(12) United States Patent
`Greene et al.
`
`(10) Patent N0.:
`45 Date of Patent:
`
`US 6,271,825 B1
`*Au . 7 2001
`a
`
`US006271825B1
`
`(54) CORRECTION METHODS FOR
`BRIGHTNESS IN ELECTRONIC DISPLAY
`
`5,206,633 * 4/1993 Zalph ................................... .. 345/92
`5,416,890 * 5/1995 Beretta ............................... .. 345/431
`5,555,035 * 9/1996 Mead et al.
`. 348/757
`
`
`
`
`
`
`
`(75) Inventors: Raymond G. Greene, Ovid; Robert H. gattylif?lstaé’lg? PeterHK¥u§$s€S0nS . - . -
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`, , sfangerk Z: 6,005,968 * 12/1999 Granger ............................. .. 382/162 azure e a ~~
`
`05 ’ O O
`
`aca’ a O
`
`(
`
`)
`
`6,020,868 * 2/2000 Greene et al. ....................... .. 345/88
`
`
`
`Asslgl'lee: _
`
`_
`
`
`
`Displays, IIlC., EndiCOtt, NY _ _
`
`
`
`(*) Nonce:
`
`Th1s_ patent lssufzd on a Con?rmed pros'
`ecut1on application ?led under 37 CFR
`1.53(d), and is subject to the tWenty year
`liggzntxzgerm Provisions of 35 USC-
`
`a
`
`.
`
`Subject to any disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C. 154(b) by 0 days.
`
`This patent is subject to a terminal dis
`claimem
`
`.
`(21) Appl. No.. 09/173,468
`
`(22) Filed:
`
`Oct. 14, 1998
`
`(63)
`
`(51)
`(52)
`
`(58)
`
`(56)
`
`Related US. Application Data
`
`Continuation of application No. 08/636,604, ?led on Apr.
`23, 1996.
`
`Int. Cl.7 ..................................................... .. G09G 5/10
`US. Cl. .......................... .. 345/147; 345/207; 345/89;
`345/88
`Field of Search .............................. .. 345/87, 103, 98,
`345/100, 147, 1_2, 3, 92, 903, 207, 89,
`88, 63, 431, 199, 348/383, 757, 687, 631,
`607, 609; 382/167, 162
`
`References Cited
`
`U.S. PATENT DOCUMENTS
`
`I. Gorog, “Displays for HDTVzDirect View CRTs and Pro
`
`jection Systems”, Proceedings of the IEEE, vol. 82, No. 4,
`pp 520_536, 1994*
`
`(List continued on next page.)
`
`iri'li‘tlrytifami'ler?charq lgllerpe
`Ssls an xammer fangs guyen
`74 Attorney, Agent, or Fzrm—SalZman & Levy
`(57)
`ABSTRACT
`
`The present invention features methods and apparatus for
`the correction of spatial non-uniformities in brightness that
`arise from materials, manufacturin , o erational and li ht
`g p
`g
`in
`arameter variations in electronic color, ?at- anel dis
`g P
`P
`plays. The methods apply both to gradual non-uniformities
`usually found in monolithic displays as Well as to abrupt
`variations present in displays composed of a multitude of
`tiles. Corrections are performed on the electronic drive
`signals used to control the brightness of selected display
`pixels. Parameters required for these corrections are
`acquired via brightness measurements over selected pixels
`and stored after suitable transformations. The stored param
`eters are then used to scale and/or interpolate drive signals
`in real time. Corrections are performed such that any
`remaining gradual and abrupt brightness non-uniformities
`fall beloW the detectable threshold under the intended vieW
`ing conditions. The correction methods can also be used for
`correcting brightness non-uniformities arising from uneven
`aging of the display. Apparatus for an automatic self
`calibrating function is also described.
`
`4,825,201 *
`
`4/1989 Watanabe et al. ..................... .. 345/1
`
`26 Claims, 6 Drawing Sheets
`
`@
`
`i
`Providing a tiled, Flat-panel diaplay having an array of pixcla
`each having a color space de?ned by color coordinatea
`
`l Determine the trtstirnulua valuea of eelected pixela l
`l
`I ?tore data repreaentative of these tri-etirnulua valuea l
`
`l
`
`Provide e’wred data to diaplay controller
`
`Provide an input data atrearn repreoenting an
`image to the diaplay controller
`
`Normalizing the input data etrcarn by applying the
`stored values to the input data
`
`Kecornputlng color coordinatea of the normalized data
`stream to produce dynamic, varying drive signals
`i
`Applying dynamic drive algnala to the dlaplay to produce
`an image having variationa in luminance below a visual
`perceptual level
`
`

`

`US 6,271,825 B1
`Page 2
`
`OTHER PUBLICATIONS
`
`H. Henck Van Leeuven et al., “A Digital Column Driver IC
`for AMLCDs”, Euro—Display, pp. 453—456, 1993.*
`H. Okada et al., “An 8—Bit Digital Data Driver for AML
`CDs”, Society for Information Display International Sym
`posium Digest of Technical Papers, vol. XXV, pp. 347—350,
`1994.*
`M. HijikiWa et al., “Future Prospects of Large Area Direct
`VieW LCDs”, Society for Information Display International
`Symposium Digest of Technical Papers, vol. XXVI, pp.
`147—149, 1995.*
`N. MaZurek et al., A 51—in Diagonal Tiled LCD VGA
`Monitor; SID International Symposium, Digest of Technical
`Papers, vol. 24, pp. 614—617, 1993.*
`
`D. Nickerson, “History of the Munsell System, Company
`and Foundation, 1—111”, Color Research Applications, vol.
`1, pp. 7—10, 69—77, 121—135, 1976.*
`
`S. Hecht, “The Visual Discrimination of Intensity and the
`Weber—Fechner LaW”, Journal of General Physiology, vol.
`7, p. 214, 1924*
`
`K. B. Benson editor, Television Engineering Handbook
`Featuring HDTV Systems, McGraW—Hill, 1992.*
`
`G. WysZecki et al., Color Science, 2”“ Edition Wiley, NeW
`York, 1982*
`
`* cited by examiner
`
`

`

`U.S. Patent
`
`Aug. 7, 2001
`
`Sheet 1 0f 6
`
`US 6,271,825 B1
`
`Tile Row [12
`
`[14
`
`[16
`
`QONEQEEB
`
`Position
`
`

`

`U.S. Patent
`
`Aug. 7, 2001
`
`Sheet 2 0f 6
`
`US 6,271,825 B1
`
`Tile Row [12
`
`[14
`I
`
`[16
`I
`
`13
`[18
`
`wt:
`
`0L
`
`2|
`
`6
`
`m 0
`m C
`U C
`C t
`F. 6 0 W
`t 1m
`
`O 5 ,
`
`QONEQSFEDJ
`
`Poaition
`
`

`

`U.S. Patent
`
`Aug. 7, 2001
`
`Sheet 3 0f 6
`
`US 6,271,825 B1
`
`52
`
`6:95:00
`@EEQ
`
`523 X
`
`@625
`@625
`
`26M
`
`;
`
`i
`
`26M
`52a]
`
`52b
`v f —\ v
`Column
`Column
`Drivoro
`Drivoro
`
`Column
`Drivoro
`
`LCD
`Array
`
`Column
`Drivoro
`
`Column
`Drivers
`
`Column
`Drivero
`
`l
`
`ll
`
`Luminanaco
`Scalar/Adder
`
`LR/LRI
`LG/LGr
`L5/L5I
`Memory
`
`Video Clock ?ync
`Input Input Input
`
`Go'
`
`4
`
`

`

`US. Patent
`
`Aug. 7, 2001
`
`Sheet 4 0f 6
`
`US 6,271,825 B1
`
`Q®
`
`
`
`\\~N+_Etgoo:;:E:BQ
`
`_:E369
`
`®®
`
`ion
`
`528
`
`30%:E3?203%
`
`2XJE¢~mucmz::34uocm:::34
`
`
` ‘mmgNM
`
`LufiomEmom
`
`uocmEEi
`
`Egan/x
`
`Lufiom
`
`:Eso\._o_Q®®®®
`:E30
`
`®®
`
`N®
`
`XQ®
`
`®®
`
`:Esfiu
`
`L925
`
`m.§r®©
`
`
`
`Lmuoomd5:300
`
`
`
`

`

`U.S. Patent
`
`Aug. 7, 2001
`
`Sheet 5 0f 6
`
`US 6,271,825 B1
`
`CQEQELA
`
`><—m0tion
`
`72
`
`76
`
`

`

`U.S. Patent
`
`Aug. 7, 2001
`
`Sheet 6 6f 6
`
`US 6,271,825 B1
`
`5TART
`
`V
`
`Providing a tiled, flat-panel dieplay having an array of pixele
`each having a color epace defined by color coordinatee
`
`Determine the tri-etimulue valuee of eelected pixele
`
`V
`
`Store data repreeentative of theee tri-etimulue valuee
`
`V
`
`Provide stored data to dieplay controller
`
`Provide an input data etream repreeenting an
`image to the dieplay controller
`
`Normalizing the input data etream by applying the
`etored valuee to the input data
`
`Pecomputing color coordinatee of the normalized data
`etream to produce dynamic, varying drive signals
`
`Applying dynamic drive signals to the dieplay to produce
`an image having variations in luminance below a vieual
`perceptual level
`J
`
`@W 7
`
`

`

`US 6,271,825 B1
`
`1
`CORRECTION METHODS FOR
`BRIGHTNESS IN ELECTRONIC DISPLAY
`
`This application is a continuation of Ser. No. 08/636,604
`Apr. 23, 1996
`
`FIELD OF THE INVENTION
`
`This invention pertains to the ?eld of electronic displays
`and, more particularly, to all electronic gray-tone and color
`displays, irrespective of their construction. The invention
`details a method for correcting spatial non-uniformities in
`the brightness of displays, Whether they be of monolithic
`construction or assembled from small tiles.
`
`BACKGROUND OF THE INVENTION
`
`Images on electronic displays are derived from a tWo
`dimensional array of pixels, each of Which represents one
`small element of the image. The resulting image is presented
`to the observer in a 1:1 siZe in direct-vieW displays, While
`projection displays magnify the image siZe, using an optical
`lens system. In black-and-White displays, each pixel displays
`one of tWo colors, black or White; in a gray-tone display,
`pixels can produce a speci?ed number of gray tones betWeen
`black and White. Since colors can be formed by combining
`primary colors, i.e., red (R), blue (B) and green (G) light, in
`speci?ed ratios, electronic color displays use primary-color
`elements in each pixel, in order to form a desired image via
`additive color mixing. In order to shoW still images, pixels
`can carry the same information all of the time; for moving
`images, the content of each pixel must be rede?ned peri
`odically. Full-motion images are usually required to be
`redraWn 30 to 75 times per second.
`Pixels can be accessed by using several techniques,
`including scan-, grid-, shift-, matrix- and direct-addressing.
`If, for example, the display carries an array of N><M pixels
`and it has to be redraWn n times each second, the data sent
`to each pixel must be provided in 1/(n*N*M) seconds and
`then held constant for (N*M—1)/(n*N*M) seconds, as other
`pixels are being de?ned. In the current American television
`(TV) standard (NTSC), each frame has about 250,000 pixels
`in a display area, With an aspect ratio of 4x3, Which are
`refreshed at the rate of 30 frames/second. One of the neW
`picture formats proposed for American high-de?nition tele
`vision (HDTV) to the Federal Communications Commission
`(FCC) has an aspect ratio of 16><9 and a refresh rate of 60
`frames/second. Pixels are arranged into 1280 horiZontal and
`720 vertical lines or, alternatively, 1920 horiZontal and 1080
`vertical lines (I. Gorog, “Displays for HDTV: Direct VieW
`CRTs and Projection Systems”, Proceedings of the IEEE,
`vol. 82, no. 4, pp. 520—536, 1994). The typical, loW
`resolution computer display (VGA) has 480 roWs of 640
`pixels, or, a total of 307,200 pixels at a refresh rate of 72
`frames/second.
`Electronic displays can be implemented by using a mul
`titude of different technologies, including, for example, the
`cathode-ray tube (CRT), electroluminescent displays
`(ELDs), light-emitting diode displays (LEDs) and liquid
`crystal displays (LCDs). While CRT displays have a depth
`comparable to the height of the screen, ELDs and LCDs
`belong to the class of ?at-panel displays (FPDs), the dimen
`sion of Which, in their direction perpendicular to the image
`plane, is much smaller than that of the CRT. With the CRT,
`either one (gray-tone) or three (color) electron beams scan
`along horiZontal lines in order to access each pixel. All color
`data is thus carried to the pixels via the electron beam
`current. FPDs (such as the LCDs) use matrix-addressing, in
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`2
`Which each pixel is accessed via roW and column lines. The
`pixel at the cross-point of a speci?c roW and column line can
`be selected via passive or active techniques. In the passive
`case, the non-linearity of the LCD pixel’s element is used for
`the selection. Active, matrix-addressed LCDs (AMLCDs),
`on the other hand, require a device (e.g., a transistor) for the
`selection of the pixel. In active matrix-addressing, a roW of
`pixels is usually selected at once by placing a speci?c
`voltage on the roW line (usually the gate line of the
`transistor). Pixel data is then made available via column
`lines to each of the pixels (usually the drain of the transistor)
`in the selected roW. An entire roW of pixels can be accessed
`in parallel in active matrix-addressing. Coupling betWeen
`pixels and roW and column lines is one of the disadvantages
`of matrix-addressing.
`The siZe of an electronic display is usually speci?ed by
`the length of the diagonal of the pixel array. Computer
`displays generally have siZes of betWeen 10“ and 21“; home
`television displays generally have siZes of betWeen 19“ and
`31“. Large public displays (e.g., used in sports arenas)
`generally feature siZes that range betWeen 200“ and 700“.
`The resolution of the image on an electronic display is
`determined by the pitch of the pixels, i.e., the smaller the
`pixel pitch, the ?ner the details that can be displayed.
`Typical computer displays have pixel pitches on the order of
`0.3 mm, and they can be vieWed from as close as 30 cm
`Without the human eye resolving the mosaic structure of the
`pixels. Large-screen, public displays have pixel pitches as
`large as 30 mm [see, e.g., Panasonic Astrovision, AZ-3000
`Series High-Density Fluorescent Displays, Panasonic
`Corporation, Japan, 1995]. VieWing distances of at least 10
`meters are required for such displays.
`A duty cycle is de?ned as the time spent for turning on
`individual pixels or a roW of pixels. With a CRT, each pixel
`is accessed individually and sequentially by sWeeping the
`electron beam. Thus, for example, in a VGA display With
`N><M=640><480 and n=72 HZ, the dWell time of the electron
`beam on each pixel is 46 ns. By de?nition this equals the
`duty cycle of this CRT. In an FPD-VGA display With the
`same frame rate, the dWell time is 640 times longer, or, 29
`us, due to parallel matrix-addressing.
`The brightness of an image on an electronic display is
`characteriZed by using the photometric quantity of lumi
`nance measured in candela per unit area (cd/m2=1 nit). The
`luminous ef?ciency is used to describe hoW much light the
`display produces per the amount of electrical energy pro
`vided to the display. LCDs operate With highly ef?cient
`backlights (such as ?uorescent lamps) With a luminous
`ef?ciency as high as 55 lm/W and a typical light transmit
`tance of about 4%. This gives a typical luminous ef?ciency
`of 2.2 lm/W for AMLCDs, Which exceeds the performance
`of all other display technologies. The brightness of LCDs
`can be increased by simply turning up the intensity of the
`backlight.
`The contrast in a display is another important attribute. It
`describes the achievable light intensity modulation in the
`image betWeen the brightest and dimmest pixels. Images
`With greater contrast are more sparkling in appearance. The
`best AMLCDs achieve contrast ratios as large as 100:1.
`Ambient illumination affects the contrast of the displayed
`image. The component of the ambient illumination that is
`re?ected from the display’s surfaces Will be added to the
`emitted intensity of the image to be displayed. The higher
`the contrast, the more tolerant the display is to ambient light.
`Of all displays, AMLCDs have the highest tolerance to
`ambient light, because of the presence of polariZers and
`
`

`

`US 6,271,825 B1
`
`3
`because of the ability of AMLCDs to independently adjust
`the intensity of the backlight.
`The viewing characteristics of electronic displays are
`speci?ed by the viewing distance and vieWing angle ranges.
`The minimum vieWing distance is related to the pixel pitch
`via the resolution ability of the observer’s retina. Displayed
`images are usually best vieWed at normal incidence. Maxi
`mum horiZontal and vertical vieWing angles aWay from the
`normal are determined by the type of the display and the
`layout and the optical design of the pixels. VieWing angle
`ranges of 130° horiZontal and 115° vertical are average for
`typical AMLCD displays.
`Full-color displays are expected to be able to display 256
`(8-bit) shades of each of the highly saturated primary colors
`red, blue and green. This results in a total of 25633 or
`16,777,216 colors that (in principle) can be displayed.
`Full-color capability has been available on CRTs for quite
`some time via the selection of the R, B and G phosphor
`materials, as Well as the control of the electron beam. Full
`color Was demonstrated for the ?rst time With LCDs in 1993,
`by developing 8-bit data driver circuits [G. H. Henck Van
`Leeuven et al., “A Digital Column Driver IC for AMLCDS”,
`Euro-Display, pp. 453—456, 1993; see also H. Okada, K.
`Tanaka, S. Tamai and S. Tanaka, “An 8-Bit Digital Data
`Driver for AMLCDs”, Society for Information Display
`International Symposium Digest of Technical Papers, vol.
`XXV, pp. 347—350, 1994]. To date, several manufacturers
`have demonstrated full color AMLCDs by using amorphous
`silicon (a-Si), thin-?lm transistors (TFT) as the sWitches.
`Saturated primary colors are de?ned by using a uniform
`“White” backlight in combination With three color ?lters.
`Driver electronics is used to provide an optimal lineariZation
`of the liquid crystal response, in order to facilitate the
`additive mixing of colors.
`Direct-vieW electronic displays With diagonals up to about
`31“ are usually manufactured in monolithic form, With the
`entire pixel array fabricated on a single continuous medium.
`The siZe of a commercial CRT is limited by the de?ection
`optics and the Weight of the unit to about 35 “. Commercial,
`monolithic AMLCDs are currently limited to siZes less than
`12“ because of manufacturing yield and cost. Commercial,
`16“ AMLCD displays are in product development. Display
`siZes of up to 21“ have been demonstrated in research (M.
`HijikigaWa and H. Take, “Future Prospects of Large-Area
`Direct VieW LCDs”, Society for Information Display Inter
`national Symposium Digest of Technical Papers, vol. XXVI,
`pp. 147—149, 1995). Very large electronic displays cannot be
`made in a monolithic fashion. Rather, each pixel is sepa
`rately fabricated, and then the display array can be
`assembled by accurately arranging pixels into roWs and
`columns. The alignment process is dif?cult and cannot be
`made With high precision over large areas. As a
`consequence, the pixel pitch in large-screen displays usually
`is on the order of at least 30 mm.
`Intermediate-siZed electronic displays With pixel pitches
`from about 0.6 to 3 mm can be assembled, in principle, from
`smaller monolithic pieces, With each carrying many pixels
`[see e.g., N. MaZurek, T. Zammit, R. Blose, and J. Bernkopf,
`“A 51-in Diagonal Tiled LCD VGA Monitor,” SID Interna
`tional Symposium, Digest of Technical Papers, vol. 24, pp.
`614—617, 1993]. These monolithic pieces are then arranged
`into a regularly tiled array to form a full display. In tiled
`displays, the pixel pitch on all tiles is preferably the same.
`Because of the small siZe of the tiles, this can be achieved
`With a tightly-controlled manufacturing process. The seam
`betWeen tWo adjacent tiles should be large enough to facili
`tate assembly. The seam Will be visible to the human
`
`4
`observer, unless the pixel spacing across the seam is the
`same as the pixel spacing on the tiles. This is very difficult
`to achieve. Consequently, to date, commercial prototype,
`tiled displays have had visible seams betWeen the tiles. The
`minimum achievable pixel pitch in tiled displays is,
`therefore, determined by the available assembly technology.
`
`SUMMARY OF THE INVENTION
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`This invention describes methods and apparatus for the
`correction of spatial non-uniformities in brightness that arise
`from the materials, manufacturing and operational param
`eter variations in electronic color displays. Such non
`uniformities can introduce gradual or abrupt variations in the
`displayed image. Gradual non-uniformities are usually
`found on monolithic display screens, While displays that are
`composed of a multitude of smaller tiles (With each carrying
`a single pixel or an array of pixels), exhibit abrupt non
`uniformities at the edges thereof. Combinations of abrupt
`and gradual non-uniformities may also exist in displays.
`Brightness non-uniformities are visible to the observer if the
`magnitude of the disturbance crosses the human eye’s
`threshold sensitivity for brightness.
`This invention comprises methods and apparatus for
`correcting brightness variations across the pixels of elec
`tronic displays, Whether these variations arise from the
`primary colors themselves, or their additive mixing to pro
`duce other colors from these primary colors. These methods
`applies only to the correction of brightness non-uniformities.
`Methods for correcting color variations, or combinations of
`both brightness and color variations, are described in
`co-pending patent application, Ser. No. 08/628,308, ?led on
`Apr. 5, 1996, herein incorporated by reference.
`The correction methods incorporate the measurement of
`the brightness characteristics of the display. Corrections can
`be applied to selected pixels or all pixels of the display.
`Corrections are performed on the electronic drive signals
`that control the brightness of the primary color elements of
`the chosen pixels on the display. The measured brightness
`characteristics are expressed in terms of a set of parameters
`Which are stored after suitable transformations in non
`volatile memory. The stored parameters are selectively
`retrieved during the operation of the display and used to
`scale and/or interpolate drive signals in real time.
`The modi?cations of the drive signal can be realiZed using
`serial or parallel electronic implementations. In the serial
`case the controller, pixel processor, and parameter storage
`are centraliZed and operate on the entire pixel data stream.
`In the parallel implementation these functions are performed
`by multiple smaller functional units on the pixel data stream
`arriving at each column driver circuit. Corrections are per
`formed With respect to a chosen reference system such that
`any remaining gradual and abrupt brightness non
`uniformities over the selected display pixels fall beloW the
`detectable threshold under the intended vieWing conditions.
`These correction methods can also be used for correcting
`brightness non-uniformities arising from an uneven aging of
`the display.
`Apparatus for an automatic self-calibrating function is
`described herein as Well. It provides a capability to correct
`the display for brightness non-uniformities arising from
`manufacturing, operational conditions, aging, and the vieW
`ing environment (ambient light).
`
`65
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`Acomplete understanding of the present invention may be
`obtained by reference to the accompanying draWings, When
`
`

`

`US 6,271,825 B1
`
`5
`considered in conjunction With the subsequent, detailed
`description, in Which:
`FIG. 1 illustrates a schematic vieW of tristimulus R, G and
`B color coordinates;
`FIG. 2 portrays the luminance of a tiled, electronic color
`display as a function of position along a line crossing over
`several display tiles, With the upper panel depicting the line
`across the tiles and the loWer panel, the brightness along the
`line, and both abrupt and gradual luminance variations
`shoWn at the edges of the tiles and on the tiles, respectively;
`FIG. 3 depicts the luminance of a tiled, electronic color
`display as a function of position before and after correction,
`With the detection thresholds for both maximum abrupt and
`maximum gradient non-uniformities shoWn as inserts; and
`FIG. 4 shoWs a block diagram of one embodiment of a
`luminance correction method of this invention;
`FIG. 5 shoWs a block diagram of one embodiment of the
`luninance correction method of this invention, depicting a
`parallel implementation With luminance scaling scalers and
`coef?cient memories operating on the bit stream arriving at
`each column driver circuit of a normally dark display;
`FIG. 6 depicts a self-calibrating apparatus mounted on a
`?at panel display With a photodetector mounted on a mov
`able arm in order to provide an x-y pixel scan capability; and
`FIG. 7 is a ?oWchart of the steps of the color and
`luminance correction method of the invention.
`
`DESCRIPTION OF THE PREFERRED
`EMBODIMENT
`Generally speaking, this invention features methods for
`correcting brightness variations across the pixels of elec
`tronic displays, Whether these variations arise from the
`primary colors themselves, or their additive mixing to pro
`duce other colors from these primary colors, or other com
`ponents of a display (such as backlight sources).
`Color Classi?cation, De?nition and Uniformity
`Every color has three basic characteristics: hue, lightness
`and chroma. Hue is the property that distinguishes and gives
`each color its name. Lightness measures the amount of light
`re?ected from the hue. Chroma measures the amount of
`saturation or concentration thereof. There are tWo common
`Ways to classify colors by using these characteristics. The
`Munsell system Was devised by the American portrait
`painter Albert Munsell in the early 1900s (D. Nickerson,
`“History of the Munsell System, Company and Foundation,
`1—111”, Color Research Applications, vol. 1, pp. 7—10,
`69—77, 121—130, 1976). The Munsell system classi?es each
`color (hue) according to value (Which is related to lightness)
`and chroma. Munsell’s classi?cation is subjective due to the
`differences betWeen individuals in perception of colors. The
`CIE system of colors Was developed by the International
`Commission on Illumination, or, CIE (see, e.g., G. WysZecki
`and W. S. Stiles, Color Science, 2nd edition, Wiley, NY,
`1982). The CIE system is based on the use of spectropho
`tometers and the concept of a standard observer, expressed
`in color tables, and, thus, independent of a speci?c observer.
`Display colors are formed by additively combining three,
`primary, saturated colors, for example, red (R), green (G)
`and blue
`A speci?ed number (for example: 28=256
`shades) of each primary color is generated by the respective
`color element in a pixel of the display. In this case each pixel
`Will carry the R, G and B colors in three color elements. For
`example, in a CRT, When hit by the electron beam, the
`selected element of a color pixel emits light With its intensity
`approximately proportional to the electron beam ?ux. The
`same happens in the other primary color elements of the
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`6
`same pixel. The actual sensation of color occurs When light,
`emitted from each of the primary color elements Within a
`pixel, blends in the eye and the visual cortex of the vieWer.
`Because of this, human factors are signi?cant in the percep
`tion of displayed colors. Aspeci?c, illustrative model for the
`de?nition of colors is hereinafter discussed. The invention is
`not limited to this illustrative color combination model, but,
`rather, applies to all possible Ways of combining additively
`primary colors.
`Assume that the red (R), green (G) and blue (B) primary
`colors have been de?ned. This de?nition includes the tabu
`lation of the intensity-Wavelength dependence for each of
`the primaries. According to the standard color theory, any
`other color Q Within the color triangle formed by the
`primaries R, G and B can be expressed as a linear combi
`nation of the primaries
`
`Where the coef?cients R, G and B are the tristimulus values
`(see FIG. 1). No attempt is made in this disclosure to
`distinguish betWeen linear and non-linear, e.g., gamma
`corrected R, G and B color coordinates. The color de?ned in
`Equation (1) can also be speci?ed by using the normaliZed
`chromaticity coordinates (r, g) de?ned by
`
`In this model, each color Q is uniquely de?ned by specifying
`the three tristimulus values for each pixel. The complete set
`of colors formed in this fashion, via Equation (1), are called
`the gamut of colors for the chosen primaries. By specifying
`these three tristimulus or color coordinate values for each
`pixel in the N><M array of the display, the entire image has
`been de?ned. The tristimulus values correspond to elec
`tronic drive signals controlling each color element. For
`example, in an AMLCD the drive signals are voltages
`applied to the liquid crystal cells in order to modulate their
`optical rotation and thus change the optical transmission.
`Brightness uniformity describes the ability of the display
`to de?ne all colors uniformly across the entire pixel array of
`the display for any prede?ned combination of the primaries.
`This requires very good control over both the primaries and
`the tristimulus values. There are many potential sources of
`non-uniformities. Electron beam de?ection and spot siZe are
`the primary sources of color non-uniformity in CRTs, While
`materials, manufacturing- and backlighting-related issues
`are the most common factors responsible for non
`uniformities in AMLCD displays. Another mechanism giv
`ing rise to non-uniformity originates from the additive color
`formation process under the vieWing conditions of the
`display. For example, an ambient light gradient may intro
`duce a non-uniformity into the additive color sum When the
`re?ected light interferes With the emitted light. This phe
`nomenon limits the use of electronic displays in bright
`ambient light.
`Perception tests With human observers have shoWn that
`tristimulus value differences as small as 2 to 4% are observ
`able under demanding vieWing conditions. Perception tests
`also shoW that gradual color non-uniformities occurring
`continuously over many pixels are less perceptible, because
`the observer loses the reference over the area of the display
`screen. In fact, gradual color coordinate changes as large as
`10 to 20% over the siZe of the display screen may not be
`disturbing to an average vieWer. Under normal vieWing
`conditions, brightness uniformities are more observable
`When vieWed from a greater distance, rather than from up
`close.
`
`

`

`US 6,271,825 B1
`
`7
`Brightness non-uniformities in monolithic electronic dis
`plays are caused by process variations, Which tend to cluster
`into gradual changes over large sections of a display.
`Therefore, monolithic displays can be manufactured With
`relatively large process tolerances. On the other hand, abrupt
`brightness changes betWeen adjacent pixels or groups of
`pixels are disturbing. Such abrupt non-uniformities arise in
`displays Where each pixel, or array of pixels, has been
`manufactured separately and then assembled to form a
`complete, tiled, pixel array. Materials, manufacturing and
`design-related factors introduce abrupt non-uniformities in
`tiled displays. Another possible source of non-uniformities
`in tiled displays arises from the possibility that pixels close
`to the edge of a tile have different characteristics than do the
`interior pixels. If uncorrected, this effect may either cause
`scalloped luminance or chromaticity gradients close to the
`edge of tiles.
`Referring noW to FIG. 2, the combination of both gradual
`and abrupt brightness non-uniformities on a tiled display is
`illustrated. The upper portion of FIG. 2 depicts a portion of
`a roW of a tiled, color FPD 10, consisting of three adjacent
`tiles 12, 14 and 16. Luminance of the tiles 12, 14 and 16 is
`measured at a number of positions along line 18, placed at
`an arbitrary position of the tiles 12, 14 and 16. The loWer
`portion of FIG. 2 is a graphical representation of luminance
`measured along line 18. In this example, abrupt transitions
`in luminance occur at the boundary of tiles 12 and 14 (as
`shoWn as reference numeral 20 on the luminance graph) and
`at the boundary of tiles 14 and 16 (as shoWn as reference
`numeral 22 on the luminance graph). Gradual variations in
`luminance occur Within tile 12, tile 14, and tile 16, as
`indicated by the respective sloped line portions 24, 26 and
`28 of the luminance graph.
`This invention covers methods and apparatus that correct
`for brightness non-uniformities in electronic displays. While
`the methods Work both for gradual and abrupt non
`uniformities, they are most useful for the latter, especially
`for displays that are assembled from single pixels or are tiled
`from rectangular arrays of pixels.
`Description of Brightness Correction Methods
`In order to accurately match colors on electronic displays,
`the perceived brightness and color have to match Within the
`human eye’s discrimination threshold. “Brightness”
`describes the appearance of the radiant ?ux of an object. The
`brightness of an object depends on the vieWing conditions of
`the display and the adaptation of the observing eye. The
`psychophysical equivalent to brightness is luminance, Which
`is, of course, independent of vieWing and observation con
`ditions. Luminance is quanti?ed by using the concept of
`luminous ?ux per projected area of the source of light.
`Luminance is often normaliZed so that White light is
`assigned a value of 1 on a relative scale. The ability of the
`human eye to discriminate betWeen tWo luminances is
`measured using Weber’s fraction. Assume that tWo objects
`are vieWed side by side, With one object having the lumi
`nance of B, and the other B+AB. Assume further that AB is
`increased from 0 to a value that makes the brightness of the
`tWo objects detectably different. The discrimination thresh
`old value, then, for AB de?nes Weber’s ratio as AB/B.
`According to extensive visual discrimination studies,
`Weber’s fraction is not a constant (i.e., Weber’s original
`laW), but, rather, depends on the luminanc