Organization

This thesis is organized as follows. In Chapter 2, the mechanism of different categories of the CBU phenomenon caused by different human eyes movements will be discussed. Then, some prior arts of the CBU suppression will be described. In Chapter 3, proposed method, Stencil Field Sequential Color (Stencil-FSC) method, will be introduced in detail, and some algorithm optimizations will be discussed. In Chapter 4, two experimental demonstrations will be presented and verified the Stencil-FSC method, and a comparison of display performance will be made between a conventional LCD and the FSC-LCD with the proposed method.

In Chapter 5, by using created simulation program, hardware parameters will be optimized to get best CBU suppression. Finally, conclusions and future works will be summarized in Chapter 6.

Chapter 2 Color Breakup and Prior Solutions

The FSC-LCD is an effective display mechanism without CFs and has several advantages when compared to the conventional LCD. However, it faces a serious issue: color breakup (CBU). In order to suppress CBU, some basic concepts need to be understood. First, the eye movement response will be mentioned, and the different categories of CBU phenomenon caused by different human eyes movements will be discussed. Next, color difference (ΔE) between the target images and the CBU images was introduced and utilized to evaluate CBU phenomenon. Finally, several prior works about CBU suppression will be mentioned.

2.1 Mechanism of Color Breakup in Conventional FSC-LCDs

The Human eye is a complex visual system, and its structure is shown in Fig. 2-1.

Light goes through the cornea and then passes through the pupil controlled by the iris to adjust the incident magnitude of light. Then the lens focuses the light on the retina, and there are two kinds of receptors to detect intensity and color of light. The rod cells are sensitive to light intensity; the cone cells are sensitive to color. The receptors detect light and convert it into electrochemical signals. Finally, the signals pass through the optic nerve to the brain, and humans can perceive images[17]. The CBU phenomenon in the FSC-LCD has strong dependence on perceiving. According to research, humans use two major types of eye movement to perceive objects, saccade and pursuit. When watching an FSC-LCD, different eye movements will cause different CBU phenomenon; static CBU and dynamic CBU.

Fig. 2-1. Structure of human eye (From The Internet Encyclopedia of Science)

2.1.1 Physiology of eye movements

The fovea lies near the projection point at the center of the retina. This region has the highest resolution and is 0.1% of retina area. In order to see objects clearly, humans need to move their eyes to focus on the fovea. In perceiving research, two major types of eye movements are mentioned: saccade and pursuit[18]. They are caused by different observed object movements.

The first type of eye movements is saccade. Saccade is a rapid, random movement while perceiving static objects. Saccade movement moves around objects to focus on the fovea and gathers correct visual information. Take Fig. 2-2 for example. The test image is shown in the left picture, and the eye movement for the static image is lined with black on the right picture. In order to recognize the woman, the observer’s eye moves around her face to collect information. Generally, saccade is spontaneous, and the speed happens quickly at about 200 degree/sec[19].

The other type of eye movements is pursuit. Pursuit is a smooth and predictable movement when perceiving dynamic objects. When perceiving a moving object, the human eye will follow at the same velocity of the object to focus on the fovea and catch clear images. Compared to saccade, pursuit velocity is much slower, and some

research proposes the velocity is about 90 degree/sec. If a target moves at a velocity faster than 90degree/sec, eye cannot pursuit it, as shown in Fig. 2-3. Moreover, there is one thing worth to noticing about pursuit movement: pursuit latency. Pursuit latency means the delay in eye pursuit, and it is defined as the difference in beginning motion time between the target and the eye. Refering to research, the pursuit latency is about 100 ms to 150 ms.

Fig. 2-2 (a) Perceived image, and (b) Saccade movement (black line) (Eye, Brain, and Vision, p.80, D.H. Hubel)

Fig. 2-3 Pursuit movement curve. If a target moves with a velocity faster than 90degree/sec, eye pursuit will not catch up.

(a) (b)

2.1.2 Static CBU and dynamic CBU

There are two major kinds of eye movements when perceiving images, and the CBU phenomenon is strong dependent on the eye movement. Therefore, the CBU phenomenon can also be categorized into two types according to the image and the eye movement: static CBU and dynamic CBU.

Static CBU phenomenon happens when perceiving static images. While humans perceive a static image, the eyes will move around the image to gather a clear image.

The mechanism of static CBU can be explained by Fig. 2-4[20]. Fig. 2-4(a) is a static image. When perceiving the image, eyes will move between white bars with saccade movement to get visual information, like the dash line in the figure. In a conventional FSC-LCD with RGB color sequence, the display will show red, green, and blue fields in time sequence. When moving with saccade, like the dash line, the field images of different colors will separate. The pre-field is on the left side, and post-field is on the right side, so the CBU phenomenon appears as shown in Fig. 2-4(b). The effective factors of static CBU are image difference between each field and filed rate. Many methods have been proposed based on the adjustment of the factors to suppress the CBU. The mechanism of the methods will be given in the next section.

The other kind of CBU, dynamic CBU, happens when perceiving dynamic images. In order to focus on the fovea, the eyes will pursuit the moving object at the same velocity, and it will cause dynamic CBU. The mechanism of dynamic CBU can be explained by Fig. 2-5. When a white bar is displayed on an FSC-LCD with RGB color sequence, and viewer perceives it with the pursuit movement, the spatial-temporal relation can be indicated by Fig. 2-5(a). The horizontal axis is the display position, the vertical axis is time, and the eye trace line is indicated by the orange line. In order to make the eye integration easier to be understood, a coordinate

E y e tr a c e lin e

transformation is used. The horizontal axis is transformed to the retina position, and the eye trace line change to a vertical line as shown in Fig. 2-5(b). After eye integration, the dynamic CBU will be perceived, like Fig. 2-5(c). The Dynamic CBU depends on the relation between the eye movement and the pursued object, and it also be affected by the display frequency. Some methods also proposed for suppressing dynamic CBU, and the mechanisms will be discussed in the next section.

(a) (b)

Fig. 2-4 Mechanism of static CBU. (a) Static image and eye trace line with saccade movement (dash line) and (b) static CBU phenomenon.

(a)

(b)

(c)

Fig. 2-5 Mechanism of dynamic CBU. (a) The spatial-temporal relation of displaying white bar in RGB color sequence, (b) a coordinate transformation of (a) to make eye trace line to be a vertical line, and (c) dynamic CBU phenomenon.

2.2 Prior Works of CBU Suppression

If the CBU issue on the FSC-LCDs can be suppressed effectively, the FSC-LCDs will have high potential to be a novel display with lower power consumption, higher color saturation, and lower cost. Therefore, how to suppress the CBU phenomenon has become an attractive research topic in the display technology, and many methods have been proposed. Those methods can be divided into three major categories: increasing field rate, inserting multi-primary color fields, and utilizing motion compensation. .

2.2.1 Field rate increasing

Field rate increasing is a direct method for suppressing CBU[21]. The CBU width (CBUW) depends on the field rate (F) and the relative velocity (V) between the viewer eye and the image. CBUW can be obtained by Eq. 2-1.

CBUW= V/F (2-1) If the field rate is increased, according to Eq. 2-1, the CBUW will be reduced, and human eye will be less sensitive to the CBU phenomenon. Moreover, prior research has implicated the field rate human can not perceive CBU phenomenon is about 1000Hz. For projector with color sequential technique, the field rate has been raised to more than 1000 Hz successfully by color wheel, and the CBU phenomenon is almost eliminated. However, utilizing color sequential technique on an LCD is extremely hard to raise field rate to 1000 Hz because the LC response time and TFT scanning time are limited.

2.2.2 Multi-primary color fields

Inserting multi-primary color fields to suppress CBU have been done by Tatsuo Uchida research group in Tohoku University for years, and RGBY, RGBWmax, RGBCY color sequences have been proposed[22-23]. The concept of these color sequences is to reduce the image difference between each field and prevent from the appearance of the sensitive colors on the CBU band. The CBU simulations of each color sequence compared to RGB color sequence are sown in Fig. 2-6. By the simulations, the colors of CBU widths are less sensitive to human eye compared to those of RGB color sequence, so CBU can be suppressed. However, these methods face some issues. The RGBWmax and RGBY color sequence may cause color distortion and reduce contrast ratio; the RGBCY color sequence needs 300 Hz field rate to be achieved, and it is hard to be realized on hardware.

Fig. 2-6 CBU simulations of different color sequences. (a) RGB, (b) RGBWmax, (c) RGBY, and (d) RGBCY

RGBWmax

240Hz RGB

180Hz

RGBCY 300Hz RGBY

240Hz (a) (b)

(c) (d)

2.2.3 Motion compensation

Motion compensation method aims for suppressing dynamic CBU[24]. By image signal processing related to eye movements, compensated images are displayed to make each field of one frame at the same position on the retina. The compensation mechanism can be explained by Fig. 2-7. The figure of spatial-temporal relation has been used in the last section. In order to compensate the eye movement, field images will shift a distance according to the eye movement. After eye integration along the eye trace line, a white bar without CBU phenomenon will be perceived. Therefore, the dynamic CBU with predictable eye movement can be eliminated completely by the motion compensation method. However, if predicted eye movement is mistaken, or even in the opposite direction, this method will cause more serious CBU, like Fig. 2-8.

Therefore, if an FSC-LCD is for multi-viewers, or there are more than two motion objects on the display, the motion compensation method will be failed for suppressing CBU phenomenon.

Fig. 2-7 Motion compensation method. CBU phenomenon with predictable eye movement

Fig. 2-8 Motion compensation method. CBU phenomenon with opposite eye movement

2.3 CBU Evaluation Index: ΔE*

_ab

Many methods were proposed to reduce CBU. Therefore, how to evaluate the CBU phenomenon or improvement is important. The CBU phenomenon happens on the edge of images and causes difference between the perceived images and displayed target images. Therefore, utilizing the color difference between the target image and the CBU image to evaluate CBU phenomenon is straight and convenient. The concept of color difference had been discussed and proposed in Chromatics since early 1990s, and some optimized indexes of color difference were presented in recent decades.

In order to define the color difference, a uniform color space was established at first. In the late 1920s, two experiments were done by Wright and Guild to estimate the color matching function, and the experimental setup is shown in Fig. 2-9[25-26].

There was a circle region with the reference and test fields, and the size was defined by the visual angle with 2°. The reference field was projected by a monochromatic light with specific frequency, and the test field was composed by three adjustable

primary lights at wavelengths of 435.8, 546.1, and 700.0nm. In the experiment, the tester would be asked to control the components of the three primary lights and make the color of the test field the same as that of the reference field. Finally, the components of each reference monochromatic light would be recorded, and the experiment results of the tristimulus values of three primary lights could be gathered as r, g, and b as shown in Fig. 2-10. There was one thing worth to notice, the primary light component might be negative! In the experiment, it meant the adjustable primary color of the negative component was place on the side of reference field. However, the negative tristimulus value would lead to some calculation trouble in application, so it was transformed into the CIE system of colorimetry in 1931. This was accomplished by choosing two imaginary primaries, x and z, such that they produce no luminance response, leaving all of the luminance response in the third primary, y.

The transformation matrix is presented in Eq. 2-1, and the figure of the tristimulus values is shown in Fig. 2-11. By utilizing the process, there was no negative tristimulus value in the figure.

⎥⎥

Then, because color is determined by three components, light source (P), object reflection(R), and the tristimulus values of human eye(x, y, and z) as shown in Fig.

2-12, the colored tristimulus could be defined by XYZ tristimulus by those components, like Eq. 2-2.

∑

∑

= ⁷⁸⁰

380

) ( ) ( /

100 P λ y λ

k

Fig. 2-9 The experimental setup of color matching

Fig. 2-10 Curve of r, g, and b tristimulus values

(Ref: http://www.math.ubc.ca/~cass/courses/m309-03a/m309-projects)

Reference field

Test field Test field

Reference field

r

g b

Fig. 2-11 Curve of x, y, and z tristimulus values

(Ref: http://www.creativepro.com/printerfriendly/story/13036.html )

Fig. 2-12 Color is determined by three components, light source (P), object reflection (R), and the tristimulus values (x, y, z).

The XYZ tristimulus values were a three-dimensional space in which each axis was a primary, and a sample’s tristimulus values were defined by a coordinate.

However, according to prior Chromatics research, color information was independent of luminance[25], and the objective was to establish a chromaticity diagram with

z

y x

Eye tristimulus

(P)

(R)

(x,y,z)

taking the ratio of the tristimulus values to the sum, X+Y+Z, like Eq. 2-3, and it was meant to make a projection on the plane defined by X+Y+Z=1.

)

Finally, a chromaticity diagram called CIExyY was completed and presented in Fig.

2-13. However, the color difference on the CIExyY chromaticity diagram presented by MacAdam in 1942 is shown in Fig. 2-14[26]. Colors within an ellipse on the diagram denote that testers can not distinguish. From the results, the color difference is not uniform in the CIExyY chromaticity diagram. Therefore, the other optimized chromaticity diagram, CIELAB (1976) was proposed to overcome the issue. The equations for the CIE LAB chromaticity diagram are shown in the Eq. 2-4 where Xn, Yn, and Zn are the tristimulus values of the reference white.

In the CIELAB chromaticity diagram, the L*, a*, and b* denote luminance, color component of yellow-blue, and color component of red-green, respectively, as shown in Fig. 2-15. By utilizing CIELAB, a more uniform chromaticity diagram could be established. Therefore, the index of color difference was defined by ΔE*ab in the CIE LAB chromaticity diagram, as in Eq. 2-5. Consequently, the CBU phenomenon will be evaluated by the ΔE index between the target image and the CBU image to verify the proposed Stencil-FSC method.

2 / 1 2 2

2 ( *) ( *) ]

*) [(

* L a b

E _ab = Δ + Δ + Δ

Δ (2-5)

Fig. 2-13 CIExyY chromaticity diagram

(Ref. http://www.colblindor.com/2007/01/18/cie-1931-color-space)

Fig. 2-14 Experimental results of color difference in CIExyY chromaticity diagram (Ref. Hunt, John Wiley and Sons, The Reproduction of Colour, 6^th ed, p.293)

Fig. 2-15 CIE LAB chromaticity diagram

(Ref. http://www.newsandtech.com/issues/2002/02-02/ifra/02-02_greybalance.htm)

2.4 Summary

The physiology of eye movements and the mechanism of CBU caused by different eye movements have been introduced. In order to suppress the CBU issue on FSC-LCDs and achieve novel type displays with lower power consumption, higher color saturation, and lower cost, many researchers have proposed several methods for CBU suppression, like increasing field rate, inserting multi-primary color fields, and utilizing motion compensation. However, these methods have their own challenges, such as LC limitation, image distortion, and uncertain eye movement, so they are hard to be applied on the hardware. Therefore, we proposed the “Stencil Field Sequential Color (Stencil-FSC) method to effectively suppress CBU with field rate of 240 Hz, and it is easier to be implemented on hardware.

Chapter 3 Stencil Field Sequential Color Method

The “Stencil Field Sequential Color (Stencil-FSC)” method with 240 Hz field rate was proposed for suppressing CBU. The concept and algorithm will be introduced at first, and the two processes for optimization, backlight colorful method and Fourier transformation process in will be presented in the following. Finally, the performance and the feasibility of Stencil-FSC method could be obtained.

3.1 Stencil Field Sequential Color Method

3.1.1 Concept

The conventional RGB color sequence utilized to display full color image on the FSC-LCD causes serious CBU because the field images are colorful and bright, like Fig. 3-2(b). Therefore, the “Stencil” concept was utilized to suppress CBU. The meaning of “stencil” is a unique technique of painting. Take Fig. 3-1 for example, if a boy will be painted on the wall by the stencil technique, a base color, white (Fig. 3-1 (b)), is drawn firstly, and then blocks of detail parts with different colors are put on the base color in order to add each detail color (Fig. 3-1(c)). Finally, the painting can be completed by the stencil technique as shown in Fig. 3-1(d). By the concept of stencil, the “Stencil Field Sequential (Stencil-FSC) Method” was proposed to suppress CBU. The Stencil-FSC method is a color sequence with four fields. A multi-color image is displayed in the first multi-color field instead conventional single-color fields, and the red, green, and blue field images are displayed to add red, green, and

B-field R-field G-field

R-field G-field B-field Multi-color field

(b)

(c) (a)

Target image

blue image details as shown in Fig. 3-2(c). By the method, the most color and luminance information are shown at the first multi-color field, so the red, green, and blue field images are darker and less colorful compared to those of conventional RGB color sequence, and it will be helpful for suppressing CBU. Additionally, the concept of utilizing multi-color fields instead of conventional single-color fields to display images was first proposed in FSC technique.

Fig. 3-1 The flowchart of stencil method. (a) Put block of base image, (b) paint base color, (c) put blocks of detail parts with different colors and paint color, and (d) complete the painting. (From http://www.wretch.cc/blog/Bbrother&article)

Fig. 3-2 (a) Target image. The field images of (b) RGB color sequence and (c) Stencil-FSC method.

(a) (b) (c) (d)

3.1.2 Display module

In order to get the first multi-color field, the FSC-LCD with Stencil-FSC method can be based on a locally controlled backlight system. The backlight system is divided into several regions, and it can be locally controlled according to the displayed image as shown in Fig. 3-3. Furthermore, the locally controlled backlight can be intensity or color, like Fig. 3-3(b) and Fig. 3-3(c) respectively. The technique is also called High Dynamic Range (HDR) technique [27-28], and the flowchart is given in Fig. 3-4. At first, the LC signal of target image with full-on white backlight (Io) is inputted, and then backlight LED signals (ILED) of each region are gotten by an algorithm of locally controlled backlight, such as maximum, root, or Inverse of a Mapping Function (IMF)[29]. Next, in order to get the compensated LC signal (Compensated LC), the backlight distribution of locally controlled backlight is obtained by convolution of ILED and light spread function (LSF). The compensated LC can be gotten by comparing the backlight distribution to the target image. Finally, the image can be displayed by composing the real backlight distribution of ILED and compensated LC.

The backlight can be locally controlled according to the image, so it can enhance the contrast ratio, save power consumption, and increase color saturation compared to those of conventional full-on backlight system. Therefore, by implementing the technique on the FSC-LCD, not only can the first multi-color field be generated, but

the display can also have the advantages of local controlled backlight technique.

在文檔中 利用Stencil-FSC法抑制色分離並達低功率之色序型液晶顯示器 (頁 18-0)