KR100606854B1 - Optimized Design Method for Light Reflection Pattern of Light Guiding Panel - Google Patents

Abstract

An optimal design method of a light reflection pattern of a light guide plate which can be executed automatically by using a computer is disclosed. The ink dot printed on the lower surface of the light guide plate is defined by the density function. The backlight unit is automatically modeled by assigning specific values to the coefficient of density function and the remaining design elements of the backlight unit. Simulation of the modeled backlight unit using the ray tracing technique is performed to calculate the objective function value indicating the uniformity of the brightness of the light guide plate. Modification of the density function using the simplex search technique, remodeling the backlight unit applying the pattern of the ink dot determined by the modified density function until the calculated objective function value satisfies the set limit value, and the remodeled backlight Iterate through a series of loops called simulations for the unit.

Description

Optimal Design Method for Light Reflection Pattern of Light Guiding Panel

Detailed description of the embodiments of the present invention will be made with reference to the accompanying drawings, in which numerals designate corresponding parts in the drawings.

1 illustrates a structure of a conventional back-light unit (BLU).

2 illustrates a general pattern of ink dots printed on the bottom surface of the light guide plate.

FIG. 3A illustrates a dot arrangement method of a circular dot-on-grid type, and FIG. 3B illustrates a dot arrangement method of a circular-dot-on-honeycomb type. 3C illustrates a hexagonal dot-on-honeycomb type dot arrangement method, respectively.

4 may be defined as the density function d (x) of the dot printing pattern according to the distance from the lamp.

FIG. 5 is a diagram for describing a method of calculating a radius r _i of a dot from a density function d (x) in a circular dot on grid model.

6 shows symmetry (a) and asymmetry (b) in the 'circular dot on honeycomb' pattern.

FIG. 7 is a view for explaining a method of calculating a radius of a dot from a density function given in a 'circular dot-on-honeycomb' ink dot pattern.

FIG. 8 is a diagram for explaining a strip in an asymmetrical arrangement of a 'circular dot on honeycomb' ink dot pattern.

FIG. 9 is a view for explaining a method of calculating the size of a dot in a hexagonal-dot-on-honeycomb ink dot pattern.

10 is a view for explaining the BLU modeling and ray tracing technique.

FIG. 11 is a diagram for describing a method of predicting a luminance distribution of a light guide plate using a ray tracing technique. FIG.

12 is a flow chart showing the overall overview of the BLU optimization design method according to the present invention.

13 is a graph showing the uniformity of plane light according to the density function for adjusting the size of the dot.

14 shows a grid count value obtained as a simulation result of the BLU according to the dot pattern calculated by the density function D (x) .

15 is a flowchart illustrating a method of finding a density function of an ink dot that provides an optimal luminance uniformity using a simplex search technique.

16 shows a case where a density function is made of three three-dimensional polynomials.

FIG. 17 is a detailed flowchart illustrating a process of finding a new simplex required for density function conversion by a Nelder-Mead simplex search technique.

18A and 18B are luminance distribution graphs displaying luminance generated in front of the light guide plate in two and one dimensions by an optimal ink dot printing pattern.

19 is a graph illustrating a process of changing a density function in the process of performing a simplex search algorithm.

20 is a 2D luminance graph illustrating a process of changing the light emitted to the front of the LGP while the density function is changed during the simplex search algorithm.

** Description of the symbols for the main parts of the drawings **

112: light guide plate 114a, 114b: cold cathode fluorescent lamp

116a, 116b: lamp reflector 118: prism sheet

120: ink dot 122: diffusion sheet

124: reflector 126: LCD panel

130: grid plane

The present invention relates to a backlight unit (BLU), and more particularly, to the design of a light reflection pattern of a light guide plate, which is one of the main components of the BLU.

The structure of a conventional general BLU is shown in FIG. The BLU includes a light guide plate 12 made of acrylic, a cold cathode fluorescent lamp 14 provided on the side of the light guide plate 12, and a reflector 16 reflecting light generated by the fluorescent lamp 14 into the light guide plate. ), And a diffuser plate 22 for diffusing light emitted from the front surface of the light guide plate 12 to improve luminance uniformity, and a prism plate 18 for narrowing the viewing angle of the emitted light to increase front luminance. In addition, a light reflection pattern 20 is formed on the lower surface of the light guide plate 12 to diffuse light by diffuse reflection to change the direction of light input from the light guide plate side to the front of the light guide plate. The reflection sheet 24 is disposed below.

Among the characteristics of BLU, it is important to produce uniform flat light on the front of the LCD and to emit light with high luminance enough to illuminate the LCD brightly even in bright places. Such high brightness and excellent brightness uniformity depend largely on the light reflection pattern formed on the lower part of the light guide plate. There are various kinds of light reflection patterns. Among them, a printing pattern method of printing a diffusion ink having a light diffusion function in a dot pattern on the lower surface of the light guide plate and a plurality of V-shaped grooves formed on the lower surface of the light guide plate V-cut method.

In addition, in the BLU for large LCDs, a direct type BLU is being developed in which several light sources are directly installed on the bottom of the LCD. In the direct type BLU, the geometric shape of the reflector that transfers light from the back of the light source to the front of the BLU is very important.

It is well known that when the light reflection pattern is a dot pattern of diffusion ink, it is adjusted by the density and arrangement of the ink dots. 2 illustrates a general pattern of ink dots printed on the bottom surface of the light guide plate 12. If the size of the ink dot farther from the light source is larger than that of the ink dot closer to the light source as shown in FIG. You can make In the ink dot pattern, the ink is printed with hundreds of thousands of small dots of different sizes. In general, it is designed in a dense pattern where ink density is low near the light source and where the ink density is far away. As such, the brightness and uniformity of the plane light emitted to the front surface of the BLU vary depending on the printing pattern of the diffusion ink dot. Therefore, in order to obtain a desired light distribution, it is very important to find a dot pattern that generates uniform plane light under given conditions.

The task of finding the optimal dot pattern is to perform BLU simulation on a specific dot pattern to predict the uniformity of luminance provided by the pattern, and to correct the improved dot pattern to provide better luminance uniformity based on the prediction result. In addition, the BLU simulation process is repeated for the modified dot pattern, and a process of determining a dot pattern that finally provides a desired level of luminance uniformity is generally performed. BLU simulation mainly uses ray tracing technique. The purpose of the BLU simulation using ray tracing techniques to calculate the optical efficiency that indicates that the optical energy injected from the lamp of the BLU somewhat efficiently front emitted by the BLU effectively, and further for example, the brightness uniformity of the generated on the front of the BLU-side and It is to. These BLU simulations are implemented in programs and are already used for BLU design. However, as a result of BLU simulation, it is necessary to repeatedly correct the printed dot pattern to improve luminance uniformity, and finally determine the optimized printed dot pattern based on trial and error based on the designer's sense and experience. Passive design was performed. This manual design method not only requires a large number of trial and error experiments, but also requires a very long time. Therefore, there is a need for an improvement that enables the design work of the optimum print pattern to be performed more efficiently.

Therefore, the present invention can automatically calculate the optimized light reflection pattern that can provide the desired brightness uniformity in the design of the light reflection pattern of the light guide plate of the BLU to reduce the time and effort required for BLU design Its purpose is to provide a way.

According to an aspect of the present invention, a method for optimal design of a light guide plate for a backlight unit using a programmed computer, the design related to the physical specifications of the backlight unit including at least the size of the backlight unit and the light guide plate and the number and placement of lamps Among the design elements related to the properties of the optical components employed in the backlight unit, including the light transmittance of the elements and the light guide plate and the reflectance of the light reflection pattern, the light reflection pattern is used as a variable design element that is an object of optimal design and correspondingly in a polynomial manner. By defining the characteristic function to be expressed and giving an initial value to the coefficient of the polynomial of the characteristic function, and the other design elements are regarded as fixed design elements, the backlight unit is given a predetermined constant value for each. A first step of automatically modeling; Calculating a target function value representing luminance uniformity of the light guide plate by performing a backlight unit simulation using a ray tracing technique with respect to the modeled backlight unit; And a coefficient of the characteristic function polynomial which causes the objective function value to be changed to improve the luminance uniformity until a predetermined termination condition is satisfied, wherein the calculated objective function value satisfies a preset threshold value. Is newly calculated through the execution of a simplex search algorithm, the calculated new function is applied to calculate the new light reflection pattern of the light guide plate based on the modified characteristic function, and the newly calculated light is calculated. After applying the reflection pattern to the model again the backlight unit, there is provided an optimal design method of the light reflection pattern of the light guide plate comprising a third step of feeding back to the second step.

The design method further includes the step of calculating a light reflection pattern defined by the characteristic function at that time when the termination condition is satisfied, and determining the final design proposal.

In this design method, the standard deviation of the grid counts ( c _ij ) or the grid counts representing the number of photons passing through a small square divided into a grid in front of the light guide plate when the BLU is simulated using the light tracking technique. It is preferable to use a function defined as the ratio of the maximum value and the minimum value of ( c _ij ) as the objective function.

An example of the characteristic function is a density function that defines the density according to the distance from the light source of the ink dot printed on the lower surface of the light guide plate to provide a light diffusion function. Another example of the characteristic function is a function that defines the spacing and / or size according to the separation distance from the light source of the V-cut formed on the lower surface of the light guide plate to provide a light diffusion function. Another example of the characteristic function may be a function that defines the number of light sources installed on the bottom of the BLU in the direct type BLU, the spacing between the light sources, and the geometric shape of the reflector that transfers the light from the rear of the light source to the front of the BLU.

The backlight unit simulation specifically measures the total reflection coefficient, absorption coefficient, and diffusion coefficient of each component of the backlight unit to determine which direction the photon changes path when the photon hits the surface of the specific component according to these coefficients. Determine by using a random number, record the information on the output direction and the output position of all the photons output to the front of the light guide plate, each square partitioned into a grid on the front of the light guide plate using the recorded information of the photons The number of output photons passing through is counted, and the value of the objective function is calculated by matching the counted output photons with luminance.

In the design method, the terminating condition is (1) when the number of times of repeating the execution of the second and third steps exceeds a predetermined number of times; (2) when the number of calculations of the simplex search is greater than a predetermined maximum number of calculations during the execution of the simplex search algorithm; (3) the sum of the distances between the mean of the remaining points except for the largest vertex of the function value in the simplex and the remaining vertices is less than or equal to a specified error; (4) the sum of the distances from the vertex with the smallest value to the remaining vertices in the simplex is less than or equal to the specified error; And (5) at least any one or a combination thereof when the same minimum value repeats for a predetermined time or more in every iteration of the simplex search algorithm.

Further, the simplex search algorithm can minimize the objective function value by performing symmetry, expansion, contraction, and reduction operations, which are operations related to the simplex movement, for the simplex defined by the existing coefficients of the characteristic function. Calculating a new simplex, wherein the symmetry, expansion and contraction operations create one symmetry point, one expansion point and one contraction point at the new position, and each vertex of d + 1 (where d is a natural number) of the existing simplex The newly created symmetry point, the expansion point, and the contraction point are added to the remaining d vertices from which the point having the highest function value is removed, and a new symmetry, expansion, and contraction simplex is respectively created by d + 1 points, and the reduction operation is performed. Creates a new d minus vertex and adds it to the point with the lowest function value among the vertices of the existing simplex. Phase, characterized in that operation to create a cloud reduction simplex; And modifying the coefficient of the existing characteristic function using the newly calculated vertex of the simplex.

The ray tracing technique is preferably performed by a program module implemented using Monte Carlo light tracing technique or photon mapping technique.

According to the present invention, BLU simulation is performed on a specific light reflection pattern using Monte Carlo light tracking technique, and when the simulation result does not satisfy a desired condition, the light reflection pattern is modified by using the simplex search method. By repeating the process of BLU simulation, the optimized light reflection pattern can be found quickly, accurately and very efficiently.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. For convenience of explanation, a case where the ink dot pattern is used as the light reflection pattern of the light guide plate will be exemplarily described.

The present invention introduces a function for defining the characteristics of the light reflection pattern in calculating the positional size of the light reflection pattern, for example, the concept of the density function representing the density according to the distance from the light source of the ink dot, and optimizes the simplex search method. It is used as a means to find the density function. Based on a simulator for calculating the luminance occurring on the front surface of the light guide plate with a given light reflection pattern, an optimal printing pattern is automatically calculated to emit the most uniform luminance to the front surface of the light guide plate. Designing the light reflection pattern is one of the most important parts of the design of the BLU. In the present invention, Nelder-Mead's Simplex Search algorithm is used among direct search techniques to design an optimal ink pattern.

<1> Print pattern design of ink dot

Patterns printed on the bottom of the light guide plate with more than hundreds of thousands of small diffusion ink dots of different sizes are mainly circular and hexagonal. The dots are arranged in a grid and a honeycomb pattern. Honeycomb) In the grid dot arrangement method, there is a 'circular-dot-on-grid' type (FIG. 3 (a)) which positions the center of the circular dot 120a at the center of a square belonging to the grid. The honeycomb dot placement method includes a 'circular-dot-on-honeycomb' type (FIG. 3 (b)) which positions the center of the circular dot 120b at the center of the regular hexagon belonging to the honeycomb shape. There is a hexagonal-dot-on-honeycomb type (FIG. 3 (c)) which positions the center of the regular hexagon dot 120c at the center of the regular hexagon.

Variables required for the dot placement method include the maximum size, minimum size, minimum distance between dots, the size of the gap to be printed and printed on the edges of the light guide plate, and the size of the dot according to the position of the light guide plate. Function. In general, the dots at the same horizontal distance from the lamp do not all have the same size, but make the size slightly larger toward both edges. Since the light emitted from the lamp reaches the edge relatively, the size of the dot is increased to correct the scattering of the light, so that a little more light energy is emitted to the outside. In the present invention, to simplify the problem, it is assumed that dots at the same distance from the lamp are made the same size. To change the size of dots at the same horizontal distance, the method described below can be easily extended by expanding the method described below.

The role of the ink dot scatters the light reaching the dot, and serves to emit the scattered light to the outside of the light guide plate in a direction smaller than the critical angle of total reflection occurring inside the light guide plate among the scattered light. In the light guide plate, the farther side of the light source is weaker than the close side, so the density of ink dots per unit area is increased to increase the scattering of light to make the luminance emitted to the front of the light guide plate irrespective of the distance from the light source. Therefore, in order to derive uniform plane light irrespective of the distance from the light source, the density of the ink dot according to the distance from the light source must be calculated first.

<2> Relationship between density function of ink dot printing pattern and size of ink dot

In the present invention, since the dots at the same distance from the lamp is made the same size, the density of the ink dot can be defined as a one-dimensional density function d (x) as shown in FIG. Here, the variable x represents the distance from the light sources 114a and 114b, and the density function becomes symmetric when the light sources are disposed on both sides of the light guide plate. The method of numerically approximating the size of the dot according to the density function given in each printing pattern of the ink dot is as follows.

In the 'circular-dot-on-grid' ink dot arrangement model of FIG. 3 (a), all dots are arranged in a grid form with the same size square, and each dot is circular in the square. The printed pattern. In this pattern, a line of squares parallel to the light source as in FIG. 4 is called a strip S _i . Given the required density function and other variables for the dot placement method, these data are used to first calculate the length w of the sides of the square belonging to the grid and the number N of squares belonging to one strip S _i . Since the dots arranged in the square belonging to one strip S _i are the same size, given the density function d (x) , the radius r _i of the dot belonging to each strip S _i can be calculated. The method of calculating the radius r _i from the density function d (x) is shown in FIG. 5. The density function is a continuous function, and the radius of the squares belonging to the strip to the total area of each non-continuous is determined in position, the central axis of density on the strip S _i in the x _i d (x _i) is a strip S _i The ratio of the area of the arranged dots is shown. Accordingly,

...... (1)

...... (One)

If you calculate the radius

...... (2)

...... (2)

to be.

The circular-dot-on-honeycomb pattern shown in FIG. 3 (b) is a pattern in which regular hexagons are arranged in a honeycomb shape, and one circular dot is printed inside each regular hexagon. In this pattern, the strip has a square whose width is equal to the width of one regular hexagon, as shown in Fig. 6, and whose length is equal to the height of the print area. One strip has a complete regular hexagon or a half regular hexagon, and if a full regular hexagon is placed at both ends of the strip, it is called a symmetric strip, and the arrangement of the regular hexagon in the form of a symmetric strip is called symmetry. ) Reference). In addition, when a complete regular hexagon is arranged at one end of the strip and two half hexagons are attached at the other end, it is called an asymmetric strip, and the arrangement of the regular hexagon in the form of an asymmetric strip is called an asymmetrical arrangement (see FIG. 6 (b)).

The method of calculating the dot size from a given density function in the circular-dot-on-honeycomb ink dot arrangement model is a half strip unit (H) divided into one strip in half as shown in FIG. Calculate with To find the size of the radius of each circle belonging to the half strip from the density function, as in the circular-dot-on-grid pattern, the density function d (x at half strip H _i where the central axis is located at distance x _i from the light source. _i) is a group represented by the ratio of the dots to be printed on the inside of the _i H H _i the total area of the half-strip area. H _i half of the strip, as shown in Figure 8, the size of the radius r _i-1, the size of the circle C _i-1 and the radius of the halves contains two types of sources of the circle C _i r _i. Let n _i and the area A _i be the number of semicircles C _i included in the half strip H _i . Therefore, as shown in FIGS. 7 and 8, the density function d (x _i ) in the half strip H _i having the position x _i , the width w and the length l is as follows.

...... (3)

...... (3)

The half strip H _i of an asymmetric strip contains the same number of two types of half circles, so n _{i =} n _i-1 , and the half strips of a symmetric strip show the number of two types of half circles as follows: .

...... (4)

...... (4)

Therefore, in the asymmetrical arrangement, N = n _i ,

...... (5)

...... (5)

to be. Since C ₀ is actually printed from C ₁ without printing, assuming that A ₀ = A ₁ , the width of the dot is A _i = π r _i ^2, so dots from each strip are sequentially from r ₁ from the following recursion: You can derive the radius of.

...... (6)

(6)

From the above equation, r _i can be expressed as the following equation.

...... (7)

...... (7)

In a symmetrical arrangement of the 'Circular-dot-on-honeycomb' print pattern, the radius of each circle printed on each rectangular hexagon can be calculated in a similar way.

Next, the 'hexagonal-dot-on-honeycomb' shape illustrated in FIG. 3 (c) refers to a pattern in which an area where one dot is printed is a regular hexagon and a dot shape is also a regular hexagon therein. The method for calculating the dot size of a regular hexagon in this pattern is similar to the method for generating a circular-dot-on-honeycomb pattern. Since the dot printed inside the regular hexagon becomes a regular hexagon instead of a circle, the variable A _i representing the area of the circle in Equation (5) for calculating the size of the circle in the circular-dot-on-honeycomb arrangement is shown in FIG. The area of the regular hexagon

...... (8)

...... (8)

We can calculate the size r _i by substituting for.

When the density function is determined for the specific print pattern, it can be seen that the dot size of the print pattern can be determined from the density function.

<3> Modeling and Simulation of BLU

In the design of the BLU, the ray tracing technique (Ray-) is a means for effectively calculating the light efficiency indicating whether it is emitted to the front of the BLU, and for predicting various characteristics such as luminance uniformity occurring on the front of the BLU. BLU simulator based on tracing method is used. Monte Carlo ray tracing or photon mapping is a ray tracing technique that handles phenomena such as light diffusion. Therefore, it is desirable to make a BLU simulator program module based on this. Photon Path-Tracing is one of the Monte Carlo techniques in which Path-Tracing is applied as a ray tracing technique. In this technique, one ray is regarded as one photon and the photon Tracks in only one direction.

10 is a diagram for explaining a BLU modeling and ray tracing technique. The BLU simulator has the function of modeling the shape of the BLU by defining data on the shape of the BLU core parts. The data to be set for BLU modeling includes, for example, the size of the light guide plate 112, the shape and number of the lamps 114a and 114b, the geometric structure of the cross section of the lamp reflector 116a and 116b, and the printing of the diffusion material. Pattern 120, reflective sheet 124, and the like. In addition, the BLU simulator is configured to consider the material properties such as reflection coefficient, diffuse reflection coefficient, and absorption coefficient of each component.

, 흡수계수

와 확산계수

를 측정하여 이들 계수들에 따라서 광자가 특정 부품의 면에 맞았을 때, 어느 방향으로 광자가 경로를 바꿀 지를 난수를 사용하여 결정하도록 한다. 예를 들면, 난수

가 조건

을 만족하는 경우에는 광자를 전반사시키고, 조건

을 만족하는 경우에는 광자를 난반사시키고, 조건

을 만족시키는 경우에는 광자를 흡수시킨다. The BLU simulator also has the ability to predict luminance measurements and uniformity for modeled BLUs. That is, after the shape of the BLU is modeled, the BLU simulator tracks the light generated from the internal light sources 114a and 114b using a ray tracing technique and measures the light reaching the front surface of the light guide plate to measure the diffusion sheet 122 and the prism. It is possible to predict the uniformity of the plane light exiting the sheet 118 toward the LCD panel 126. The stochastic model for applying the Monte Carlo technique is applied using the Russian Roulette using the reflection coefficient among the material properties of each component of the BLU as follows. Total reflection coefficient as reflection coefficient of each part

, Absorption coefficient

And diffusion coefficient

We then measure by using a random number to determine in which direction the photon will change its path when the photon hits the face of a particular component based on these coefficients. For example, random numbers

Autumn conditions

If it satisfies the total photon, the condition

If it satisfies the condition, the photon is diffusely reflected, and the condition

If it satisfies the photon absorption.

Finally, the location of the light guide plate from which the photons exit, the direction of the photons, etc., are recorded. Based on this, the luminance distribution of photons on the front surface of the light guide plate is calculated by the following method. The luminance distribution virtually places a grid plane 130 taking the shape of a grid pattern square on the front surface of the BLU as shown in FIG. 11, and counts photons exiting from each grid pattern and stores the data in a two-dimensional array. The data can be represented as a two-dimensional or three-dimensional graph, as shown in the figure. The three-dimensional graph represents the number of photons from each lattice position, indicating surface luminance, and the two-dimensional graph shows lamps 114a and 114b. It is a graph that calculates the average value by integrating the number of photons of all gratings located at the same distance from). These graphs show the uniformity of the luminance of the plane light generated by the BLU, and the flatter the graph, the more uniform the positional brightness.

As mentioned above, since luminance uniformity is very sensitive to the printed pattern, it is essential for BLU simulation function to model the BLU that automatically calculates the efficient light reflection pattern and outputs the optimized surface luminance in order to generate uniform surface luminance. It is one of the functions to be provided.

<4> simplex search algorithm

The problem of calculating the optimal ink dot pattern is represented by a nonlinear, discontinuous, and non-differential function. And the direct search method (Direct Search Method) is often used to calculate the optimal point of these functions. In the present invention, a simplex search algorithm, which is widely applied among direct search techniques, is used to design an optimal ink pattern.

에 대하여, 목적함수 f(x)를 최대화(혹은 최소화) 시키는 것이다. BLU의 광선추적 기법을 이용한 시뮬레이션 모델에서 목적함수는 여러 가지가 있을 수 있으나, 본 발명에서는 BLU가 전면으로 출사하는 휘도의 균일도를 목적함수로 사용하고, 그 목적함수 f(x)를 최소화하고자 한다. 휘도의 균일도를 표시하는 목적함수에서 사용하는 변수들은 위에서 설명한 BLU의 구조를 나타내는 변수들을 사용할 수 있다. BLU 구조에 관한 여러 가지 변수들 중에서 도광판의 크기 및 투과율, CCFL의 크기 및 휘도 등은 미리 정해져 있으나, 램프 반사판의 구조, 확산물질의 인쇄패턴 즉, 잉크도트 패턴은 자유롭게 변경할 수 있는 변수들이다. 따라서 최적화 작업을 통해서는 주어진 BLU 구조에서 가장 효율적인 인쇄패턴이나, 반사판의 구조를 표현하는 변수를 찾게 된다. Mathematically, the optimization problem is a vector of d variables

For maximizing (or minimizing) the objective function f (x). In the simulation model using the BLU ray tracing technique, there can be various objective functions, but in the present invention, the uniformity of the luminance emitted by the BLU to the front is used as the objective function, and the objective function f (x) is minimized. . Variables used in the objective function indicating the uniformity of luminance may use variables representing the structure of the BLU described above. Although the size and transmittance of the light guide plate, the size and luminance of the CCFL are predetermined among various variables regarding the BLU structure, the structure of the lamp reflector and the printing pattern of the diffusion material, that is, the ink dot pattern, are variables that can be freely changed. Therefore, the optimization work finds the most efficient printing pattern in the given BLU structure or the variable representing the structure of the reflector.

Among the variable variables representing the structure of the BLU, the most influential variable on the efficiency of the BLU is the light reflection pattern of the BLU, that is, the printing pattern of the ink dot. Hereinafter, an example of an optimization method for calculating a print pattern of an ink dot of a BLU that provides the best uniformity of luminance emitted to the front of the BLU will be described. The optimization method for the other variables may be optimized by applying the same method as the optimization technique for the BLU printing pattern described below.

In the calculation method of the objective function f representing the uniformity of the luminance emitted to the front of the BLU, it is preferable to use an optimization technique that can calculate the minimum value of the objective function f with as few calculations as possible. Such an optimization technique includes a direct search technique using only a function value and not requiring a gradient value. The most popular direct search technique to date is the Simplex Search Method developed by J. Nelder and R. Mead in 1965. For reference, this method is distinguished from the simplex search technique used in the linear programming method. In the d -dimension, the simplex is a polytope made of non-degenerate d + 1 vertices. . For example, in two-dimensional simplex is triangular, in three-dimensional simplex is tetrahedron (Tetrahedron). The Nelder-Mead technique creates a simplex with d + 1 vertices in d -dimensional space, and performs reflection, expansion, contraction, and shrinking operations on the simplex. The symmetry, expansion, and contraction operations create a point at a new location, remove the point with the highest function value among the vertices of the simplex, and add d + 1 points with the newly created point at the remaining d vertices. Create a simplex In the reduction operation, a new d vertex is created, and the new simplex is created by adding the one with the lowest function value at the vertex of the existing simplex. Iteratively performs four operations on this new simplex, extending the simplex so that the newly created vertex is close to the lowest point, and causing the simplex to contract in the direction far from the lowest point so that the vertices of the simplex are in the optimal solution. Move closer and closer The application of the Nelder-Mead simplex search algorithm to the present invention will be described later.

<5> Optimization Design Method of BLU

12 is a flow chart showing the overall overview of the BLU optimization design method according to the present invention. The design of the optimized BLU is outlined as follows. This design method is implemented by a computer program. The BLU designer can then run the program on a computer to design an optimized BLU. The algorithm of the BLU optimization design program is largely based on the initial design process (S10 ~ S16) that gives initial values for the design elements of the BLU structure, and the optimization of the BLU which repeatedly searches for the improved BLU structure starting from the initial designed BLU structure. It consists of a structure calculation process (S18-S24). The initial BLU design process assigns specific values as initial values for all design elements. However, the printing pattern of the ink dot to be treated as a variable design element is represented by the density function and initialized by giving an initial value to the coefficient of the density function. Fix the remaining design elements by assigning them constant values. In the process of calculating the optimized structure, the final design of the optimized BLU structure is determined by repeating the task of finding an improvement of the density function of the ink dot, which is a variable design element, by using the BLU simulation based on the BLU structure light tracking technique and the simplex search method.

Initial design work for the BLU, which is performed as part of the basic element setting work for the BLU design (step S10), involves setting the physical specifications of the BLU and the attribute values of the optical component (step S14). Examples of physical specifications of the BLU include the size of the BLU and the light guide plate, the number of lamps and the placement position. The attribute values of the optical components include the light transmittance of the light guide plate, the reflectance of the ink dot pattern disposed on the bottom surface of the light guide plate, and the light efficiency. have. The luminance uniformity of the BLU can be expressed as a function of various variables used to model the structure of the BLU. The variables used herein typically include the size and transmittance of the light guide plate, the size and brightness of the lamp, the structure of the reflector, the ink dot pattern, and the like.

One of the most important factors in manufacturing a BLU is the uniformity of plane light generated on the front of the light guide plate. The uniformity of the planar light varies depending on the size of the dot printed on the lower part of the light guide plate, and the size of the dot is controlled by the density function described above. For example, FIG. 13 is a graph showing the uniformity of plane light according to the density function for adjusting the size of the dot. In this example, it is assumed that the light sources are mounted on both sides of the light guide plate, so that the printing pattern printed on the light guide plate is symmetrically on both sides with respect to the center of the light guide plate. In FIG. 13, the left side shows the density function, and the right side shows the luminance generated to the entire surface of the light guide plate obtained by ray tracing simulation when the dot size is determined by each density function. In Fig. 13A, the constant density function is used as an example in which dots of the same size are printed regardless of the distance from the light source. In this case, as shown in the figure on the right, a strong luminance is generated near the light source, and a relatively weak luminance is generated in the middle portion of the light guide plate far from the light source. FIG. 13 (b) uses a density function that increases the size of the dot toward the middle portion of the light guide plate. As compared with FIG. 13 (a), a relatively uniform luminance can be obtained. In addition, in FIG. 13 (c), the middle portion of the light guide plate uses a density function such that the dots have the same size, and the luminance is more uniform than that of FIGS. 13 (a) and 13 (b). This slightly higher luminance can be obtained. As shown in the experimental results, it can be seen that the variation of the density function of the ink dot causes a large change in the uniformity of the plane light generated by the BLU. Therefore, it can be seen that the calculation of the density function for optimizing the printing pattern of the light guide plate to generate uniform plane light to the front of the light guide plate is very important in developing the BLU.

To derive the optimal design for the ink dot pattern, treat only the ink dot pattern, which is the design element that is the target of the optimal design, as a variable design element, and all other design elements are fixed to a specific value in advance in the initial design of the BLU. It needs to be treated as a fixed design element. Therefore, the density of the ink dot according to the distance from the light source disposed on the side of the light guide plate can be set as a function, and the coefficient of the density function can be used as a variable design element. This is because, once the coefficient of the density function of the ink dot is determined, the density function is determined, and the size of the ink dot can be easily calculated from the above-described relationship between the density function and the radius of the dot.

이다. 따라서 Nelder-Mead 알고리즘에서 사용되는 심플렉스는 이들 변수를 나타내는 12차원 공간에서 13개의 점으로 구성이 된다. 또한, Nelder-Mead 알고리즘에서 사용되는 목적함수는 다음과 같이 정의한다. 밀도함수 D(x) 에 의해 계산된 도트패턴에 따라 BLU의 시뮬레이션 결과로 구해지는 그리드 카운트(도 14 참조)를 이용하여, 목적함수 f 는 아래 식 (9)로 나타내지는 그리드 카운트들의 표준편차 혹은 식 (10)으로 나타내지는 그리드 카운트 c_ij 의 최대값과 최소값의 비율로 표시된다. 참고로, 그리드 카운트 값은 광선추적 기법을 이용한 시뮬레이션을 통하여 도광판 전면에 그리드로 분할된 작은 사각형을 통과한 광자의 개수를 나타낸다. Further, the objective function for design optimization of the BLU and the limit value for the objective function are also set as part of the basic element setting of the BLU design (step S16). The Nelder-Mead algorithm is used as a simplex search algorithm for optimizing the ink dot pattern. The variable to be optimized through this algorithm is the density function D (x) , where 12 coefficients are used in the three three-dimensional polynomials (see Equation 12) used to represent the density function.

to be. Thus, the simplex used in the Nelder-Mead algorithm consists of 13 points in 12-dimensional space representing these variables. In addition, the objective function used in the Nelder-Mead algorithm is defined as follows. Using the grid counts (see Fig. 14) obtained as a result of simulation of the BLU according to the dot pattern calculated by the density function D (x) , the objective function f is the standard deviation of the grid counts represented by the following equation (9) or Grid count c _ij represented by equation (10) It is expressed as the ratio of the maximum value and the minimum value of. For reference, the grid count value represents the number of photons passing through a small square divided into a grid on the front of the LGP through simulation using a ray tracing technique.

...... (9)
여기서, m과 n은 그리드카운트 c_ij의 행(i)과 열(j)의 개수의 최대값을 각각 나타낸다.

(9)
Here, m and n represent the maximum values of the number of rows i and columns j of grid count c _ij , respectively.

...... (10)

...... (10)

Since the simplex search algorithm is a method of calculating the simplex to move closer to the optimum point by repeatedly moving the simplex, the algorithm must be terminated when various conditions such as a close to the minimum point are satisfied. In the present invention, the following limit value or limit value is set to terminate the algorithm, and when the limit value or the limit value range is satisfied, the algorithm is terminated to prevent the algorithm from repeating indefinitely and find a predetermined value. To do that.

(a) When the number of iterations of the algorithm is greater than the number of iterations specified

(b) When the number of calculations of the Simplex Search function is greater than the maximum number of calculations specified

(c) The sum of the distances from the average of the remaining points, except for the largest vertex, to the remaining vertices in the simplex is less than or equal to the specified error.

(d) The sum of the distances between the vertices with the smallest function values and the remaining vertices in the simplex is less than or equal to the specified error.

(e) When the same minimum value repeats for a certain number of times in each iteration of the algorithm

After the basic elements for the BLU design are set as described above, the process of calculating the BLU optimization structure (S18 to S24) for calculating the optimal design condition for the ink dot pattern which is a variable design element is performed. In this process, a BLU simulation using Monte Carlo light tracing technique is performed on the BLU structure determined through the initial design (S14) or the density function conversion (S22) to calculate the objective function value (S18). Check whether the calculated objective function satisfies the preset limit value (S20), and if it is not satisfied, modify the value of the variable design element of the BLU to change the structure of the BLU to improve the efficiency of the BLU (S22). ) And the above-described series of operations (S18, S20, S22) are repeated again for the modified BLU. If the objective function value meets the limit value through this iteration, the BLU design plan including the density function at that time is determined as the optimal design plan. That is, while loops from step S18 to step S22 are repeatedly performed, if the objective function value satisfies a preset limit condition, the loop is executed and the BLU design scheme at that time is determined as an optimized BLU structure. A more specific procedure for this task is shown in the flowchart of FIG. 15.

The optimized density function of the ink dot is obtained by simplex search. In order to perform a simplex search, an initial density function and an initial simplex of an ink dot are first set (steps S30 and S32). In order to express the density function, a function for expressing an accurate density is required, but in the present invention, as shown in FIG. 16, the maximum value of three three-dimensional polynomial functions is used as the density function. In other words, the density function D (x) is

...... (11)

...... (11)

And each three-dimensional polynomial is

...... (12)

...... (12)

to be. Here, the variable x represents the distance from the light source, and a ⁱ ₃ , a ⁱ ₂ , a ⁱ ₁ , a ⁱ ₀ are coefficients of the three-dimensional polynomial. The density function defined in this way can generate a wide variety of density functions, and does not require a very complex function calculation. However, other density functions can be defined and used to calculate more accurate densities.

, σ )를 초기화하여야 한다. 이러한 심플렉스 꼭지점의 초기값과 계수의 초기값에 따라 최적의 해로 빨리 수렴할 수도 있고 늦게 수렴할 수도 있다. 이러한 변수나 계수들의 초기값들은 사용자가 임의로 변경 가능하며, 그 변수의 변화 범위 또한 사용자가 임의로 변경 가능하다. 목적함수의 d 개의 변수에 대한 초기값이 설정이 되면, 이 초기값으로 심플렉스의 한 개의 꼭지점으로 사용한다. 이 꼭지점을 기준점으로 하여 초기 심플렉스를 구성하는 나머지 d 개의 꼭지점을 초기화한다. 기준점을 중심으로 d 차원의 공간에서 한 축으로 일정한 범위를 이동한 위치에 새로운 꼭지점을 만들면 모두 d 개의 새로운 꼭지점을 만들 수 있다. 이 때, 하나의 축으로 이동한 새로운 꼭지점의 위치는 그 축을 나타내는 변수의 초기값으로부터 변수값의 범위를 심플렉스탐색 알고리즘의 최대 가능한 반복횟수로 나눈 값을 더하고 뺀 두 개의 값 중 더 좋은 결과를 보이는 위치를 선택한다. 여기서, 심플렉스 이동에 관련된 대칭, 확장, 수축, 축소 등의 네 개의 계수는 일반적으로 많이 적용되는 ρ =1, χ =2,

=1/2, σ =1/2를 사용한다. 하지만 이 값은 절대적이지 않고 상황에 따라 변형 가능하며, 여러 번의 실험을 통하여 상대적으로 더욱 빠르고, 보다 정확한 해로 수렴할 수 있는 새로운 값을 사용할 수 있다. In addition to the initialization of the density function, the simplex initialization is also performed in preparation for the optimization of the printing pattern using the simplex search. In the simplex search algorithm such as Nelder-Mead, when optimizing the objective function with d variables, the simplex made of d +1 vertices moves in the d- dimensional space and approaches the optimal solution. To apply this optimization, we first move the coefficients of motion (ρ, χ,) for the d +1 vertices and four movement operations (symmetry, expansion, contraction, and reduction) that make up the simplex.

, σ) should be initialized. Depending on the initial value of the simplex vertex and the initial value of the coefficient, the convergence may be quick or late. Initial values of these variables or coefficients can be arbitrarily changed by the user, and the change range of the variable can be arbitrarily changed by the user. Once the initial values for the d variables of the objective function are set, they are used as the vertices of the simplex. Using these vertices as a reference point, we initialize the remaining d vertices that make up the initial simplex. If you create a new vertex at a position shifted a certain range from one space in the d- dimensional space around the reference point, you can create d new vertices. At this time, the position of the new vertex moved to one axis is the better of the two values minus the initial value of the variable representing that axis divided by the range of variable values divided by the maximum possible number of iterations of the simplex search algorithm. Select the visible location. Here, the four coefficients related to the simplex movement, symmetry, expansion, contraction and contraction, are generally applied to ρ = 1, χ = 2,

Use = 1/2 and σ = 1/2. However, this value is not absolute and can be modified according to circumstances, and new experiments can be used to converge to a relatively faster and more accurate solution through several experiments.

After such initialization, in order to obtain the structure of the BLU having the optimum luminance uniformity, the operation of finding the optimum value of the coefficient of the density function that defines the ink dot pattern (steps S34 to S44) is repeatedly performed. The calculation of the optimal density function requires a lot of calculations, and the numerical optimization technique is required to perform these calculations automatically.

First, the size of the individual ink dot is calculated based on the density function given through initialization or loop execution (step S34). For calculating the size of an individual ink dot from the density function, see the above-described relationship between density function and ink dot size.

Once the size of the individual ink dots, the only variable design element, is determined, the structure of the BLU is specified. The BLU is executed using the Monte Carlo light tracking technique described above with respect to the BLU (step S36), and the value of the objective function is calculated using the information obtained through the BLU simulation (step S38). Then, it is determined whether the calculated objective function value is less than or equal to a preset threshold value (step S40). If the calculated objective function value does not fall below the preset limit, the simplex search method is used to find the coefficient of the density function of the ink dot (the vertex of the new simplex) that provides improved brightness uniformity. The procedure of modifying the BLU structure applying the ink dot pattern calculated by the found new density function is performed (steps S42 and S44).

에 대하여 목적함수를 계산한다(S52). 주어지는 심플렉스는 최초에는 초기화된 심플렉스가 될 것이지만 아래에서 설명하는 '정렬, 대칭, 확장, 수축, 축소' 루프를 반복하면서 계속 바뀌어 나간다. Figure 17 illustrates in more detail the process of finding a new simplex required for the conversion of the density function by Nelder-Mead's simplex search technique. A simplex initialization is first performed for the simplex search (S50), and then d + 1 vertices and a center point of the simplex for a given simplex

Compute the objective function for (S52). The given simplex will initially be an initialized simplex but will continue to change as we repeat the 'align, symmetric, expand, contract, shrink' loop described below.

(1) Ordering

The objective function values for the d + 1 vertices are aligned to satisfy the following (S54).

...... (13)

...... (13)

(2) Reflection

Calculate the reflection vertex by the following equation,

...... (14)

...... (14)

을 계산한다(S56). 여기서,

는

을 제외한 나머지 d 개 꼭지점의 중심점

이다. 대칭점의 함수값이 조건

을 만족하면 대칭점

을 새로운 꼭지점으로 받아들이고, 그렇지 않은 경우에는 다음 단계를 수행한다(S58, S60). Function Values for Symmetry Points

Calculate (S56). here,

Is

Center point of d vertices except

to be. Function value of symmetry point

If you satisfy the symmetry point

Accept as a new vertex, otherwise perform the following steps (S58, S60).

(3) Expansion

을 만족하면(S62), 다음 식에 의하여 확장점(Expansion Vertex)

를 계산하고,if

If (S62), the expansion point (Expansion Vertex) by the following equation

And calculate

...... (15)

...... (15)

에 대한 함수값

를 계산한다(S64). 만약

이면(S66), 확장점

를 새로운 꼭지점으로 받아들이고(S68), 그렇지 않은 경우에는 대칭점

를 새로운 꼭지점으로 받아들인다(S60).Extension point

Function value for

Calculate (S64). if

Back side (S66), extension point

Is taken as a new vertex (S68), otherwise the symmetry point

Accept as a new vertex (S60).

(4) Contraction

을 만족하면,

과

중에서 작은 함수값을 가지는 꼭지점과 중심점

사이에 다음과 같은 수축연산을 수행한다. if

If you satisfy

and

Vertices and center points with small function values

Perform the following shrinkage operation in between.

   (4.1) Outside Contraction

을 만족하면(S70), 다음과 같은 수축점(Contraction Vertex)

를 계산하고,if

If satisfied (S70), the following contraction (Contraction Vertex)

Calculate,

...... (16)

...... (16)

에 대한 함수값

를 계산한다(S72). 만약

이면(S74), 수축점

를 새로운 꼭지점으로 받아들이고, 그렇지 않은 경우에는 아래의 '(5) 축소(Shrinking)'(S84)를 수행한다.Shrink point

Function value for

Calculate (S72). if

Back surface (S74), shrink point

Accept as a new vertex, otherwise perform '(5) Shrinking' (S84) below.

   (4.2) Inside Contraction

을 만족하면(S70), 다음과 같은 수축점(Contraction Vertex)

를 계산하고,if

If satisfied (S70), the following contraction (Contraction Vertex)

Calculate,

...... (17)

...... (17)

에 대한 함수값

를 계산한다(S78). 만약

이면(S80), 수축점

를 새로운 꼭지점으로 받아들이고(S82), 그렇지 않은 경우에는 아래의 '(5) 축소(Shrinking)'(S84)를 수행한다.Shrink point

Function value for

Calculate (S78). if

Back surface (S80), shrink point

Is taken as a new vertex (S82), otherwise it performs '(5) Shrinking' (S84) below.

(5) Shrinking

For all existing vertices, new vertices are calculated using the following equation (S84).

...... (18)

...... (18)

과 새로 만들어진 꼭지점

을 다음 단계에서 심플렉스의 새로운 꼭지점으로 사용한다.Existing vertices

And newly created vertices

As the new vertex of the simplex in the next step.

, σ 에는 임의의 값을 지정할 수 있으나, 대표적으로 많이 적용되는 값은 ρ =1, χ =2,

=1/2, σ =1/2이다. 위의 Nelder-Mead 알고리즘을 종료하는 조건에는 심플렉스의 크기가 너무 작거나, 심플렉스의 꼭지점의 함수값에 대한 변화가 너무 적거나 하는 등이 있다. 또한 최초의 심플렉스의 크기와 모양을 정하는 것에 는 시행착오나 여러 실험을 통하여 정할 수 있다. In the above process, it is checked whether the convergence conditions are satisfied with respect to the newly obtained symmetry point, extension point, external contraction point, internal contraction point, and reduction point, and if the convergence condition is not satisfied, the process returns to step S52 and the existing simplex points are newly obtained. Substitute the simplex defined by and repeat the procedure of recalculating the objective function for the vertex of the new simplex (S86, S52 ~ 84). Above, the constants ρ, χ,

, σ can be any value, but the most common values are ρ = 1, χ = 2,

= 1/2 and σ = 1/2. The conditions for terminating the above Nelder-Mead algorithm may include the size of the simplex being too small or the change in the function value of the vertex of the simplex being too small. In addition, the size and shape of the first simplex can be determined by trial and error or by various experiments.

After the simplex correction and the conversion of the density function according to the above operation are performed, the ink dot pattern according to the converted density function is again determined (step S36). In addition, BLU simulation using Monte Carlo light tracking technique is executed for the BLU design according to the new ink dot pattern, and the objective function value is calculated again from the simulation results. Furthermore, the process of comparing the calculated value of the objective function with the threshold again is repeated (steps S36 to S40).

If the loops of steps S34 to S44 that perform the BLU simulation and the density function correction as described above continue to run, one day, the objective function value will converge below the preset limit value. When the threshold value is satisfied, the execution of the above loop is terminated, and the density function of the ink dot at that time is determined to be the optimum density function to be obtained (step S46).

<6> Experimental Results

The present inventors actually implemented the above-described optimization technique in a program to actually design the BLU. The BLU used was a symmetrical type with a cold cathode fluorescent lamp (CCFL) on each side of the light guide plate, which was a 288x351x8 (mm) sized BLU used in LCD monitors of notebook computers. In order to express the density function of the diffusion ink pattern printed on the bottom of the light guide plate of the BLU, three-dimensional polynomials as shown in Eq. (12) were used, and the standard deviation as in Eq. (9) was used as the objective function. In this experiment, a 100x100 grid was used, and a total of 400,000 rays were generated for the Monte Carlo ray tracing technique. 18 (a) and 18 (b) are graphs showing luminance generated on the front surface of the light guide plate in two and one dimensions, respectively. Here, the one-dimensional display function displays the result of integrating the luminance generated in the two-dimensional luminance at the same distance from the lamp. Table 1 shows the coefficients and standard deviation values of three density functions that produce optimal uniform brightness.

[Table 1] Density coefficient and optimal function value for optimal printing pattern

i Coefficient of cubic polynomial Optimal Function Value (Standard Deviation)

One 0.422089 0,622510 0.395499 1.596931  20.217103 2 0.480890 0.677667 0.483086 1.367568 3 0.420862 2.224944 0.558160 2.456686

The process of changing the density function through the simplex search algorithm, and the process of changing the luminance generated in front of the light guide plate as a result of performing the ray tracing simulation using the pattern of the ink dot calculated by the density function is performed as follows. . Here, only one cubic polynomial of the three cubic polynomials used in Eq. (12) was used to simplify the density function. In addition, D (x) = 1 was used as the initial value of the density function, and the standard deviation as shown in Eq. (9) was used as the objective function to compare the uniformity during the optimization process. In this experiment, a 100x100 grid was used and a total of 200,000 rays were generated for the Monte Carlo ray tracing technique. In this experiment, the optimal ink dot printing pattern was calculated through the repetition of 27 simplex search algorithms. Table 2 shows the coefficients of the density function used in the simplex search algorithm each time and the standard deviation, which is the value of the target function calculated through the ray tracing technique. Here, the case where there is little change in the value of the objective function while repeatedly performed is omitted.

[Table 2] Density function coefficients and objective function values during simplex search algorithm

Number of Iterations in Simplex Search Coefficient of cubic polynomial Objective function value (standard deviation)

0 0.0 0.0 0.0 1.0 102.313014 One 0.0 3.0 0.0 1.0 22.478581 5 0.843750 0.843750 1.968750 1.093750 22.462243 6 0.421875 0.421875 -0.515625 -1.953125 22.298491 10 0.316406 1.066406 -0.386719 -1.214844 22.094292 16 0.012531 1.090656 -0.342464 -1.524487 21.891181 18 0.052468 1.306375 -0.488688 -1.393706 21.752102 22 -0.044362 1.32816 -0.365656 -1.418618 21.568177

FIG. 19 is a graph illustrating a process of changing a density function during a simplex search algorithm, and each graph is a three-dimensional polynomial represented by coefficients calculated during a simplex search in Table 2. FIG. Here, the value on the x-axis represents the distance from the light source, and the interval is normalized to [0, 10.0]. The reason for not specifying the normalized interval as [0.0, 1.0] is to reduce the occurrence of errors in the density function value by preventing the value of the density function from being too small.

FIG. 20 illustrates a process of changing a two-dimensional luminance graph emitted to the front of the light guide plate as the density function is changed during the simplex search algorithm. Each of the graphs (a) to (h) is a luminance graph emitted to the front of the light guide plate when simulated using eight density functions represented by coefficients calculated during the simplex search in Table 2 repeatedly. Graph (a) is a case where a constant density function is used. Since the size of the ink dot printed on the lower part of the light guide plate is constant regardless of the distance from the light source, relatively high luminance is generated near the light source. The graph (b) has a relatively small luminance uniformity compared to (a), but there are many differences in the luminance emitted from a portion close to the light source and the luminance emitted from the center of the light guide plate far from the light source. In the graph (c), the luminance of the central portion of the light guide plate is higher than that of the graph (b) in order to reduce the uniformity of the luminance occurring near and far from the light source. Graphs (d) to (h) show a process in which the luminance changes uniformly as the luminance at the center decreases. It can be seen that the standard deviation value of luminance, the objective function value in this process, becomes smaller as shown in Table 2.

The foregoing description focuses on the optimal design method of the ink dot pattern, but the present invention can also be applied to the optimum design of other types of light reflection patterns. For example, in the case of a V-cut reflection pattern, a function of the spacing and / or size of the V-cut may be used as a variable design element. Furthermore, the optimized design technique of the present invention can be applied to a light guide plate design that emits the best luminance in a given environment, and the light guide plate itself has a light scattering function without a printing process of a diffusion material which has been studied a lot recently. It can be applied to designing a light guide plate or developing a light guide plate design technology in which a prism function is added. The problem of optimizing the characteristics other than the light reflection pattern in the production of the BLU can be easily applied by modifying the method of minimizing the uniformity of the luminance output to the front surface of the BLU proposed in the present invention.

Furthermore, in the present BLU industry, various types of BLUs are attempted to increase the efficiency of BLUs and to produce BLUs suitable for large LCDs. Among them, BLU production of various types such as direct type BLU which places the light source under the light guide plate, BLU that uses LED by replacing CCFL with light source, and BLU made of prism-inserted light guide plate to form prism shape on the surface and back of the light guide plate It is trying. In this type of BLU, ultimately, the luminance emitted to the front of the BLU should be increased and manufactured to have a uniform luminance. Therefore, the optimization technique proposed in the present invention can be applied to the production of such a BLU.

On the other hand, the density function of the dot described above can be seen as an example of the characteristic function to determine the light reflection pattern. Therefore, the characteristic function for determining the light reflection pattern may vary depending on the type or shape or structure of the BLU. For example, in the case of the direct type BLU, a function defining the number of light sources installed on the bottom of the BLU, the spacing between the light sources, and the geometric shape of the reflector for transferring the light from the back of the light source to the front of the BLU may be determined as a characteristic function.

Although the optimization design technique of the present invention has been described in terms of the printing pattern of the diffusion ink printed on the lower surface of the light guide plate, the efficiency of the BLU is not only the printing pattern of the diffusion material but also the bending of the lamp reflector, the various forms of the unprinted light guide plate, the lamp reflector It depends on several factors, including the position of the lamp at Therefore, it is necessary to develop a technique for optimizing in consideration of the shape of the lamp reflector and the light guide plate as well as the density function of the light guide plate printing pattern by applying these various factors.

, 확장

, 수축

, 축소

의 값에 따라서 그 결과가 달라질 수 있고, 최저점에 수렴하는 속도도 달라진다. 그러므로 이러한 최적화 변수의 초기값, 이동 범위, 알고리즘의 계수의 설정에 대한 연구가 함께 진행될 필요가 있다. In addition, while performing optimization using the simplex search algorithm, the initial value of the variable of the objective function, the range of the simplex movement, and the symmetry that is the coefficient of the simplex search algorithm

, expansion

, Shrink

, reduction

The result depends on the value of and the rate of convergence to the lowest point. Therefore, studies on the initial value of the optimization variable, the moving range, and the setting of the coefficient of the algorithm need to be conducted together.

The BLU optimal design method described above may be implemented as a program and included as a part of a design packet of the BLU of the LCD. Using the present invention implemented as a program, it is possible to significantly improve the BLU design capability, such that the entire process of the BLU design can be fully automated. Not only can an optimized BLU structure design be obtained very quickly, but also the degree of optimization of the design can be further improved, which greatly improves the brightness uniformity of the light guide plate. In addition, it is possible to secure conditions for efficiently responding to the rapidly developing new models of BLU.

Although described above with reference to a preferred embodiment of the present invention, those skilled in the art various modifications and changes to the present invention without departing from the spirit and scope of the invention described in the claims below You can. Accordingly, all changes that come within the meaning or range of equivalency of the claims are to be embraced within their scope.

Claims (10)

As a method for optimal design of a light guide plate for a backlight unit using a programmed computer, Design elements related to the physical specifications of the backlight unit, including at least the size of the backlight unit and the light guide plate, the number and location of the lamps, and the properties of the optical components employed in the backlight unit, including the light transmittance of the light guide plate and the reflectance of the light reflection pattern. Among the design elements, the light reflection pattern is used as a variable design element that is an object of optimal design, and a characteristic function represented by a polynomial is defined correspondingly, an initial value is given to the coefficient of the characteristic function polynomial, and other design elements A first step of automatically modeling the backlight unit by assigning a predetermined constant value to each of them as fixed design elements; Calculating a target function value representing luminance uniformity of the light guide plate by performing a backlight unit simulation using a ray tracing technique with respect to the modeled backlight unit; And Coefficient of the characteristic function polynomial which causes the objective function value to be changed to improve the luminance uniformity until a predetermined termination condition is satisfied, including when the calculated objective function value satisfies a preset threshold value. New calculation through execution of a simplex search algorithm, the calculated new coefficients are applied to modify the characteristic function, a new light reflection pattern of the light guide plate is calculated based on the modified characteristic function, and the newly calculated light reflection And re-modeling the backlight unit by applying a pattern, and then feeding back to the second step, the third step of feeding back the light reflection pattern of the light guide plate. The method of claim 1, further comprising the step of: calculating the light reflection pattern defined by the characteristic function at that time and determining the final design proposal when the termination condition is satisfied. Optimal design method. The method of claim 1, wherein the objective function is a standard deviation of grid counts ( c _ij ) representing the number of photons that pass through a small square divided into a grid in front of the light guide plate when the BLU is simulated using the light tracking technique. And a function defined as a ratio of the maximum value and the minimum value of the grid counts ( c _ij ). The light reflection pattern of the light guide plate according to claim 1, wherein the characteristic function is a density function that defines the density according to the distance from the light source of the ink dot printed on the lower surface of the light guide plate to provide a light diffusion function. Optimal design method. The method of claim 1, wherein the characteristic function is a function of defining a distance and / or size according to the separation distance from the light source of the V-cut formed on the lower surface of the light guide plate to provide a light diffusion function. Optimal design method of light reflection pattern. The method of claim 1, wherein the characteristic function is a function that defines the number of light sources installed on the bottom of the BLU in the direct type BLU, the spacing between the light sources, the geometric shape of the reflector plate to transfer the light from the back of the light source to the front of the BLU Optimal design method of direct type BLU, characterized in that The method of claim 1, wherein the backlight unit simulation measures the total reflection coefficient, the absorption coefficient, and the diffusion coefficient of each component of the backlight unit, and according to these coefficients, the photon path in any direction when the photons hit the surface of the specific component. Determine whether to change the value, record information on the output direction and output position of all photons output to the front surface of the light guide plate, and divide the information into the grid on the front surface of the light guide plate using the output information of the photons. And counting the number of output photons that pass through each square, and calculating the value of the objective function by matching the counted output photons with luminance. The method of claim 1, wherein the termination condition is one of: (1) when the number of times of repeating the execution of the second and third steps exceeds a predetermined number of times; (2) when the number of calculations of the simplex search is greater than a predetermined maximum number of calculations during the execution of the simplex search algorithm; (3) the sum of the distances between the mean of the remaining points except for the largest vertex of the function value in the simplex and the remaining vertices is less than or equal to a specified error; (4) the sum of the distances from the vertex with the smallest value to the remaining vertices in the simplex is less than or equal to the specified error; And (5) at least one or a combination thereof when the same minimum value is repeated at least for each iteration of the simplex search algorithm. The simplex search algorithm of claim 1, wherein the simplex search algorithm performs a symmetry, expansion, contraction, and reduction operation on the simplex defined by the existing coefficients of the characteristic function to perform the symmetry, expansion, contraction, and reduction operations. Computing the new minimization that can be minimized, the symmetry, expansion and contraction operations create one symmetry point, expansion point and contraction point at the new position, and d + 1 of the existing simplex (where d is Remove the point having the highest function value among the vertices of the natural number), and add the newly created symmetry, expansion and contraction points to the remaining d vertices to form a new symmetry, expansion and contraction simplex with d + 1 points, respectively. The reduction operation creates a new d reduction vertices and sums them with the one with the lowest function value among the vertices of the existing simplex. W phase, characterized in that the operation of creating a new reduced simplex; And modifying the coefficients of the existing characteristic function by using the newly calculated vertices of the simplex. The method of claim 1, wherein the ray tracing method is performed by a program module implemented using Monte Carlo light tracing or photon mapping.

Images