KR20030001770A - Display Apparatus with the pseudo-contour noise detector using skin-color filter and Method of processing an image Thereof - Google Patents

Abstract

PURPOSE: A display device having a virtual-outline noise detector using a skin color filter and a method for processing image of the same are provided to process selectively only an image of a virtual-outline noise region without lowering image quality. CONSTITUTION: A virtual-outline noise detector has a motion detector(62), a skin color filter(64), and an edge detector(66). The motion detector(62) is connected in parallel to the first field delay portion(46). The motion detector(62) is used for detecting motions of pixels stored by the first field delay portion(46) and the next pixel prior to the stored pixels and predicting brightness according to the motions. The skin color filter(64) is connected serially with a reverse gamma corrector in order to detect a skin color part from corrected digital data. The edge detector(66) is connected serially with the reverse gamma corrector in order to detect an edge part of the corrected digital data.

Description

Display Apparatus with the pseudo-contour noise detector using skin-color filter and Method of processing an image Thereof}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a display device such as a plasma display panel or a digital micro-mirror element, and in particular, a virtual-contour using a skin filter to selectively reduce image quality by selectively performing image processing only in a region in which virtual-contour noise occurs. A display device having a noise detector and an image processing method thereof.

With the recent increase in the size of the display device, a thin display device is required. For this reason, various types of display devices are provided. For example, matrix panels for displaying digital signals, gas discharge panels such as plasma displays, digital micromirror devices (DMDs), EL displays, fluorescent displays, liquid crystal displays, and the like are provided. In the case of such a thin display device, in particular, the gas discharge panel is a simple process, so the large screen is easy to display, the display quality of the self-emission type is good, or the response speed is high. It is attracting attention as a display device. However, in such a display device, there is a problem that the display quality is disturbed, especially in the halftone display of the moving image part. On the other hand, it is considered to reduce the rough contour by superimposing positive or negative equalization pulses on the original signal.

Plasma Display Panel (hereinafter referred to as "PDP") and Digital Micromirror Device (hereinafter referred to as "DMD") use a subfield method, which has a binary memory, each of which has a weighted number of A dynamic image is processed by temporarily overlapping a binary image and processing a half tone. The following discussion covers PDPs, but the same applies to DMDs.

The PDP emits a phosphor by ultraviolet rays of 147 nm generated when the He + Xe or Ne + Xe inert mixed gas is discharged to display an image including characters or graphics. Such a PDP is not only thin and easy to enlarge, but also greatly improved in quality due to recent technology development. In particular, the three-electrode AC surface discharge type PDP has advantages of low voltage driving and long life because wall charges are accumulated on the surface during discharge and protect the electrodes from sputtering caused by the discharge.

1 is a perspective view showing a discharge cell of a three-electrode alternating surface discharge type PDP.

Referring to FIG. 1, a discharge cell of a three-electrode AC surface discharge type PDP includes a scan electrode 12Y and a sustain electrode 12Z formed on an upper substrate 10, and an address electrode formed on a lower substrate 18. 20X). The upper dielectric layer 14 and the passivation layer 16 are stacked on the upper substrate 10 having the scan electrode 12Y and the sustain electrode 12Z side by side. Wall charges generated during plasma discharge are accumulated in the upper dielectric layer 14. The protective layer 16 prevents damage to the upper dielectric layer 14 due to sputtering generated during plasma discharge, and increases emission efficiency of secondary electrons. As the protective film 16, magnesium oxide (MgO) is usually used. The lower dielectric layer 22 and the partition wall 24 are formed on the lower substrate 18 on which the address electrode 20X is formed, and the phosphor 26 is coated on the surfaces of the lower dielectric layer 22 and the partition wall 24. The address electrode 20X is formed in the direction crossing the scan electrode 12Y and the sustain electrode 12Z. The partition wall 24 is formed in parallel with the address electrode 20X to prevent ultraviolet rays and visible light generated by the discharge from leaking to the adjacent discharge cells. The phosphor 26 is excited by ultraviolet rays generated during plasma discharge to generate visible light of any one of red, green, and blue. Inert gas for gas discharge is injected into the discharge space provided between the upper and lower plates and the partition wall.

The PDP cell of this structure is selected by the counter discharge between the address electrode 20X and the scan electrode 12Y, and then sustains the discharge by the surface discharge between the sustain electrode pairs 12Y and 12Z. In the PDP cell, the fluorescent material 26 emits light by ultraviolet rays generated during sustain discharge, so that visible light is emitted outside the cell. As a result, the PDP having cells displays an image. In this case, the PDP implements a gray scale required for displaying an image by adjusting the discharge sustain period of the cell, that is, the number of sustain discharges, according to the video data.

As a driving method of such a PDP, an address / display period-separated subfield (Address ADS) driving method as shown in FIG. Representative. In the ADS driving method, one frame is divided into n subfields corresponding to each bit of n-bit image data, and each subfield is further divided into an address period and a discharge sustain period. Here, ²⁰ in the address period are the same, and the sustain discharge period of each sub-field: 2 ^1: 2 ^2: ... The gray scale is represented by a combination of display periods by weighting a ratio of: 2 ^n-1 . The address period of each subfield includes a reset period for initializing the full screen.

In addition, the driving method of the PDP divides one field into subfields represented by brightness of 1,2,4,8,16,32,64,128, and performs an address in units of subfields as shown in FIG. By deciding on / off, all 256 people will be represented. Displaying the brightness of an image in this way creates annoying contours around moving objects during video display and degrades the image quality. This is called false contour noise. The cause of the pseudo contour is accumulated along the propensity of chasing a moving object on the screen and the brightness of cells adjacent to the moving object from a fixed position on the retina for 1 frame display period (16.7 ms), which is different from the actual brightness. Is in the perception of a person.

FIG. 4 illustrates pseudo contouring occurring at 127 and 128 degrees with respect to the basic subfield arrangement shown in FIG. 3.

Referring to FIG. 4, FIG. 4A shows a state in which the display image is scrolled by one pixel per frame from left to right, and reading only 127 and 128 degrees as shown in FIG. 4B while reading from 127 to 128 (A) and ( C) looks normal, but in the case of (B) reading the boundary that changes from 127 to 128 people, a pseudo contour like 255 appears. In this case, black lines appear on the screen, which degrades the quality of the image.

Such video pseudo contour noise may be generated in subfields within one frame, and may also be generated between frames if a large number of emission or gray level difference occurs between the frames.

As a method for removing the pseudo pseudo contour noise, a subfield is divided into one or two subfields, a subfield is rearranged, a subfield is added, and the subfields are ordered. Method of rearranging and error diffusion, putting equalizing pulse, motion compensating with motion vector, light emission sequence of subfield using limited number of gray levels Moving Picture Distortion Free Code, which is driven by a monotonically increasing code, has been proposed. However, adding subfields causes insufficient sustain periods, making it difficult to increase the resolution and complicate the circuit configuration.

A display device for displaying an image having a gray level using a plurality of subfields has a problem in that virtual-contour noise appears while a moving image is displayed. Virtual-contour noise arises from human visual characteristics. This is caused by the subfield display characteristics in a display device that displays human visual characteristics and an image having gray levels using the subfield method. In other words, when the human eye moves, a subfield different from the original gray level is projected onto the retina, where the original gray level is incorrectly detected. Next, the virtual-contour noise will be described.

Assuming that the areas A, B, C, and D have been moved from the state shown in FIG. 5 by one pixel width to the right as shown in FIG. 6, the result is that the user faces the screen to follow the areas A, B, C, and D. The eye of the viewer also shifts to the right. Then, after one field, three vertical pixels B1 in the B area will be replaced by three vertical pixels A1 in the A area. At this time, when the displayed image is changed from FIG. 5 to FIG. 6, the human eye has an area B1 having (0000 0000), which is a form of AND of the B1 area data (0111 1111) and the A1 area data (10000000). Will be recognized. That is, the region B1 does not display the original 127 luminance level, but rather displays a luminance level of zero. As a result, apparently dark borders appear in the B1 region. If a clear change from '1' to '0' is applied to the higher bits, then a clear dark border appears.

On the contrary, when there is an image change from FIG. 6 to FIG. 5, at the time of change to FIG. 5, the observer has (1111 1111) in the form of an OR of OR data of A1 area data (1000 0000) and B1 area data (0111 1111). Recognize the area A1. That is, the maximum bit is forcibly changed from '0' to '1', so that the A1 region is displayed at the luminance level of approximately 255 FOLDs instead of the original 128 luminance level. Therefore, an apparently bright borderline appears in region A1. If an explicit change from '0' to '1' is applied to the upper bits in this way, a clear border appears.

In the case of a video only, the boundary line that appears on the screen is called virtual-contour noise, which causes a deterioration of the image quality.

As a technique for reducing such virtual-contour noise, there is a display device shown in Japanese Patent Nos. 09-258689 and 10-39830. The display device of Nos. 09-258689 attempts to reduce virtual-contour noise by selecting different modulation signals for every n pixels and performing different modulations for every n pixels using the selected modulation signal. However, in such a device, a reduction process is performed on a region where virtual-contour noise does not actually appear for the entire image, thereby degrading the quality of the display image for the entire image.

Further, the display device of Nos. 10-39830 detects a dynamic area of an image and performs modulation processing on each pixel in this area to reduce virtual-contour lock. However, such a device performs virtual-contour noise reduction processing for the entire dynamic region, and thus performs virtual-contour noise processing even in regions where virtual-contour noise does not appear. As a result, the quality of the image is degraded when viewing the entire image.

SUMMARY OF THE INVENTION An object of the present invention is to provide a display device having a virtual-contour noise detector using a skin filter that selectively performs image processing only in a region where virtual-contour noise occurs so as not to deteriorate the quality of an entire image, and an image processing method thereof. .

1 is a perspective view showing a discharge cell structure of a conventional three-electrode AC surface discharge plasma display panel.

FIG. 2 is a frame diagram illustrating a subfield driving method of a discharge cell shown in FIG. 1.

3 illustrates a basic eight subfield arrangement.

FIG. 4 is a diagram illustrating pseudo contour phenomenon occurring at 127 and 128 degrees for the basic subfield arrangement shown in FIG. 3; FIG.

5 is a diagram illustrating an example of plasma display panel screen luminance distortion;

FIG. 6 is a diagram illustrating a case where one pixel shifts from the screen luminance distortion shown in FIG. 5; FIG.

7 is a diagram showing the configuration of a plasma display panel having a virtual-contour noise detector using a skin color filter according to the present invention;

FIG. 8 is a detailed view of the virtual-contour noise detector of FIG. 7. FIG.

FIG. 9 is a flowchart illustrating a method of detecting a skin color region using a skin color filter illustrated in FIG. 8.

FIG. 10 is a view for explaining movement and setting of the draw hold area in the flowchart of FIG. 9; FIG.

FIG. 11 is a view illustrating an image in which only a skin color region is detected through the flowchart illustrated in FIG. 9.

12A to 12C are diagrams showing types of filters used when driving the edge detector shown in FIG.

13A to 13D are diagrams for explaining edge detection results detected by the filter shown in FIG. 12.

<Explanation of symbols for the main parts of the drawings>

10: upper substrate 11: discharge cell

12Y, 42 scan electrode 12Z, 44 sustain electrode

13a: initial zone 13b: final zone

14 upper dielectric layer 16 protective film

18: lower substrate 20X: address electrode

22: lower dielectric layer 24: partition wall

26 phosphor 30 PDP

40: input signal 42: analog-to-digital converter

44: reverse gamma corrector 46: 1 field delay

48: virtual-contour noise detector 50: virtual-contour noise processor

52: display data processor 54: subfield processor

56: data driver 58: scan driver

60: PDP 62: motion detector

64: skin color filter 66: edge detector

In order to achieve the above object, a display device having a virtual-contour noise detector using a skin filter according to the present invention is a display for displaying the input image having a gray level using a plurality of subfields obtained by dividing one field of the input image. An apparatus, comprising: a motion detector for detecting movement of one pixel and a next pixel of the input image; A skin color filter detector for detecting a skin color region of the input image; An edge detector for detecting an edge region of the input image; And a virtual-contour noise reduction processor for controlling the data detected from the motion detector, the skin color filter detector, and the edge detector to reduce the virtual-contour noise.

The skin color filter detector is characterized by using a skin color filter.

The edge detector may be any one of a Roberts filter, a Prewitt filter, and a Sobel filter.

An image processing method of a display device having a virtual-contour noise detector using a skin color filter according to the present invention uses a plurality of subfields obtained by dividing one field of an input image to remove virtual-contour noise of the input image having a gray level. The image processing method of reducing the display device, the method comprising: predicting the brightness of the input image by detecting the movement of one pixel and the next pixel in each subfield of each pixel of the input image, and using the skin color filter Detecting a skin region of the input image, detecting an edge region of the input image, and controlling a region detected through each of the steps to perform virtual-contour noise reduction processing; .

The detecting of the skin color region of the input image according to the present invention comprises constructing a flat histogram using a threshold color hold value of a preset skin color of the input image, and an area having a larger value among cell values displayed on the flat histogram. Moving the threshold hold region around the controller; constructing a stereo histogram using the reset threshold hold value; and detecting a identifiable region by controlling a cell value displayed on the stereo histogram. Characterized in that.

The flesh color of the input image applied in the present invention is characterized in that it is determined based on hue, brightness, and saturation.

In this case, detecting the edge region of the input image may include detecting any one of a Roberts filter, a Prewitt filter, and a Sobel filter.

Other objects and advantages of the present invention in addition to the above object will be apparent from the description of the preferred embodiment of the present invention with reference to the accompanying drawings.

Hereinafter, exemplary embodiments of the present invention will be described with reference to FIGS. 7 to 13.

7 is a diagram illustrating a configuration of a PDP having a virtual contour noise detector according to the present invention.

Referring to FIG. 7, the PDP includes an analog / digital converter 42 for converting the input analog data 40 into a digital form, an inverse gamma corrector 44 for inverse gamma correction of the gamma corrected image data, and a pixel. 1 field retarder 46 for holding the movement of the PDP 60 for one field, scan driver 58 for driving the scan and sustain electrodes of the PDP 60, and address electrodes of the PDP 60; And a virtual-contour noise detector 48 and a virtual-contour noise processor 50 for detecting and controlling the virtual-contour noise of the image data.

In addition, the display data processor 52 and the subfield processor 54 display a control signal for driving the input image signal controlled by the virtual-contour noise detector 48 and the virtual-contour noise reduction processor 50 in the PDP 60. ).

An analog-to-digital converter (A / D) 42 is connected between an input line for inputting an RGB image signal and an inverse gamma corrector 44 to convert analog data from the input line into digital data, thereby converting the analog data into digital data. To feed.

The inverse gamma corrector 44 is connected between the analog / digital converter 42 and the one-field delay unit 46 to inverse gamma correct the gamma-corrected digital data and supplies it to the display data processor 52.

The one field retarder 46 is connected between the inverse gamma corrector 44 and the display data processor 52 to store the data of one pixel for one frame in order to detect movement of one pixel and the next.

As illustrated in FIG. 8, the virtual-contour noise detector 48 includes a motion detector 62 for detecting movement of one pixel and the next pixel, and a skin color filter for detecting a flesh color portion of the image data of the pixel. 64, and an edge detector 66 for detecting an edge area on the screen.

The motion detector 62 is connected in parallel to the one field retarder 46, and detects the motion of the pixel stored in the frame and the next pixel before the frame storage by the one field retarder 46, and predicts the brightness according to the motion. Since the virtual-contour noise is generated only where there is motion, the virtual-contour noise reduction processing unit 50 is executed only on the moving part to remove the virtual-contour noise.

The skin color filter 64 is connected in series with the inverse gamma corrector 44 to detect the flesh color portion of the person among the inverse gamma corrected digital data. Detecting human skin part is because skin color is regarded as important memory color by viewers in display device, and accurate reproduction of skin color in image is a measure of fidelity of color reproduction.

The edge detector 66 is connected in series with the inverse gamma corrector 44 to detect data corresponding to an edge region of the inverse gamma corrected digital data.

As a result, the virtual-contour noise detector 48 estimates the brightness of the digital data by the motion detector 62, detects the skin region by the skin color filter 64 and the edge region by the edge detector 66. This control signal is supplied to the virtual-contour noise reduction processor 50.

The virtual-contour noise reduction processor 50 is connected in series to the virtual-contour noise detector 48 to control the bits of each pixel to reduce the virtual-contour noise detected by the virtual-contour noise detector 48. It supplies to the display data processing part 52.

The input data supplied to the display data processing unit 52 is controlled by the subfield processor 54 and then to the scan driver 58 for driving the scan electrode and the sustain electrode and the data driver 56 for driving the address electrode. Supplied.

Through the driving signals supplied to the scan driver 58 and the data driver 56, the PDP 60 displays image data with reduced pseudo contours.

FIG. 9 is a diagram illustrating a flowchart for detecting a skin color region in the skin color filter 64 of FIG. 8.

Referring to FIG. 9, after generating 2D and 3D color histograms, clustering cells having a color corresponding to a flesh color value and classifying only pixels corresponding to the flesh color.

At this time, the color of the input image is R (Red), G (Green), B (Blue) space instead of Hue (hereinafter referred to as "H"), Saturation (hereinafter referred to as "S"), Brightness (Value; Is called "V"). The HSV representation of color represents human perception better than the RGB representation.

First, the highest and lowest threshold values of each color element for the input image 70 are determined. Thus, when the inverse gamma corrected input image 70 is input, a 2D color histogram composed of SVs is generated according to a predetermined threshold hold value.

A threshold hold area is created using the threshold hold value, and the maximum value of the histogram in the threshold hold area is m. At this time, the threshold hold area is moved around the place where cells having a value larger than 0.1 m are collected.

Determine if the transitioned distance between the next previous and current threshold hold area is smaller than the predetermined threshold hold value. (76) If the transitioned distance is large, reset the predetermined threshold hold value and set the 2D color histogram. Step 72 and moving the threshold hold area 74 are repeated until the transitioned distance becomes small.

When the repeated transition distance becomes smaller, the threshold hold area surrounding the area on the histogram having a cell value greater than 0.1 m is reduced as it is changed from the initial area 13a of FIG. 10 to the final area 13b.

The reset threshold hold area then forms a 3D color histogram of HSV up to H (color).

When the 3D color histogram is formed, it is checked whether the cell value is larger than the newly formed threshold hold value.

The previous step 82 detects an identifiable area larger than the threshold hold value.

In this case, it is checked whether the detected region is configured as one region. (86) If the detected region is configured as one region, only the region is sent to the virtual-contour noise processor 50 of FIG. 7 for control. (90)

However, if it is composed of more than one area, the areas are merged until the two areas remain and only the most dominant area is selected. 88 The area is then sent to the virtual-contour noise reduction processor 50 for control.

11 shows a picture in which only the skin part is detected by the skin color filter.

The skin part is virtually eccentric and is well recognized by the human eye. Accordingly, the image quality generated when the skin-colored portion is detected and the virtual-contour noise reduction processor 50 is applied only to the detected portion can be prevented from being lowered and the virtual-contour noise can be reduced.

12A to 12C are diagrams illustrating filters having various coefficients when the edge detector is implemented, that is, a Robert filter, a Pre filter, a Sobel filter, and the like.

In this case, edge detection using each filter becomes possible, thereby maximizing the edge detection effect of image analysis.

Figure 13 shows the results of the edges obtained with each mask method. It performs edge detection and performs virtual-contour noise processing on only the remaining portions except the region of the edge portions.

When FIG. 13A is referred to as an original input image, FIG. 13B is image data obtained by applying the Robert filter of FIG. 12A to an edge detector. FIG. 13C illustrates image data obtained by applying the pre-filter of FIG. 12B to the edge detector, and FIG. 13D illustrates image data illustrated by applying the Sobel filter of FIG. 12C to the edge detector.

In conclusion, the present invention performs virtual-contour noise calcination on the entire image to reduce the virtual-contour noise, thereby reducing the virtual-contour noise on the region where the virtual-contour noise does not actually appear. In order to solve the problem of deterioration of quality, the virtual-contour noise detection unit is used to execute the virtual-contour noise reduction processor only in an area in which virtual-contour noise is generated.

As described above, a display device having a virtual contour noise detector using a skin color filter according to the present invention exhibits a large amount of virtual contour noises in the skin part of a person when the input image is displayed on the display panel. By using the propensity to recognize contour noise, the human skin color is detected and the virtual contour noise processor is applied only to the part to eliminate the deterioration of the entire image. In addition, the detection of the movement of the image and the detection of the edge region are also controlled to remove the degradation of the image quality.

Those skilled in the art will appreciate that various changes and modifications can be made without departing from the technical spirit of the present invention. Therefore, the technical scope of the present invention should not be limited to the contents described in the detailed description of the specification but should be defined by the claims.

Claims (12)

A display device for displaying the input image having a gray level using a plurality of subfields obtained by dividing one field of the input image. A motion detector for detecting movement of one pixel and the next pixel of the input image; A skin color filter detector for detecting a skin region of the input image; An edge detector detecting an edge region of the input image; And a virtual-contour noise detector using a skin-color filter, characterized in that it comprises a virtual-contour noise reduction processor for controlling the data detected from the motion detector, the skin color filter detector, and the edge detector to reduce virtual-contour noise. Display. The method of claim 1, An analog / digital converter for converting the input analog data into digital data; A scan driver for driving the scan electrode and sustain electrode of the display device; And a data driver for driving an address electrode of the display device. The display device having a virtual contour noise detector using a skin color filter. The method of claim 1, An inverse gamma corrector connected to the analog / digital converter for inverse gamma correction of gamma-corrected digital data; And a one-field retarder connected in series with the inverse gamma corrector to frame-store data of one pixel for one frame to detect movement of one pixel of the digital data and inverse gamma corrected pixel. A display device having a virtual-contour noise detector using a skin filter. The method of claim 1, And the skin color filter detector uses a skin color filter. The method of claim 1, The edge detector is a display device having a virtual-contour noise detector using a skin filter, characterized in that any one of the Roberts filter, Prewitt filter, Sobel filter. In the image processing method of the display device for reducing the virtual-contour noise of the input image having a gray level using a plurality of subfields obtained by dividing one field of the input image, Predicting the brightness of the input image by detecting movement of one pixel and the next pixel in each subfield of each pixel of the input image; Detecting a skin color region of the input image using a skin color filter; Detecting an edge region of the input image; And controlling the area detected through each of the steps to perform a virtual-contour noise reduction process. 12. An image processing method of a display apparatus having a virtual-contour noise detector using a skin color filter. The method of claim 6, And predicting the brightness of the input image comprises measuring a brightness felt by the observer's retina in each of the pixel regions of the image. The method of claim 6, And a skin color filter, wherein the skin color of the input image is determined based on hue, brightness, and saturation. The method of claim 6, Detecting the skin color region of the input image comprises: constructing a planar histogram using a preset threshold value of skin color of the input image; Moving the threshold hold area around an area having a larger value among the cell values shown on the planar histogram; Constructing a stereo histogram using the reset threshold values, And detecting the identifiable area by controlling the cell value displayed on the stereo histogram. The method of claim 9, The planar histogram is configured based on brightness and saturation. And forming the stereo histogram based on hue, lightness, and saturation, wherein the stereoscopic histogram comprises a virtual contour noise detector using a skin color filter. The method of claim 10, And the threshold hold area is formed based on the highest and lowest values of the hue, lightness, and saturation. The method of claim 6, The detecting of the edge region of the input image may include detecting any one of a Roberts filter, a Prewitt filter, and a Sobel filter. An image processing method of a display device having a virtual-contour noise detector.