KR101936627B1 - Apparatus and method for enhancing image quality using fuzzy technology - Google Patents

Abstract

The present invention relates to an apparatus for enhancing image quality using a fuzzy technique and a method thereof. Further, according to the present invention, an apparatus for enhancing image quality using a fuzzy technique comprises: a knowledge database for storing fuzzy rules, in which the fuzzy rules use a video motion, a location, an interpolation method and a window size as fuzzy variables and are based on a language value defined by a fuzzy set on a discussion area of each fuzzy variable; a fuzzification unit for receiving an image stream and converting the video motion and location into a fuzzy set; a fuzzy inference unit for applying the fuzzy set received from the fuzzification unit to the fuzzy rules stored in the knowledge database, and inferring the fuzzy rules that can be used for interpolation; and a de-interlacing unit for performing de-Interlacing using the fuzzy rules inferred by the fuzzy inference unit.

Description

[0001] Apparatus and method for improving image quality using fuzzy technology [

The present invention relates to an apparatus and method for improving image quality using a fuzzy technique.

Expert systems are based on common sense, but common sense is generally ambiguous [1-3]. Therefore, this ambiguous concept can not be directly used as an engineer knowledge and is used after being processed in the pre-processing stage.

Then this knowledge becomes meaningful and the engineer uses it at the same level of understanding on the computer [4].

The problem here is how to manage ambiguous and ambiguous expertise in certain concepts of computers [5].

To solve this problem, some researchers studied artificial intelligence, including fuzzy set theory.

Fuzzy set theory is used to manage the degree of fuzzy and ambiguity (FV). This FV can be applied to many subjects such as the degree of room temperature, the degree of person's height, the degree of vehicle speed, the degree of school distance, and the degree of beauty of ladies [6-9]. All this information can be described on a scale. In general, this scale is difficult to distinguish between "yes" and "no" categories.

In 1937, the philosopher M. Black published a paper entitled "Ambiguity: Exercise in Logical Analysis". Later, in 1965, L. Zadeh published a "fuzzy set." In this paper, Zadeh used the fuzzy likelihood theory as the formal logic system.

In addition, he proposed a new concept of applying natural language terms to fuzzy sets. Natural language is concrete, immediate, and informative. Zadeh has coded binary logic as a membership grade.

Thus, the classical two-valued Boolean logic has become a multi-valued fuzzy logic. The fuzzy logic can also represent a logical value between 0 and 1. 0 and 1 are completely false and totally true [11-14].

In image processing, noise reduction is an important issue to be solved [15-21].

There are two types of noise. One is impulse noise and the other is random noise [22-23]. In general, impulse noise is referred to as the "Soul-and-Pepper noise", and random noise is referred to as "Gaussian noise". Gaussian noise is represented by two parameters: mean (μ) and variance (σ ² ).

Noise is generated during image / video capture, transmission, and storage. Therefore, during TV communication, information may be changed due to noise.

Fuzzy logic is a process of multi-valued computation, and a variable can be a value between 0 and 1. Crips logic is the opposite of fuzzy logic, which is called Boolean logic. Only two values of "0" and "1" are used.

Most of the traditional edge-based single field de-interlacing methods employ an average filter with a small number of taps, so the restoration accuracy is low as the frequency response of the average filter looks like a dome.

[1] B. Reusch, M. Fathi and L. Hildebrand, "Fuzzy color processing for quality improvement", in Soft Computing, Multimedia and Image Processing - Proceedings of the World Automation Congress. Albuquerque, NM: TSI, (1988), pp. 841-848. [2] S. Bothorrel, B. Bouchon and S. Muller, "A fuzzy logic-based approach for semi-parametric analysis of microcalcification in mammographic images", Int. J. Intell. Syst., Vol. 12, (1997), pp. 819-843. [3] K. Kang, G. Jeon and J. Jeong, "A single field interlaced to progressive format conversion using edge map in the image block", in Proc. IASTED SIP2009, Innsbruck, Austria, (2009). [4] C. C. Lee, "Fuzzy logic in control systems", IEEE Trans. Syst., Man, Cybern., Vol. 20, no. 2, (1990), pp. 404-435. [5] T.-H. Tsai and H.-L. Lin, " Design and implementation for deinterlacing using the edge-based correlation adaptive method ", SPIE Journal of Electronic Imaging, vol. 18, no. 1, (2009), pp. 013014. [6] Z. Wang, A. C. Bovik, H. R. Sheikh and E. P. Simoncelli, "Image quality assessment: from error visibility to structural similarity", IEEE Trans. Image Process, vol. 13, no. 4, (2004), pp. 600.612. [7] F. Michaud, C. T. L. Dinh and G. Lachiver, "Fuzzy detection of edge direction for video line doubling", IEEE Trans. Circuits Syst. Video Technol., Vol. 7, no. 3, (1997), pp. 539-542. [8] F. Russo and G. Ramponi, "Edge detection by FIRE operators", in Proc. FUZZ-IEEE.94 - 3rd IEEE Int. Conf. Fuzzy Systems, (1994), pp. 249-253. [9] K. Basterretxea, J. M. Tarela and I. del Campo, "Digital Gaussian membership function circuit for neuro-fuzzy hardware", IET Electron. Lett., Vol. 42, no. 1, (2006), pp. 44-46. [10] L. A. Zadeh, "Fuzzy sets", Inf. Control, vol. 8, (1965), pp. 338-353. [11] W. Wu, Z. Liu, W. Gueaieb and X. He, "Single-image super-resolution based on Markov random field and contour transform", J. Electron. Imaging, vol. 20, (2011), pp. 023005. [12] J. Wu, T. Li, T.-J. Hsieh, Y.-L. Chang and B. Huang, " Digital signal processor-based 3D wavelet 116 error-resilient lossless compression of high-resolution spectrometer data ", Journal of Applied Remote Sensing, vol. 5, (2011), pp. 051504. [13] K. He, J. Sun and X. Tang "Guided image filtering", in Proc. of the 11th European Conf. on Computer Vision (ECCV), vol. 6311, (2010), pp. 1-14. [14] I. Pekkucuksen and Y. Altunbasak, "Gradient based threshold free color filter array interpolation", in Proc. of IEEE Int. Conf. on Image Processing (ICIP), (2010), pp. 137-140. [15] W. Wu, Z. Liu and D. Krys, "Improving Laser Image Resolution for Markov Random Field Method", Autom. Const., Vol. 21, (2012), pp. 172-183. [16] G. Jeon, M. Anisetti, V. Bellandi, E. Damiani and J. Jeong, "Designing a type-2 fuzzy logic filter for improving edge-preserving restoration of interlaced-to-progressive conversion", Inf. Sci. vol. 179, no. 13, (2009), pp. 2194-2207. [17] G. Jeon, M. Anisetti, D. Kim, V. Bellandi, E. Damiani and J. Jeong, "Fuzzy rough sets hybrid scheme for motion and scene complexity adaptive deinterlacing", Image Vision Comput. 27, no. 4, (2009), pp. 425-436. [18] G. Jeon, M. Anisetti, J. Lee, V. Bellandi, E. Damiani and J. Jeong, "A Concept of Linguistic Variable-based Fuzzy Ensemble Approach: Application to Interlaced HDTV Sequences", IEEE Trans. Fuzzy Systems, vol. 17, no. 6, (2009), pp. 1245-1258. [19] J. Wu, A. Paul, Y. Xing, Y. Fang, J. Jeong, L. Jiao and G. Shi, "Morphological dilation image 107 coding with context weights prediction" . 25, no. 10, (2010), pp.717.728. [20] W. Wu, Z. Liu and X. He, "Learning-based super resolution using kernel partial least squares", Image Vision Comput., Vol. 29, (2011), pp. 394.406. [21] J. Wu, J. Huang, G. Jeon, J. Jeong, and L.C. Jiao, " An adaptive autoregressive de-interlacing method ", Optical Engineering, vol. 50, no. 5, (2011), pp. 057001. [22] H. S. Malvar, L. W. He and R. Cutler, "High-quality linear interpolation for demosaicing of Bayer patterned color images", in Proc. IEEE International Conference on Speech, Acoustics, and Signal Processing, (2004). [23] X. Zhang and B. A. Wandell, "A spatial extension of CIELAB for digital color image reproduction", J. Soc. Inf. Display, vol. 5, no. 1, (1997), pp. 61.67.

An object of the present invention is to provide an apparatus and method for enhancing image quality using an efficient fuzzy technique with increased accuracy by performing deinterlacing based on a fuzzy rule.

According to the present invention, an apparatus for improving image quality using a fuzzy technique includes a fuzzy rule generation unit for generating a fuzzy rule based on a language value set in a fuzzy set on a discussion area of each fuzzy variable, A knowledge database storing the information; A fuzzy unit for receiving a video stream and converting a video operation and a position into a fuzzy set; A fuzzy inference unit for applying fuzzy sets received from the fuzzy unit to fuzzy rules stored in the knowledge database to deduce fuzzy rules that can be used for interpolation; And a deinterlacer for performing deinterlacing using the fuzzy rules derived from the fuzzy inference unit.

According to the present invention, there is provided a method for improving image quality using a fuzzy technique, comprising the steps of: (A) setting a video operation, a position, an interpolation method and a window size as fuzzy variables in a knowledge database, Storing a fuzzy rule based on a language value; (B) a fuzzy logic unit receiving a video stream and converting a video operation and a position into a fuzzy set; (C) applying a fuzzy set transmitted from the fuzzy inference unit to fuzzy rules stored in the knowledge database to infer fuzzy rules that can be used for interpolation; And (D) the de-interlacing unit performing de-interlacing using the fuzzy rules derived from the fuzzy inference unit.

According to the present invention, the deinterlacing is performed based on the fuzzy rule, and the accuracy is increased and efficient.

1 is a conceptual diagram for explaining three main subjects in image processing in general.
Figures 2 (a) and 2 (b) show a membership function of a small ( _SMALL ) and a large ( _LARGE ), and Figures 2 (c) and 2 (d) Quot; prodor " and " max ", and Figures 2 (e) and 2 (f) show the operators "product" and "min".
Figure 3 (a) shows "very", (b) shows "terrible", (c) shows "very very", (d) shows "somewhat" and (e) shows "definitely".
4 is a view showing a window to which the present invention is applied.
5 is a configuration diagram of an image quality enhancement apparatus using a purging technique according to an embodiment of the present invention.
6 is a flowchart illustrating a method of improving image quality using a fuzzy technique according to an exemplary embodiment of the present invention.
7 shows an image of an airplane 512x512, an airplane 512x512, a pepper 512x512, a boat 352x288, and the like.
8 to 11 are restored images, wherein (a) is a source image, (b) is a benchmark result, and (c) is a result according to the present invention.
Figure 12 (a) is the difference image between the original and the benchmark, and (b) is the difference image between the original and the proposed method.
Figure 13 shows a visual comparison of milk image and performance.

BRIEF DESCRIPTION OF THE DRAWINGS The present invention is capable of various modifications and various embodiments, and specific embodiments will be described in detail below with reference to the accompanying drawings.

The following examples are provided to aid in a comprehensive understanding of the methods, apparatus, and / or systems described herein. However, this is merely an example and the present invention is not limited thereto.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail. The following terms are defined in consideration of the functions of the present invention, and may be changed according to the intention or custom of the user, the operator, and the like. Therefore, the definition should be based on the contents throughout this specification. The terms used in the detailed description are intended only to describe embodiments of the invention and should in no way be limiting. Unless specifically stated otherwise, the singular form of a term includes plural forms of meaning. In this description, the expressions " comprising " or " comprising " are intended to indicate certain features, numbers, steps, operations, elements, parts or combinations thereof, Should not be construed to preclude the presence or possibility of other features, numbers, steps, operations, elements, portions or combinations thereof.

It is also to be understood that the terms first, second, etc. may be used to describe various components, but the components are not limited by the terms, and the terms may be used to distinguish one component from another .

In general, there are three main themes in image processing, which can be described as follows:

As follows :

(1) system considerations,

(2) image quality evaluation,

(3) parameter adjustment.

Here, the "system considerations" problem means there is no systematic consideration of subjectivity.

The "image quality assessment" issue addresses how to evaluate the agreed image quality.

Finally, the "parameter adjustment" problem means how to adjust the parameters of the algorithm. This concept is shown in FIG.

Fuzzy variables can be described by language variables. The fuzzy expert system is based on language variables and has the following format.

(1)

If A ₁ is B ₁ then C ₁ is D ₁ ,

If A ₂ is B _2, then C ₂ is D ₂ ,

...

If A _N is B _N then C _N is D _N.

The possible range of fuzzy variables is the universe of discourse. Language variables include the concept of very, fairly or extremely hedges (or hedges). Hedge works by itself. Here are some examples. Hedge "very" performs intensive operations, creates new subsets, and ultimately emphasizes meaning. Another example is "very", which gives a stronger meaning to "heavily".

Generally, hedges work with tasks that are proven to be effective. Hedge also helps to reflect the human mind because humans generally can not distinguish between "terrible" and "very".

Figures 2 (a) and 2 (b) show a membership function of small ( _SMALL ) and large ( _LARGE ), and Figures 2 (c) and 2 (d) Prodor "and" max ", and FIGS. 2 (e) and 2 (f) show the operators" product "and" min ".

The hedge can be applied to the above mentioned operators and is expressed as follows.

Very is an intensive operation. As mentioned above, the range of the set is narrowed and the membership of the fuzzy element is lowered. This operation is a squaring operation of mathematics.

(2)

Extremely, very similar effects, which are much greater. This operation is a factor of three.

(3)

Very very simple is an extension of the intensive operation. This operation is the square of the centralized operation.

(4)

Somewhat more (less or less) is an expansion operation. By increasing the set, we increase the membership of the fuzzy elements. This operation is represented as follows.

(5)

Indeed is an intensification operation. Strengthen the meaning of all sentences. When the belonging speed is 0.5 or more, the belonging speed is further increased, and when the belonging speed is lower than 0.5, it is lowered. Hedge 'surely' can be done in two ways.

(6)

Where HM is the height of the mountain. For example, if μ _HM (x) is 0.5, very, very, very very, somewhat, definitely 0.5, 0.125, 0.0625, 0.7071 and 0.5.

Figure 3 (a) shows "very", (b) shows "terrible", (c) shows "very very", (d) shows "somewhat" and (e) shows "definitely".

4 is a view showing a window to which the present invention is applied.

As shown in FIG. 4, as the window size (WS) increases, the number of pixels increases, and each pixel can be represented by a data vector for ease of calculation.

The number of reference pixels included in the window is 2WS for the vertical scan line and 2WS + 1 for the horizontal scan line, respectively, where WS is an arbitrary natural number. WS in FIG. 4 is an abbreviation of Window Size and includes any natural number meaning. Pixels located in the center of the window represent pixels to be interpolated, the vertical scan lines are scan lines that are displayed vertically, and the horizontal scan lines are scan lines that are displayed horizontally.

5 is a configuration diagram of an image quality enhancement apparatus using a purging technique according to an embodiment of the present invention.

5, an apparatus for improving image quality using a fuzzy technique according to an embodiment of the present invention includes a knowledge database 100, a fuzzifier 110, a fuzzy inference unit 120, and a de-interlacing unit 130 have.

The proposed expert system is based on a fuzzy if-then rule designed for video de-interlacing. The proposed method is fuzzy reasoning governed by the else action. The proposed fuzzy inference, governed by the else action, is written as follows and stored in the knowledge database 100. Here, video operation, position, interpolation method, and window size correspond to fuzzy variables.

Fast and slow are the language values determined in the fuzzy set on the discussion area of the video action.

And, the uniform region and the nonuniform region are the language values determined in the fuzzy set on the discussion area of the position.

Next, intra-field interpolation, inter-field interpolation, and line average interpolation are language values determined in the fuzzy set on the discussion area of the interpolation method.

The small window and the large window are the language values determined in the fuzzy set on the discussion area of the window size.

( Equation  7)

IF video motion is fast , THEN intra method is selected

Else IF video motion is slow , THEN inter method is selected

( Equation  8)

IF x ( i , j ) is located in homogeneity region , THEN larger window is

used

Else IF x ( i , j ) is located in heterogeneity region , THEN smaller window

is used

( Equation  9)

IF x ( i , j ) is located in homogeneity region , THEN inter method is

selected

Else IF x ( i , j ) is located in heterogeneity region , THEN LA method is

used

(If the IF video motion is fast then select the interpolation method in the field.

Else IF Video motion is slow Then select interpolation between fields.

IF x (i, j) is located in the uniform region, THEN uses a large window.

Else IF x (i, j) is located in a non-uniform region, THEN uses a small window.

IF x (i, j) is located in the uniform region, THEN field interpolation method is selected.

Else IF x (i, j) is located in the non-uniform region, use the THEN line average interpolation method.)

Here, the intra-field interpolation method using spatial correlation is applied to a line repetition method, a line average method (LA: Line Average), an edge-based linear average (ELA) method, A method of inter-field interpolation using information in fields as well as information between fields before and after using spatial correlation and temporal correlation includes a motion compensated method and a motion adaptive method .

The fuzzy logic unit 110 receives the video stream and converts the video motion and the position into a fuzzy set using Equation (10) to Equation (12) below.

The function to be used at this time uses the Gaussian function.

(10)

IF D _VM < v ₁ , THEN

Else IF v ₁ <D _VM < v ₂ , THEN

Else IF v ₂ <D _VM < v ₃ , THEN

Else IF v ₃ <D _VM < v ₄ , THEN

Else IF v ₄ <D _VM THEN

Where D _VM is the video operation difference and parameters v ₁ , v ₂ , v ₃ , v ₄ and v ₅ are empirically assigned 3, 5, 8, 11 and 15 respectively.

(11)

Where c _Gaussian is the center of the Gaussian fuzzy set, σ is the width of the Gaussian fuzzy set, and x is the input variable. σ is determined empirically and σ = 1.5 for the simulation. The fuzzy set must be selected from fast, slow, homogeneity and heterogeneity.

Similarly, a uniform region and a non-uniform region may be defined as follows.

(12)

IF D _HOR <D _HER + 10, THEN the region is strong homogeneity

region

Else IF D _HOR <D _HER , THEN the region is weak homogeneity

region

Else IF D _HER <D _HOR , THEN the region is weak heterogeneity

region

Else IF D _HER <D _HOR + 10, THEN the region is strong heterogeneity

region

(IF D _HOR <D _HER + 10, the THEN region is a strong homogeneous region,

Else IF D _HOR <D If _HER , the THEN region is a weak uniform region,

Else IF D _HER <D The _HOR , THEN region is a weak non-uniform region,

Else IF D _HER <D _HOR + 10, the THEN region is a strong non-uniform region)

Here, the parameters D _HOR and D _HER represent the difference between the uniform region and the non-uniform region.

The fuzzy inference unit 120 applies fuzzy sets received from the fuzzy logic unit 110 to fuzzy rules stored in the knowledge database 100 to deduce fuzzy rules that can be used for interpolation.

The deinterlacing unit 130 performs interpolation according to the result of the inference by the fuzzy inference unit 120 and outputs an interpolation result.

At this time, the output pixel by the de-interlacing unit 130 is obtained by the following expression (13).

(13)

Where (i, j) is the position of the interpolation pixel.

6 is a flowchart illustrating a method of improving image quality using a fuzzy technique according to an exemplary embodiment of the present invention.

Referring to FIG. 6, a method of improving image quality using a fuzzy technique according to an exemplary embodiment of the present invention includes a fuzzy inference process, which is based on a fuzzy if-then rule designed for video deinterlacing, (S100).

At this time, the fuzzy inference stored in the knowledge database is expressed by Equations (7) to (9).

Next, the fuzzy unit receives the video stream and converts the video motion and the position into a fuzzy set using Equations (10) to (12) (S110).

Meanwhile, the fuzzy inference unit applies the fuzzy set received from the fuzzy unit to the fuzzy rules stored in the knowledge database to infer the fuzzy rules that can be used for the interpolation (S120).

Then, the deinterlacing unit performs the interpolation according to the result deduced by the fuzzy inference unit and outputs the interpolation result (S130).

At this time, the output pixel is obtained by the de-interlacing unit in Equation (12).

Continuing now, we present a comparison of visual performance for four images.

7 shows an image of an airplane 512x512, an airplane 512x512, a pepper 512x512, a boat 352x288, and the like.

In the present invention, two methods are adopted: an LA method (benchmark 1) and an edge-based line average (ELA, benchmark 2) method.

Consider benchmark 2 only for basic benchmarks for visual performance comparisons. Both benchmark 1 and benchmark 2 methods are considered for objective performance comparisons. The simulation was performed on Intel Core 2 Duo CPU E8500 @ 3.16GHz.

To measure the performance of objective visual quality, the peak signal-to-noise ratio (PSNR) is applied and the PSNR is given by Where MSE represents the mean square error.

(14)

The restored values are shown in Figures 8-11. Each image mentioned in (a) shows the original image. Images (b) and (c) are benchmark results and are the result of the proposed method.

As you can see, the proposed method is visually superior to the benchmark. Stair artifacts were removed from the roof, edges and details. The blurred result of (b) was also resolved.

Figure 12 (a) is the difference image between the original and the benchmark, and (b) is the difference image between the original and the proposed method.

As you can see, the image of (b) has better performance than the benchmark method because it has few edges.

Figure 13 shows a visual comparison of milk image and visual performance, with (a) original image, (b) benchmark 1, (c) benchmark 2, and (d) proposed method

Table 1 shows the PSNR and MSE results for the four images.

(Table 1)

We used fuzzy conditional rules for format conversion to improve image quality. I applied membership effects to each membership function using membership hedges.

In the present invention, four fuzzy sets of "fast", "slow", "homogeneity" and "heterogeneity" are used.

By applying hedging, we were able to improve visual and objective performance. The simulation results show that the proposed fuzzy rule-based method outperforms two traditional benchmark methods.

100: knowledge database 110:
120: fuzzy inference unit 130: deinterlacing unit

Claims (8)

A knowledge database storing a fuzzy rule based on a language value determined by a fuzzy set on a discussion area of each fuzzy variable, using a video operation, a position, an interpolation method, and a window size as fuzzy variables;
A fuzzy unit for receiving a video stream and converting a video operation and a position into a fuzzy set;
A fuzzy inference unit for applying fuzzy sets received from the fuzzy unit to fuzzy rules stored in the knowledge database to deduce fuzzy rules that can be used for interpolation; And
And a deinterlacing unit for performing deinterlacing using the fuzzy rules derived from the fuzzy inference unit,
Wherein the fuzzy rule stored in the knowledge database is defined by the following equations (7) to (9).
(7)
IF video motion is fast , THEN intra method is selected
Else IF video motion is slow , THEN inter method is selected
(8)
IF x ( i , j ) is located in homogeneity region , THEN larger window is
used
Else IF x ( i , j ) is located in heterogeneity region , THEN smaller window
is used
(9)
IF x ( i , j ) is located in homogeneity region , THEN inter method is
selected
Else IF x ( i , j ) is located in heterogeneity region , THEN LA method is
used The method according to claim 1,
The fuzzy rule stored in the knowledge database
The language values determined in the fuzzy set on the discussion area of the video operation, including "fast" and "slow"
The language value defined in the fuzzy set on the discussion area of the position includes the " uniform area " and the " non-uniform area value &
Field interpolation, inter-field interpolation, and line average interpolation as language values determined in the fuzzy set in the discussion area of the interpolation method,
An apparatus for improving image quality using a fuzzy technique including a "small window" and a "large window" as language values set in a fuzzy set on a discussion area of a window size. delete A knowledge database storing a fuzzy rule based on a language value determined by a fuzzy set on a discussion area of each fuzzy variable, using a video operation, a position, an interpolation method, and a window size as fuzzy variables;
A fuzzy unit for receiving a video stream and converting a video operation and a position into a fuzzy set;
A fuzzy inference unit for applying fuzzy sets received from the fuzzy unit to fuzzy rules stored in the knowledge database to deduce fuzzy rules that can be used for interpolation; And
And a deinterlacing unit for performing deinterlacing using the fuzzy rules derived from the fuzzy inference unit,
Wherein the fuzzy unit is a fuzzy unit for receiving a video stream and converting a video motion and a position into a fuzzy set using a Gaussian function. (A) storing a fuzzy rule based on a language value set in a fuzzy set on a discussion area of each fuzzy variable, using a video operation, a position, an interpolation method, and a window size as fuzzy variables in a knowledge database;
(B) a fuzzy logic unit receiving a video stream and converting a video operation and a position into a fuzzy set;
(C) applying a fuzzy set transmitted from the fuzzy inference unit to fuzzy rules stored in the knowledge database to infer fuzzy rules that can be used for interpolation; And
(D) a de-interlacing unit performing de-interlacing using fuzzy rules derived from the fuzzy inference unit,
Wherein the fuzzy rule stored in the knowledge database in the step (A) is defined by the following equations (7) to (9).
(7)
IF video motion is fast , THEN intra method is selected
Else IF video motion is slow , THEN inter method is selected
(8)
IF x ( i , j ) is located in homogeneity region , THEN larger window is
used
Else IF x ( i , j ) is located in heterogeneity region , THEN smaller window
is used
(9)
IF x ( i , j ) is located in homogeneity region , THEN inter method is
selected
Else IF x ( i , j ) is located in heterogeneity region , THEN LA method is
used The method of claim 5,
The fuzzy rule stored in the knowledge database
The language values determined in the fuzzy set on the discussion area of the video operation, including "fast" and "slow"
The language value defined in the fuzzy set on the discussion area of the position includes the " uniform area " and the " non-uniform area value &
Field interpolation, inter-field interpolation, and line average interpolation as language values determined in the fuzzy set in the discussion area of the interpolation method,
A method for improving image quality using a fuzzy technique including a "small window" and a "large window" as language values set in a fuzzy set on a discussion area of a window size. delete (A) storing a fuzzy rule based on a language value set in a fuzzy set on a discussion area of each fuzzy variable, using a video operation, a position, an interpolation method, and a window size as fuzzy variables in a knowledge database;
(B) a fuzzy logic unit receiving a video stream and converting a video operation and a position into a fuzzy set;
(C) applying a fuzzy set transmitted from the fuzzy inference unit to fuzzy rules stored in the knowledge database to infer fuzzy rules that can be used for interpolation; And
(D) a de-interlacing unit performing de-interlacing using fuzzy rules derived from the fuzzy inference unit,
In the step (B), the fuzzing unit receives a video stream and converts a video motion and a position into a fuzzy set using a Gaussian function.

Images