions can be employed and combination with one another.Interpolating function can be level and smooth to consider the fact that light spread function is also normally level and smooth.One is that sampled point that in the horizontal and vertical directions four are adjacent is (just as S in figure 3 for the embodiment that of sampled point of effectively processing is favourable ₁₁, S ₁₂, S ₂₁and S ₂₂) build a rectangle.

In this way, as long as the light spread function value of a small amount of sampled point of such as four just can represent the light spread function value of the array with many pixels.Therefore, light spread function value is stored with the low resolution of sampled point just much of that.The light sent due to a LED being placed on the frame of LCD can scatter a big chunk of cover plate, can cover the major part of whole panel or at least panel at the light spread function of the resolution of sampled point.But, do not need the light spread function value storing each pixel.Therefore, the storer needed for light spread function value the and such as number of times of the read-write of storer being greatly diminished.

Certainly, may there is deviation in the interpolate value of interpolating pixel and the value of reality.This deviation may be such as 10%.Because interpolating function can be level and smooth and light spread function can also be smooth distribution, therefore, this deviation may not by the perception of human eye institute.The density of sampled point is adjustable, and therefore, this deviation can be limited in the reasonable scope.The spatial variations of the light spread function near LED light source is stronger compared with remote area.Fig. 4 shows an example of the LCD with the LED being placed on a frame (being the right in this case) place.LED component by symbol " " characterize.The density of the sampled point under LED frame can other parts (such as middle section) of ratio panels more be increased.It must cover the spatial frequency of all or nearly all light spread function.Such deviation of the interpolation between sampled point that can make is low, and lower than certain limit value (such as 10%).Fig. 2 describes the light spread function of a LED.Can find out, close to the spatial frequency of light source (LED component) higher than remote area.The density of sampled point usually can higher than as shown in Figure 4.Low sampling point density in Fig. 4 considers for graphic sharpness.

Certainly, the sampled point on panel can build a uniform grid, as long as the distance between two adjacent sampled points is associated with the high spatial frequency of light spread function.The advantage of uniform grid can be simple process.Such as, for the HDTV of side-light type, uniform rectangular node can contain 32x32 pixel.Storer carrys out reduced size with the factor of 1024.HDTV has the sampled point slightly above 2000.Compare non-uniform grid, light spread function needs this shortcoming of larger storer may be not serious, because the chip cost of such storer of such as SRAM and so on is low in modern semiconductor integrated circuit technology.Much lower than the resolution of the pixel of display of the resolution of sampled point.In any case but the quantity of sampled point may remain the quantity being much higher than LED, because sampled point will consider the spatial frequency of light spread function.

By the help of sampled point, the value of the light spread function of many pixels can be obtained by calculating interpolating function, instead of by a huge storer.This just requires that calculating should be very efficient, because the number of pixel is very high.For an interpolation, for display two sampled points calculated with number of pixels between distance be often used as denominator.An example of a bilinear interpolation is as shown in Equation 3:

$B(x,y)=\frac{1}{(x_2-x_1)\·(y_2-y_1)}\·\left[\begin{array}{c}B\_S_{11}\·(x_2-x)\·(y_2-y)+B\_S_{21}\·(x-x_1)\·(y_2-y)\\ +B\_S_{12}\·(x_2-x)\·(y-y_1)+B\_S_{22}\·(x-x_1)\·(y-y_1)\end{array}\right]$

Formula 3

The coordinate of four sampled points of rectangular array is: (x ₁, y ₁); (x ₁, y ₂); (x ₂, y ₁); (x ₂, y ₂).B_S ₁₁, B_S ₁₂, B_S ₂₁and B_S ₂₂it is the functional value in these four sample point.Such array can be called interleaved arrays.Can calculate according to formula 3 at the pixel value at coordinate (x, the y) place being arranged in this interleaved arrays.If denominator is bigit (such as 8,16 or 32), then a divide operations may be easy to perform.Therefore, if sampled point is the rectangle interleaved arrays of bigit around two edges, may be favourable.These two edges can have different values.This can be associated with the spatial frequency in horizontal or vertical direction.

The size of pixel compensation cell can be a bigit for interpolation easily and effectively.But, consider the spatial frequency of light spread function, the multiple of the resolution of display not always rational bigit.Such as, the line number of a HDTV is 1080, and it can not be divided by by 64,32 or 16.The spacing 8 of two sampled points may require too high or unnecessary high HW cost, the processor of such as large memories or complexity.For such situation, the resolution of display can by virtual extension.Such as, line number virtually can be extended to 1088 from 1080.In addition, may need to treat especially at the pixel compensation cell of the border of display and corner and interleaved arrays.

In Figure 5, the sampled point of corner regions is shown.Sampled point V ₁₁, V ₁₂, V ₁₃, V ₁₄, V ₂₁, V ₃₁and V ₄₁for the virtual increase when the number of row and column is not the multiple of the size of the reduction cell of pixel.Therefore, display panel has extended virtually, makes these samples in two virtual frames 51 and 52.Virtual sampled point in panel and the distance between actual samples point may be rational bigits.This allows the extrapolation array to may be less than interpolation array such as 53 also to have an interpolation easily and effectively.The light spread function value of these virtual sampled points can be pushed off and be stored in advance in LSF look-up table.Light spread function due to side-light type LCD can be level and smooth, and described deduction can be unadventurous, and can carry out in advance and be stored in advance in LSF storer.

Most of LCD only has LED on a frame, and it is also like this for comprising many TVs.In some LCD, described LED is placed on two relative frame places.Iff LED being placed on a frame place and resolution in the vertical direction of this frame is not that (such as LED is on a long margin frame of HDTV for the multiple of rational bigit, its line number is 1080), it may be favourable that the opposite border side of LED realizes virtual extended, because the light spread function at this opposite side place is more level and smooth compared with the border side at LED place.If LED is placed on two frame places and resolution in the vertical direction of this frame is not the multiple (on two long margin frames of HDTV, its line number is 1080 to such as LED) of rational bigit, virtual extended can be added to both sides.If LED is placed on a frame or two frame places and resolution in the parallel direction of this frame is not that (such as LED is on one or two short frame of HDTV for the multiple of rational bigit, its line number is 1080), virtual extended can be added to both sides.

Resolution for the treatment of display is not that the other method of the situation of the multiple of a rational bigit is illustrated in figure 6.In this example, the size of a pixel compensation cell can be 32 × 32, with due regard to the spatial frequency of light spread function.For the LCD with 1080 row, it is capable that panel can comprise 36 sampled points, and the distance wherein between two adjacent row is 32.Distance between second minimum and the 3rd minimum row is 16, and the distance between minimum row and the second lowermost row is 8.Generally speaking accurately cover 1080 pixel columns.If consider the high spatial frequency in the frame area of these LED, higher sampling point density may be favourable close to the frame of one or two LED.

For the compensation of pixel, the backlight illumination after each pixel is necessary, as shown in Equation 2.A kind of method calculating backlight illumination first uses interpolating function to calculate LED(Ak _ij) light spread function value, then can calculate backlight illumination according to formula 1.A more effective method first calculates at sampled point S according to formula 1 _ijthe backlight illumination at place.Formula 1 can convert to:

$B\_S_{\mathrm{ij}}=\underset{k=1}{\overset{m}{\mathrm{\Σ}}}\mathrm{Ak}\_S_{\mathrm{ij}}\·L_k$

Formula 1 '

The light spread function value (Ak_S of these sampled points _ij) can read from a look-up table or multiple look-up table.By the contribution (Ak_S of each LED _ijl _k) sue for peace to produce the whole backlight illumination (B_S after a sampled point _ij).Backlight illumination (B_S after all sampled points _ij) the actual image forming the backlight illumination of a reduction.Backlight illumination after pixel in array between sampled point can be interpolated.The result of interpolation is the backlight illumination of pixel resolution.Like this, the labour intensity calculating the backlight illumination of pixel resolution can be lower, because interpolating function performs with only needing next pixel.Distance between sampled point should keep bigit.

In order to higher energy-conservation, the output valve of the cell of a reduction can lower than the maximal value of the grey scale pixel value in this cell.The signal of high spatial frequency may not be retained completely.So, cut out and may occur.Cut out and mean, if backlight illumination is after one pixel less than a grey scale pixel value, then the transmission value of pixel is saturated with maximal value.It is more general formula that formula 2 can be converted into following:

$t_{\mathrm{ij}}=\min (\frac{G_{\mathrm{ij}}}{B_{\mathrm{ij}}},1)\·t_{\mathrm{MAX}}$

Formula 2 '

B _ijbe normalized into and grey scale pixel value (G _ij) identical color range/unit.T _mAXbe maximum or saturated transmission value and can 100% be standardized as.For the linear gradation color range of 12, t _mAXcan be 4095.For simplicity, t _mAXcan 1(100% be set to), after this disappear in this mathematical formulae for the calculating of pixel transmission value.

Fig. 7 shows one to adjust the dimmed computation process in local that grey scale pixel value/transmission value attaches most importance to.Based on the feature of input image data and LCD, LED value can be determined.High image resolution is reduced by image reduction step (71).The pel array forming a rectangle can be reduced into a cell.The reduction image of low resolution can be used as the input of LED optimizer (72).The output of LED optimizer (72) is LED value, can be used as the control signal of LED driver part.The LED value of a few frame can be stored in LED register (73).Can use the light spread function value of low resolution, such as stored in box 74 value, determines LED value.Determination due to LED value is not invention which is intended to be protected, therefore, is not explained in detail in this article.Determine that the method for LED value is open in US8421741B2 and US8411116B2 for one.

How of the present invention focusing on adjusts grey scale pixel value in the backlight situation that local is dimmed.In order to this object, the light spread function value of the reality reaching certain precision may be needed.The light spread function value of side-light type LCD can be determined by measurement, simulation or the combination of both.Like this, actual physical behavio(u)r just accurately can be described.Light spread function be LCD physical attribute and independent of image.Therefore, can pre-determine it.If the distribution of the spatial frequency low light in other words spread function of the light spread function of LCD is level and smooth, once the value of the look-up table of predetermined smooth spread function may be used for other LCD of same model, therefore, the look-up table of the predetermined smooth spread function of LCD model can be provided for, and not need the concrete light spread function measuring each panel.Because the distribution of light spread function is level and smooth, fluctuation caused in process of production and deviation may not by the perception of human eye institute.Can certainly measure the concrete light spread function of each panel, its shortcoming is that production cost may uprise.Subsequently, the high-resolution smooth spread function of predetermined LCD can be tapered to sampled point is the low resolution of benchmark.As disclosed in the present invention, the resolution of light spread function can take sampled point as benchmark.Such resolution is well below the resolution of display, because the spatial frequency of light spread function is well below by the given spatial frequency of the resolution of display.At light spread function value (the such as Ak_S at sampled point resolution place _ij) can be determined in advance and be stored in advance in look-up table (frame 74).

The LED value L provided by frame 73 _kwith Ak_S _ijbe multiplied and give a kth LED to sampled point S _ijthe contribution of backlight afterwards.Sampled point S _ijwhole backlight illumination (B_S afterwards _ij) be all LED or be the contribution sum of all LED substantially.This describes in formula 1 ', and performs in frame 75.In this way, the backlight illumination at certain sampled point of the LED value given by frame 72 or 73 can calculate.Described distribution is level and smooth, because described smooth spread function is level and smooth.Resolution due to sampled point is low, therefore, limits the calculating labour intensity for frame 75.

In order to the adjustment of the transmission value of each pixel/sub-pixel, need the backlight illumination after each pixel.This is provided by frame 76.It carries out interpolation between the backlight illumination of a small amount of sampled point provided by frame 75.Exemplary method illustrates in equation 3.The output of frame 76 is B _ij.Frame 77 performs the transmission value that formula 2 or more general formula 2 ' calculate pixel.Formula 2 ' considers that transmission value has saturated and maximal value.Described input is grey scale pixel value G _ij(view data) and the backlight illumination (B after pixel provided by frame 76 _ij).Described output is the transmission value (t through adjustment of pixel ij _ij).Can be sent to timing controller, TFT driving or the similar electronic control unit being used for display panel.Whole pixel compensation process is included in the frame 70 of Fig. 7.This method allows high-level efficiency and adjusts the grey scale pixel value of the dimmed LCD in backlight local at low cost.

The output of LED optimizer (72) is LED value, and it is stored in LED register (73), and for control LED.The LED value stored can use simultaneously in pixel compensation, and the transmission value of each pixel is adjusted.As previously mentioned, inverse gamma will be applied to the transmission value of a pixel.The output of inverse gamma function is used by timing controller and/or TFT driving element.There is for most of timing controller of current LCD module and TFT driving element the gray level of 8.Because local is dimmed, backlight can be uneven.The driving element of the TFT of 8 may cause false profile, is particularly strongly dimming the smooth area of dark at backlight place.Experiment shows, the gray scale of 10 can avoid contouring artifacts.The well-known method generating the extra gray scale of two positions controls based on FRC(frame rate) and error diffusion.A kind of widely used error diffusion algorithm is Freud-Staenberg shake.The application of FRC and error diffusion can demonstrate seamlessly transitting in image when avoiding contouring artifacts.When TFT driving element has gray scale resolution (true 10) of one 10, FRC and error diffusion may be unnecessary.When the TFT driving element of use 6 gray scales, FRC and error diffusion are enforceable.Otherwise, often may there is the contouring artifacts that can see.

In order to provide the gray scale color range resolution of 10 gamma codings, the precision for the calculating according to formula 1 ', 2 ' and 3 may be needed higher.It can be the linear position (bit) of such as 12 or more, as 16.View data (being generally the bits of coded of 8 gammas), can be transformed into a higher linear resolution, such as 16.The low-down gray-scale value of common gamma function can be distinguished in 16 linear positions.This LED driver part has the resolution of typical 8 linear positions, and therefore, the value of LED can have the resolution of 8 linear positions.Light spread function can be 12 linear positions, and described linear position is based on measurement.Most of image photometer (camera) has 12 linear positions.For low-down smooth spread function value, the resolution of 12 still can provide a rational precision.8 LED values and 12 light spread functions can cause 20 backlights according to formula 1 '.Backlight illumination can be that to be truncated be 16.The transmission value calculated can be 16 linear positions and be converted into 10 gamma bits of coded.

In much equipment, the gray-scale value of image (such as pixel-by-pixel basis pixel) is transfused in a pipeline fashion.When the whole view data of a frame is known (this means to be transfused to) substantially, the value of LED can be determined.The direct execution of formula 1 and formula 2 ' may need a frame buffer memory, to store whole or whole view data substantially.Such frame buffer memory and the access of such as reading and writing this frame buffer memory may cause high cost and consume remarkable power.In the computation process of Fig. 7, do not need frame buffer memory.For the calculating of backlight illumination, can use former frame or or even the LED value of the first two frame, described value is stored in register (73).Have how many frames to be staggered, this depends on the processing time of LED optimizer (72).Usually, LED value at same frame or should provide during next frame. so the LED value of pixel compensation extracts from last frame or the second last frame.Therefore, the LED value of the gray-scale value of the pixel of current view data and former frame or multiframe can be used for calculating the backlight illumination after each pixel by the compensation of pixel.The register/memory storing LED value is needed to be negligible, because the quantity of independent controlled LED is low-down.The backlight illumination of the sampled point resolution that frame 75 provides is by using the LED value (73) that stores and calculating from the light spread function value of the look-up table be stored in advance in LSF storer (74).The output valve of this frame (75) carrys out interpolation by backlight interpolater (76) with the resolution of pixel.Backlight illumination only depends on the LED value (73) of storage, and can streamline formal layout, makes backlight interpolater (76) provide the backlight illumination of pixel resolution with streamline form.It can be synchronized to vsyn and the hsyn signal of view data.This means, if when receiving the vsyn signal of this frame or before this, the backlight illumination of the first pixel of frame has been available.If when receiving capable hsyn signal or before this, the backlight illumination of the first pixel of this row has been available.Like this, the transmission value counter (77) performing formula 2 ' can transmit the transmission value of the pixel after adjustment with streamline form.From input channel to output channel, only there is the low latency of such as 30 clocks be used for performed by the transmission value calculation block (77) of formula 2 '.For the pixel clock frequency of a HDTV resolution and 148.5MHz, this is only the delay of about 200 nanoseconds.

The control signal of LED and LCD can by synchronously.Such as, for control LCD the pixel value (t that adjusts out _ij) be synchronized to be used to calculate t _ijlED value.In addition, the method for such as scanning backlight and so on capable of being combined.

Above description discloses the method for the grey scale pixel value of the dimmed process in a kind of local for adjusting LCD.The backlight illumination on pixel benchmark is obtained by the interpolation between multiple sampled point.Backlight illumination on sampled point benchmark considers the substantial contribution of all LED.For computing application light spread function value and the LED value of the contribution of a LED of a sampled point, wherein the predetermined density of sampled point is associated with the spatial frequency of light spread function.Sampling point density is the pixel resolution being significantly less than LCD, and the quantity of sampled point is considerably higher than the quantity of independent controllable LED.Therefore, the method can the accurate and high-resolution data of high efficiency process LCD display.

As shown in the frame 71 of Fig. 7, for image reduction, maximum gradation value can be the output valve of the cell of reduction.For the image of reduction like this, optimizer can provide one without the scheme sheared.This means to be equal to or higher than the given brightness of the gray-scale value of this pixel at the backlight illumination at each pixel place.If the compensation of pixel is correct execution, then each details of image can be retained and the brightness of whole image can be preserved.The method of the pixel compensation of the interpolating function of the backlight illumination between to have for sampled point disclosed in Fig. 7 is just much of that.

As previously mentioned, in the dimmed backlight situation of high energy saving, cut out and may occur.Cut out and also may occur, because the LED value of former frame can be used for the pixel compensation of present frame.If this LED value can by time filtering with suppress the brightness of LED change (especially from low to high) very unexpected time the appreciable scintillation artifacts that may produce.Therefore, in the new embodiment in a dimmed process in local, as shown in Figure 8, a termporal filter (831) is added and is that optimizer (82) arranges further restriction, limits with the scope of the LED value to a new frame.The previous LED value stored on LED register (83) can be used by termporal filter (831).These situations may cause LED value for too low current image.If present frame is brighter than past frame too many, may occur cutting out.

Cutting out always can perception or visible.But cutting out the artifact perceived may reduce picture quality.The neighbor with different gray-scale values may lose their difference each other.Cloudlike the details of profile and so on may lose.This may obstruct the dimmed power savings advantages brought in local.Cut out can be in visible region with slightly dim spot for cost preserves image detail, also can produce an image.Because display is a kind of massaging device, the picture quality of this image has the picture quality of the image cutting out artifact higher than one.When fractional energy savings remains unchanged, this can obtain trading off between energy-conservation and image detail.

Theory of the present invention is the virtual backlight illumination after may there is appreciable region of cutting out that is increased in.Formula 2 ' is converted into:

$t_{\mathrm{ij}}=\min (\frac{G_{\mathrm{ij}}}{{\mathrm{GSF}}_{\mathrm{ij}}\·B_{\mathrm{ij}}},1)=\min (\frac{G_{\mathrm{ij}}}{B\_V_{\mathrm{ij}}},1)$

Formula 4

CSF _ijbe called as and cut out inhibiting factor, it adds backlight illumination virtually.This means that LED value and backlight illumination are constant.If do not cut out in the neighborhood, CSF _ijcan be 1.Do not need the virtual increase of backlight illumination.If cut out in adjacent region, this means that real backlight illumination is less than grey scale pixel value, then higher than 1 CSF _ijcan suppress to cut out.This means, at raw image data G _ijin difference can by the pixel transmission value t that adjusts according to formula 4 _ijdifference reflect.The grey scale pixel value of the adjustment represented by formula 2 or formula 2 ' is in fact tailored inhibiting factor CSF now _ijbe divided by.

But the simple method of cutting out inhibiting factor as applied constant on whole panel may mean that whole brightness of image reduces.Ratio between brightness of image and power consumption will become grain.This and lower fractional energy savings are actual identical.Owing to cutting out mostly only on one or a small amount of point, what local was different cut out inhibiting factor can avoid the saturated of pixel transmission value.This means, suppress to cut out, and total brightness of image can be saved substantially, trading off between energy-conservation and image detail can be obtained.The highest local dependence is if the inhibiting factor of cutting out for the virtual increase of backlight illumination is independent for each pixel.Therefore, CSF _ijbe used as the symbol cutting out inhibiting factor, wherein i and j represents the index of the row and column of this pixel.

Because each pixel needs the CSF of oneself _ij, the efficiency of algorithm is again a major challenge.In the present invention, we make use of the theory of sampled point and interpolation again, thus make computational complexity fall the end in rational scope.

Cut out the pixel compensation of inhibit feature for having local, the first step be to determine sampled point (m, n) cut out inhibiting factor CSF_S _mn.The sampled point cutting out inhibiting factor can be identical with the sampled point of pixel compensation, and this can make algorithm more simple.In fig. 8, the embodiment of emphasis in the dimmed process in the local of cutting out inhibit feature is shown.Image reduction frame (81) extracts the characteristic information of the view data for LED optimizer and pixel compensation.The maximal value of pixel compensation cell can (81) have been determined and has been stored during image reduction step.Such signature grey scale value can be stored in such as GPC-RAM2(813) storer in.Can determine according to formula 1 ' at the backlight illumination of a sampled point.The cutting out inhibiting factor and can be of sampled point (m, n)

CSF_S _mn=min(G _MAX/B_S _mn,1)

Formula 5

B_S _mnit is the backlight illumination at sampled point (m, n) place.G _mAXit can be the maximal value of respective pixel compensating unit lattice.Cutting out inhibiting factor needs at once for the pipeline processes of pixel compensation, this means each pixel comprising the first pixel comprised in a frame.In order to avoid the frame buffer-stored of costliness, the G of former frame _mAXvalue can be used for formula 5.Therefore, two storeies may be needed for storing the signature grey scale value of two frame pixel compensation cells.In fig. 8, they are GPC-RAM1 (812) and GPC-RAM2 (813).Due to relatively low sampling resolution, the size of these two kinds of storeies is less.They may be implemented as SRAM.At GPC-RAM1(812) in value (they obtain during the input of former frame) be read in the pixel compensation process for present frame, and the signature grey scale value of present frame is written to GPC-RAM2(813).For next frame, the function of these two storeies is exchanged, and storer 813 is read and storer 812 is written into.If at the backlight illumination of a sample point lower than the maximum gradation value of pixel compensation cell, cut out and may occur.Cut out inhibiting factor more than 1.But cutting out always can perception or visible.Allow some pixels to be tailored (and these pixels cut out not perceived or do not found out) and the bright of shown image can be kept.If allow cutting out of the pixel of some to a certain degree at the image reduction frame of Fig. 8 and LED optimizer (81,82) in appointment LED value, then can reach higher fractional energy savings.

The method having some common arranges a definite limitation to visible and/or irritating cutting out.One is histogram.This means that a certain proportion of pixel in a region can be tailored.So, the gray-scale value of some the brightest pixel (as the brightest 1%) can be used to replace the maximal value of pixel compensation cell.This value is called as G _hIST.G _hISTmay be used for formula 5 to replace G _mAX.Which results in new formula:

CSF_S _mn=min(G _HIST/B_S _mn,1)

Formula 6

Inhibiting factor of cutting out according to formula 6 can lower than the factor according to formula 5.This means that the transmission value of the adjustment of the pixel of LCD can be higher, and cut out can not be perceived and find out.Cut out the whether perceived spatial frequency also depending on picture material.The eyes of the mankind are to comparatively low spatial frequency ratio is more responsive to higher-spatial frequencies.Therefore, the high-frequency signal of this pixel compensation cell can be restrained.The nonlinear filter function with low-pass characteristic can be used for determining unit lattice G _fREthe signature grey scale value of frequency dependent.Simple example as:

$G_{\mathrm{FRE}}=G_{\mathrm{MEAN}}+\frac{G_{\mathrm{MAX}}-G_{\mathrm{MEAN}}}{2}$

Formula 7

G _mEANit can be the average gray value of pixel compensation cell.G _fREthe G in formula 5 can be replaced _mAXor the G in formula 6 _hIST.Must be noted that for using the LCD of White LED as backlight, the gray-scale value of pixel can be the maximal value of RGB sub-pixel value.This maximum sub-pixel value can be used on average, histogram and frequency analysis.Formula 7 has two frequency components, namely given and by G by monitor resolution _mAXrepresented maximum frequency and by the size of pixel compensation cell given and by G _mEANthe frequency represented.In formula 7, the frequency range between these two frequencies is reproduced as 50%.Can use in formula 7 be different from 50% Summing Factor/or more frequency component.And, the signal of lower frequency can be considered, this means the information that also can use adjacent cells lattice.Also can use or combine other analytical approach, as Sobel gradient.

Histogram information and frequency information can be combined with each other, to preserve the brightness of image and suppress can perception or visiblely to cut out.Generally speaking, can determine that the signature grey scale value of pixel compensation cell is to avoid visible cutting out.It comprises histogram, frequency and out of Memory, and a simple method uses G _hISTreplace the G in formula 7 _mAX, that is:

$G_{\mathrm{PC}}=G_{\mathrm{MEAN}}+\frac{G_{\mathrm{HIST}}-G_{\mathrm{MEAN}}}{2}$

Formula 8

This signature grey scale value can be used for limiting cutting out as seen in pixel compensation cell.If backlight or virtual backlight are no more than this value, a small amount of pixel can be saturated, but do not have to see or to perceive cut out artifact.The signature grey scale value of this frame can be extracted and be written in storer (813) during image reduction step.Compare with the value (it only needs a storer 811) of image reduction cell, the signature grey scale value of pixel compensation cell needs two storeies (812 and 813).The operation of these two storeies exchanges frame by frame.Due to sampled point CSF_S _mncut out the function that inhibiting factor can be the signature grey scale value of one or more cells around it, the inhibiting factor of cutting out of sampled point can be the described histogram of one or more cells around and the function of frequency information:

CSF_S _mn=min(G _PC/B_S _mn,1)

Formula 9

At the virtual backlight illumination of this sampled point be:

BV_S _mn=CSF_S _mn·B_S _mn

Formula 10

Cutting out inhibiting factor and separating in the resolution punishment of sampled point according to the virtual backlight illumination of formula 10 according to formula 9.All sampled points cut out inhibiting factor (CSF_S _mn) a composition matrix.What this was not processed cut out inhibiting factor matrix is called as CSF1, described in cut out inhibiting factor matrix and will be further processed in follow-up content.A possible problem is, cuts out inhibiting factor and may accompany a sampled point and change consumingly to next sampled point.Such as, one in a pixel compensation cell very bright point may cause high cutting out inhibiting factor, and at the sampled point of adjacent cell, does not have that to cut out or cut out inhibiting factor be 1.Interpolation can be applied to level and smooth CSF1 distribution.But, around the sampled point with high shear inhibiting factor, likely there is visible shadow.In Fig. 9 a, show the stingy figure of original image.Show the stingy figure of the correspondence by cutting out the image after inhibiting factor process in figure 9b.Lamp in image is very bright and cause and cut out.Cut out and be visible without the transition between cut out areas, because the transition of cutting out inhibiting factor is steep.In figure 9b, the shade along the profile of cabin is consequent.If this may occur when pixel compensation cell is little.A large pixel compensation cell can avoid this problem.But a large pixel compensation cell significantly may reduce the brightness of whole image.It may counteract localized approach of cutting out suppression of the present invention.In addition, the histogram of large pixel compensation cell and frequency information may be too coarse.

In order to overcome cut out inhibiting factor sharply change the problem that may bring, take sampled point as cutting out inhibiting factor and can being filtered of resolution.Figure 10 a illustrates the example distribution of cutting out inhibiting factor at the resolution place at sampled point according to formula 9.Figure 10 b illustrates the distribution of cutting out inhibiting factor after filtering of example.The distribution of cutting out inhibiting factor becomes level and smooth.Like this, cutting out as seen can be suppressed, can avoid the pseudomorphism of the darkened boxes such as in a part for image and so on simultaneously.In Fig. 9 c, show with by the stingy figure cutting out the image handled by inhibiting factor of spatial filtering.Comparison diagram 9a and 9b, had not only seen and has cut out but also do not see shade.The brightness of image of the entirety of image is saved substantially.

Spatial filter function as box filtering, Gaussian smoothing filter etc. can use for this reason.For an effective mathematical operation, spatial resolution is still in the resolution of sampled point.For pixel compensation, virtual backlight illumination can decompose by pixel resolution level.Interpolation technique can be used for the backlight illumination calculating virtual pixel resolution level from the value of sampled point.Widely used bilinear function (formula 3) can be employed.

When the resolution of display is not the multiple of pixel compensation cell, the external pixels compensating unit lattice (such as C11, C12, C21) for pixel compensation are less than interior pixels compensating unit lattice as shown in Figure 5.They may be 1/4th or half of the size of internal element lattice.When external unit lattice do not comprise real pixel as C11 or C21, the value of the respective element in Matrix C SF1 can be set to 1.When outside cell contains real pixel as C12, can determine the signature grey scale value of this cell according to formula 8, the wherein different factor/coefficients or weight can be used for calculating.Because beholder focuses one's attention at the center of display usually, signature grey scale value can be less than the signature grey scale value of the internal element lattice with similar content.Similar scheme can be used for the second external unit lattice as C22 that still may comprise virtual pixel.

Figure 8 illustrates the process comprising the pixel compensation cutting out suppression.The information of the cell of the reduction of a function of the such as maximal value of present frame, mean value, frequency signal, histogram value or these values etc. and so on is stored in the image-RAM frame of RAM(reduction, and 811) in.The image with this reduction of low resolution can allow the operation of optimizer (82) iteratively carry out, and without the HW of high cost.This optimization core uses light spread function value possibly.One may be also by the light spread function value be stored in advance in LSF-storer (84).The gray-scale value of the cell (811) of a reduction can be less than the signature grey scale value of pixel compensation cell (813), although these two cells may be identical (position, sizes).Suitable optimized algorithm (82), as disclosed in US8421741B2 and 8411116B2, directly can provide LED value and indirectly provide corresponding backlight illumination.Described LED value and backlight illumination have met or exceeded the gray-scale value of reduction cell as the image of the reduction of pixel.If image reduction (81) and LED optimizer (82) provide a foregoing LED value without cutting out, then do not need the function of cutting out suppression, or to cut out inhibiting factor be all 1.But, like this for the power consumption of the enough good picture quality of generation one may be height too.For the value of the cell of reduction lower than the signature grey scale value of pixel compensation cell, the LED value that optimizer (82) provides summation (power consumption) lower than described situation above.Simple example is the gray-scale value of image reduction cell is mean value, and the signature grey scale value of pixel compensation cell is maximal value.Due to the gray-scale value that image reduction cell is lower, the power consumption for backlight is lower.Adopt local to cut out inhibiting factor by application and visiblely cut out artifact to suppress possible, and the overall brightness of image can be saved the power saving simultaneously reaching height substantially.

Light spread function for backlight sampling can have the resolution identical with the image of reduction.In this case, only there is a LSF storer (84) that LED optimizer (82) and backlight sampling (85) can be used to.Otherwise, a storer can be added for storing the light spread function value the same with image reduction (81) cell resolution, input and LED optimizer (82).As mentioned above, LED optimizer (82) is not within invention which is intended to be protected.Be not described in detail in this.

The LED value that optimizer (82) provides is stored in LED register (83).The LED value of the LED-backlit for driving present frame determined for former frame or former frame is used to calculate backlight illumination (85) in sample point according to formula 1 ', wherein can use the light spread function value be stored in advance in LSF-storer (84).The signature grey scale value being stored in the pixel compensation cell of the former frame in GPC-RAM1 (812) can be used for performing formula 9, together with the backlight illumination (outputs of 85 frames) in sampled point benchmark place, thus obtain sampled point resolution in cut out inhibiting factor Matrix C SF1.This performs in frame 851.Can come from former frame or former frame due to LED value and can use before the view data of the new frame of input, this part of cutting out inhibiting factor Matrix C SF1 may be calculated before the view data of the new frame of reception one.CSF1 obtains GPC-RAM1 value based on the LED value of former frame or former frame and former frame.The view data of former frame or former frame usually can be closely similar with new frame, so there will not be the problem that can see.

When there is scene and switching, may there is stronger deficiency in the backlight of a small amount of frame.By a scene change detection module (not belonging to scope of the present invention), faster than specified by the termporal filter for continuous scene of the increase of LED value can be allowed.This can reduce the quantity of the frame with strong backlight deficiency.When cutting out inhibiting factor and being very high, this factor can be limited in below certain rational value.The meaning of rational value refers to that it can avoid image to become too dark.The example of a rational value is 2.If this situation also may occur, because this means that the gray-scale value of image reduction cell is more much lower than the signature grey scale value of pixel compensation cell when image reduction (frame 81) performs too fiery.Another kind of situation is just now said, switches occur after the image that very dark image shift becomes clear very much in scene.For fast-changing image, as scene switch when, possibly cannot a small amount of frame of perception cut out artifact.

Cut out the backlight illumination that inhibiting factor Matrix C SF1 is multiplied by each sample point, create the backlight distribution that as shown in Equation 10 one is virtual.This by virtual backlight sample (852) come.This distribution can by spatial filter function, such as cassette filter or Gaussian filter, processes (the virtual backlight wave filter 853 of frame).Frequency filtering can higher than the given frequency of whole panel size/dimension, and therefore, virtual backlight remains local distribution.After this step, in the virtual backlight value of sample point by frame 86 interpolation.Output is the B_V of pixel benchmark _ij(virtual backlight distribution).Can be performing as in the Pixel-level of streamline in the calculating of the virtual backlight illumination of pixel benchmark place.Export (virtual backlight illumination B_V _ij) can be such as use vsyn and the hsyn signal of view data, synchronous with the gray-scale value of image pixel.In last transmission value calculation procedure (frame 87), new transmission value utilizes the grey scale pixel value of synchronous image and virtual backlight illumination to calculate according to formula 4.Whole pixel compensation process is included in the frame 80 of Fig. 8.

Performance in the frame 852 and 853 of Fig. 8 also can realize in a slightly different mode.Cut out the output of inhibiting factor Matrix C SF1(frame 851) can be processed by a spatial filter function.Be multiplied with backlight illumination (formula 10) through filtering/level and smooth inhibiting factor matrix of cutting out, produce the virtual backlight distribution of the resolution of sampled point.Output is also can be closely similar with the output of frame 853 at the virtual backlight illumination of sampled point resolution.Step is below identical with 87 with the step 86 in Fig. 8.These two kinds of methods can provide a similar picture quality and need similar calculating labour intensity.

Because great majority operation is here all carry out on sampled point benchmark, therefore restriction is obtained to the overall calculation labour intensity comprising the pixel compensation cutting out suppression.The interpolation that the execution of virtual backlight inserter (86) just can be able to be realized by simple HW, if the distance particularly between two sampled points is a bigit.Final step (frame 87) is divide operations.Due to view data often with streamline form as Pixel-level input, step 86 and 87 can be organized into a pipelined process.The HW complicacy of this step (frame 87) can be limited.

Cutting out suppressing method as above can in conjunction with additive method (as pixel compensation and/or the existing method determining backlight illumination).A kind of method for directly-down type LCD is, such as, uses a corresponding compensating factor to an image segments of being arranged by a LED.This compensating factor may be a function of this LED value.This function can be predetermined and be stored in a look-up table.Be multiplied the original pixels gray-scale value of this fragment with this compensating factor the grey scale pixel value after can being adjusted.For one higher or overdo energy-conservation, cut out and may occur.Suppress visible cutting out, compensating factor can be divided by by the inhibiting factor of cutting out of local.The business removed out like this produces a new compensating factor.Cut out inhibiting factor based on the signature grey scale value of image segments, as formula 8, and can to determine according to formula 9.Backlight illumination required for formula 9 can calculate according to formula 1 ' or according to existing method (such as point of application spread function) or estimate.This new compensating factor can be filtered and interpolation.After filtration this and/or level and smooth new compensating factor is used to adjust grey scale pixel value.As a result, the details of image can be preserved and the brightness of image also can be saved substantially.

The invention discloses a kind of method of pixel compensation accurately and effectively for the dimmed process in local.In addition, higher energy-conservation in order to allow, introduce and a kind ofly cut out inhibiting factor by application local and suppress visible method of cutting out artifact.Final result is the high image quality with the brightness substantially retaining image detail and image.It is compromise that it achieves between energy-conservation and image detail with reduction process cost.

Claims (24)

1. the method for being adjusted by the view data of the dimmed backlight in local to LCD, described LCD has the back light unit comprising multiple LED controlled separately, and described method comprises:

Pre-determine the light spread function value of a LCD model;

The resolution of the light spread function of described LCD model is tapered to the resolution of the sampled point of Two dimensional Distribution, and the light spread function value of the resolution of sampled point is stored in a lookup table;

The LED value of former frame or former frame is used for the calculating of the backlight illumination of sample point;

The contribution of this LED to the backlight of this sample point is calculated by the light spread function value and described LED value that are applied in a LED of a sample point;

By all or most LED to be calculated the backlight illumination of this sample point in the contribution summation of this sample point to backlight;

By carrying out interpolation to calculate the backlight illumination after other pixel between this multiple sampled point to the backlight illumination of multiple sample point;

The transmission value of each pixel is calculated based on the gray-scale value of pixel of image of input and the backlight illumination after this pixel; And

The transmission value of described pixel is supplied to the electronic control unit for LCD.

2. the method for claim 1, is characterized in that, the resolution of wherein said sampled point depends on the spatial frequency of described smooth spread function.

3. the method for claim 1, is characterized in that, can determine described smooth spread function value by measurement, simulation or the combination of both.

4. the method for claim 1, is characterized in that, wherein said LED value determines based on the view data of input and the feature of LCD, and the LED value of determined former frame or former frame is stored in register (73).

5. the method for claim 1, is characterized in that,

Wherein, described multiple LED is placed on one or more frames of LCD.

6. the method for claim 1, is characterized in that,

Wherein the predetermined density of sampled point is significantly less than the pixel resolution of LCD, and the quantity of sampled point is the quantity being much higher than described LED controlled separately.

7. the method for claim 1, is characterized in that, the distance between two wherein adjacent sampled points is the bigit of pixel.

8. method as claimed in claim 7, is characterized in that, the distance wherein between the sampled point of the Central District of border and panel is different.

9. the method for claim 1, is characterized in that, described method also comprises:

Virtual sampled point is added to the border of panel, wherein the size of external pixels compensating unit lattice is bigits.

10. the method for claim 1, is characterized in that, described method also comprises:

The grey scale pixel value of image described in pipeline processes, the calculating of the backlight illumination of the first pixel of one of them frame is synchronous with the vsyn signal of described frame, and the calculating of the backlight illumination of the first pixel of a row is synchronous with a hysn signal.

11. the method for claim 1, is characterized in that, described method also comprises frame rate and controls and error diffusion.

12. 1 kinds of methods that not only can retain image detail but also substantially can keep the brightness of image, described image is displayed to the LCD of the back light unit with the LED comprising multiple independent control, and described method comprises:

Based on the pixel of the surrounding of sampled point signature grey scale value and as calculated or the backlight illumination of this sampled point estimated calculate a sampled point cut out inhibiting factor;

By by as calculated or the backlight illumination estimated be multiplied insert a virtual backlight illumination with described inhibiting factor of cutting out;

From two-dimentional sampled point benchmark to virtual backlight illumination described in pixel benchmark interpolation;

The transmission value of each pixel is calculated based on the gray-scale value of pixel of image of input and the described virtual backlight illumination after this pixel;

The transmission value of described pixel is supplied to the electronic control unit for LCD.

13. methods as claimed in claim 12, is characterized in that,

Wherein said virtual backlight illumination in sampled point benchmark place by spatial filtering.

14. methods as claimed in claim 12, is characterized in that,

Wherein saidly cut out inhibiting factor in sampled point benchmark place by spatial filtering.

15. methods as claimed in claim 12, is characterized in that,

Described signature grey scale value is the maximal value of pixel compensation cell around sampled point, the function of the value of at least one in mean value, histogrammic gray-scale value and spatial frequency signal and be stored in a RAM(813) in.

16. methods as claimed in claim 12, is characterized in that,

Wherein said signature grey scale value represents a kind of to the prediction of cutting out the as seen restriction in pixel compensation cell.

17. methods as claimed in claim 12, is characterized in that,

The signature grey scale value of two continuous print frames has two RAM, and wherein the signature grey scale value of present frame is all written in a RAM, and the signature grey scale value of last frame reads from another RAM.

18. methods as claimed in claim 16, is characterized in that, also comprise:

Image reduction step is performed to generate the image of the reduction of low resolution to the high-resolution image of input, and described signature grey scale value is determined during described image reduction step and is stored.

19. methods as claimed in claim 18, is characterized in that,

The signature grey scale value of pixel compensation cell is the gray-scale value greater than or equal to the corresponding image reduction cell for determining LED value.

20. methods as claimed in claim 12, is characterized in that,

The backlight illumination of feature based gray-scale value and sampled point determines CSF1 matrix, and described backlight illumination is the light spread function value based on described LED value and described sampled point.

21. methods as claimed in claim 20, is characterized in that, if wherein do not cut out in the neighbouring area of a sampled point, the value of the respective element of described CSF1 matrix is 1.

22. methods as claimed in claim 20, is characterized in that, if cut out in the adjacent region of a sampled point, then the value of the respective element of described CSF1 matrix is greater than 1, and are limited in below 2 or other a rational value.

23. 1 kinds of systems for being adjusted by the view data of the dimmed backlight in local to LCD, described LCD has the back light unit comprising multiple LED controlled separately, and described system comprises:

For pre-determining the device of the light spread function value of a LCD model;

Resolution for the light spread function by described LCD model tapers to the resolution of the sampled point of Two dimensional Distribution, and the light spread function value of the resolution of sampled point is stored device in a lookup table;

For the LED value of former frame or former frame being used for the device of the calculating of the backlight illumination of sample point;

The device of this LED to the contribution of the backlight of this sample point is calculated for the light spread function value of the LED by being applied in a sample point and described LED value;

For the device by all or most LED to be calculated the backlight illumination of this sample point in the contribution summation of this sample point to backlight;

For the device by carrying out interpolation to calculate the backlight illumination after other pixel between this multiple sampled point to the backlight illumination of multiple sample point;

For calculating the device of the transmission value of each pixel based on the gray-scale value of pixel of image of input and the backlight illumination after this pixel; And

For the transmission value of described pixel being supplied to the device of the electronic control unit for LCD.

24. 1 kinds of systems that not only can retain image detail but also substantially can keep the brightness of image, described image is displayed to the LCD of the back light unit with the LED comprising multiple independent control, and described system comprises:

For the pixel of the surrounding based on sampled point signature grey scale value and as calculated or the backlight illumination of this sampled point estimated calculate the device cutting out inhibiting factor of a sampled point;

For by by as calculated or the backlight illumination estimated be multiplied insert the device of a virtual backlight illumination with described inhibiting factor of cutting out;

For from two-dimentional sampled point benchmark to the device of virtual backlight illumination described in pixel benchmark interpolation;

For calculating the device of the transmission value of each pixel based on the gray-scale value of pixel of image of input and the described virtual backlight illumination after this pixel;

For the transmission value of described pixel being supplied to the device of the electronic control unit for LCD.