KR100842616B1 - Method and Apparatus For Detecting Flat Panel Display By Vision Model - Google Patents

Abstract

The present invention relates to a method and apparatus for detecting a flat panel display by a vision model, wherein defects of spots present in the flat panel display appear, i.e., a system for generating a difference recognition by hardware or software. By using the photographing system, the image screen of the test panel can be obtained, and the detection reference information by the detection information generation system, which takes a reference image with the background simulation system, is composed of the following steps. In the flow of recognition generation which consists of the process of capturing the image of the test panel, adjusting the captured image, generating the test image and the reference image, preliminary processing, estimating the image information, and matching the image information. The process of creating a test image and a reference image, the process of collecting the respective test information at the end to generate the detection and detection values, and using the degree of detection and the detection values thereof, to determine the good and bad image quality of the flat panel display device. Judging process of the panel's good and bad.

Description

Method and Apparatus for Detecting Flat Displays by Visual Models {Method and Apparatus For Detecting Flat Panel Display By Vision Model}

1A and 1B are schematic diagrams showing the types of stains in the stain detection method according to the related art.

Fig. 2 is a flowchart showing a spot detection method according to the related art.

3 is a schematic diagram showing an apparatus for detecting a flat panel display by a visual model according to the present invention;

4 is a schematic diagram showing a connection of a main system according to the present invention;

5 is a flowchart showing a method of detecting a flat panel display by the visual model according to the present invention;

6 is a flowchart illustrating a process of adjusting image capture according to the present invention.

7 is a flowchart illustrating a shearing process according to the present invention.

8 is a flowchart illustrating a process of estimating image information according to the present invention.

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

a: L line b: Vertical line

c: curve e: thin line

f: thick line h: edge area

g: bright area i: dark area

j: Edge bright area 10: Flat display device

31: Panel 32: Shooting System

33: Background Simulation System 34: Image Capture System

35: generation system of difference recognition 36: output system

37: Detection degree 38: Detection value

41: Shooting System 43: Background Simulation System

45: generation system of difference recognition 46: clock condition

47: visual model 48: model correction

49: generation system of detection information

The present invention relates to a method and apparatus for detecting a flat panel display by a visual model. In particular, the present invention relates to a method for detecting a flat display device based on a detection level and a numerical value of the visual model, and to consider the entire panel and to directly detect the entire panel. By this, all the panel quality can be clearly seen as good or bad.

In accordance with the general purpose of flat panel displays, liquid crystal displays (LCDs) are widely used, for example, in televisions, monitors for personal computers, mobile phones, and various home electrical appliances, and the quality of liquid crystal display panels is high. In the state, it is important, for example, the color, contrast, reaction time, luminance, and the like, and the background uniformity of the LCD is judged by observing the luminance of the LCD and the stain, and the stain is the luminance of the display device. According to the unevenness of color, the situation of various lines is generated, and the simplest judgment method is changed into a black screen and other step screens, and it is observed well from various different angles. Various spots can be seen, and various spots include, for example, horizontal lines or 45 ° lines, blocks, and circular blocks, and blocks at any corners. There is a possibility. Since these defects are low contrast and irregular display, the degree of defects with many stains is an important reference for judging whether the quality of the flat display panel is good or bad.

However, in the prior art, since the judgment of the spot is judged by observing the screen flaw of the flat panel display directly, there are the following drawbacks.

1, inspection, the screen is difficult to take.

2, it is difficult to classify stains of various situations, and the definition of each manufacturer is different, and the standard of judgment is insufficient.

 3, there is no level of defect degree in which unevenness is described, and there is no standard of recognition even if there is.

In the international organization VESA (VEDIO ELECTRONICS Standards Association), there are definitions of various types of stains (VESA FPDM 2303-8), which are simply simple definitions, and there is no suggestion for resolving the deficiency and its judgment.

In the related art, there are a device for detecting a spot and a method thereof, and a definition of a spot is posted in US6154561 in the US invention. As shown in Fig. 1A, a flawless flat display Including the line blobs of the flat panel display 10, which are adjacent to different negative pixels, for example lines, curves c, L lines a, vertical lines b. ), Thin lines (e), thick lines (f), and the like. In addition, as in the block-like spots shown in FIG. 1B, spots such as dark areas i, bright areas g, and the like, and spots such as edge areas h and edge bright areas j that occur at the edges of the panel. It includes.

For example, US5917935 of the US patent sets a threshold value, compares the spot of the spot with the background value, finds the degree of the spot of heavy spots, creates a statistical list, and adds to the flowchart and the statistics list shown in FIG. It is published by classification as the state of the stain of a different definition. This process consists of the following processes. The process of capturing the first screen from the display panel (step 201), the process of generating a plurality of sample screens by this first screen (step 202), and filtering each subsample screen according to different needs, and various histograms of the screen. (step 203) of the predetermined feature filter by the histogram (step 203), a step in which a threshold is set in each of the screens, and a feature block is generated by the threshold (step 204), and the analysis of the feature block The process of determining the first screen spot flaw (step 205), the step of characterizing the spot flaw (step 206), and the last adjustment operation are performed, thereby eliminating error detection and removing the spot. The process of determining the classification (step 207).

According to the process of the patent, only various stain classifications, for example, line stains (Spot Mura), spot stains (Fil1 Port Mura), edge stains (Pannel Edge Mura), block stains (Block) The quality of the display panel is not determined for various patterns.

Just Noticeable Difference (JND) published at International Display Workshops (IDW) in 2003, and Mura Analysis Method by Using JND luminance and the SEMU definition It is about the degree of operation, but it is too simple and includes only a few parameters, which is insufficient in the determination of the quality of the display panel. Said so-called JND shows the conventional method of observing a difference with the naked eye.

The SEMU International Standard publishes a measurement definition (SEMI D 31-1102) of spot inspection on flat display pixels. However, the above is limited to the area of the display device, the contrast, and the background luminance along with the definition or analysis of the spots, and is too simple as a whole, so that the defects of the flat display device cannot be exhibited. Therefore, the prior art has the following drawbacks.

1, artificial inspection is used, so the reliability is low and there are questions.

2, the module is too simple because it takes only the description or classification of physics as a stain and does not consider human detection.

3, only severe stains, and the quality of the display device is not judged.

4, a plurality of detection of the request segmentation of the automatic inspection technique are executed, that is, various stains are inspected in different patterns, and a plurality of parameters cannot be determined. If there are new stains, a new calculation method is required.

However, with respect to a preferred embodiment of the present invention, a conventional JND map is employed, and regarding the application thereof, reference can be made to "method of determining image quality by JND map" published in US Pat. have.

The present invention is a method and apparatus for detecting a flat panel display by a visual model. For example, an image is taken by a photosensitive device such as a CCD (Charge Coupled Device, CCD) device, a CMOS, and a reference image. By referring to the reference image, the defects of the artificial inspection are eliminated, and a detection method and a detection map are generated by a predetermined calculation method and a vision model. For example, the JND map and the JND value are used to determine the quality of the display device. By judgment, i.e., considering the entire display panel, the definition of spots is not considered, but the quality of the entire panel is taken good or bad by a direct inspection method.

The apparatus described in the present invention includes a discriminability metric generation system, an imaging system connected to a difference recognition generation system, a background simulation system connected to a difference recognition generation system, and a reference image. (reference image) includes a visual model system for introducing the visual model into the visual model system or the generation system of difference recognition, and a generation system of detection information connected to the generation system of difference recognition and generating reference information for detection. do.

The detection method includes a step of generating an image of the test panel which captures an image of the test panel by a photographing system, a step of simulating generation of a reference image by a background, and a difference between the image of the test panel and the reference image. A process of generating a detection degree by combining a process of generating a recognition flow by the recognition generating system, test information of generating a recognition flow, a process of obtaining a detection value by mathematical means from test information of generating a recognition flow, and the detection And determining the good or bad of the flat panel display device, which judges whether the quality of the display device is good or bad based on the detection value.

The present invention relates to a method and apparatus for detecting a flat panel display device using a visual model. The present invention relates to a method for detecting a flat display device using a visual model, and to taking an image by a photosensitive device such as a CCD (Charge Coupled Device, CCD) device or a CMOS. JND maps) and detection values (JND values) are generated, the judgment of the quality of the display device allows the whole panel to be considered, and the direct inspection method shows that the quality of the entire panel is good or bad.

The present invention has at least the following advantages.

1, the conventional inspection apparatus still employs artificial detection with low reliability, and the present invention is to avoid bringing the question of inspection of the flat panel display into the artificial work.

2, the accuracy is good to the extent that the stain is severe, and submit a reasonable inspection degree.

3, not only the degree of unevenness is described but also the quality of the display panel can be obtained thereby.

4, create an inspection level of spots on the flat panel display, the maker solves the question of the quality of the display.

5, there is no classification test, i.e., no analysis is performed on each of the different spots, the whole display panel is considered, and a new calculation process required by the new spots is omitted.

6, the visual model introduces the generation of the detection degree (e.g., JND map), and in the visual model, factors such as shape, area, contrast, etc. are considered. Evaluate the quality.

7, the degree of detection (e.g., JND map) used in the present invention results in unclear and uneven spots.

Fig. 3 is a schematic diagram showing a device for detecting a flat panel display by the visual model of the present invention, and the detection diagram (preferred embodiment is JND map) and the detection value (e.g. JND value) is intended to determine whether the panel 31 of the flat panel display device is good or bad. First, an image is taken by a photosensitive device such as a CCD device or a CMOS, which is built in the imaging system 32, and the pixel of the test display device is taken by using the image capture system 34. The background simulation system 33 introduces the relevant background according to the actual atmosphere, and there is a reference image of the contrast of the captured test image, so that an objective image can be obtained.

The captured image and the reference image are introduced into the generation system 35 of difference recognition, and the detection degree 37 and the detection value 38 are generated through the output system 36 by the predetermined calculation method and the visual model. The output system 36 can be performed by a display device, a graph, or the like, and the inspector allows the quality of the panel 31 to be judged by the detection degree 37 and the detection value 38.

Fig. 4 is a connection diagram showing the respective sub-systems according to the present invention, and the present invention is directed to the generation system 45 of difference recognition by using the generation system of difference recognition, that is, to show spot defects of the flat panel display. By connecting to the imaging system 41 in the system, an image screen of the test display device is taken, and a reference image is obtained by the background simulation system 43 of the generation system 45 of difference recognition. The generation system 45 of the difference recognition further introduces a simulated visual model of the naked eye visual system, adds a viewing condition 46, and joins the adjustment value appropriately from the inspector (for example, a panel maker). In addition, the generated inspection values may be introduced into a model calibration 48 in response to the results of test environments that are appropriately different, and a model of the system may be provided to adapt to various test environments by changing the input parameters. Finally, by connecting to the detection information generation system 49 of the difference recognition generation system 45, detection reference information is generated, and the detection information generation system 49 is, for example, a display device, a graph, The present invention can be connected to an output system (see FIG. 3) such as a detection degree (JND map) and a detection value (JND value) of the present invention.

Fig. 5 is a flowchart showing a method for detecting a flat panel display by the visual model according to the present invention, which is a process of acquiring a test panel and introducing a background, a process of generating a test image and a reference image, and a generation system of difference recognition. The process of generating a flow of recognition by this is included. There are the following steps.

First, the test panel image is captured from the photographing system (step 51), and then, for example, adjustment of image phase difference correction, noise cancellation, adjustment of image angle, ratio and size to luminance value is performed. Adjust the captured image (step 52), generate a test image (step 53), and also simulate the background image (step 54), taking into account the inspection environment, generate the reference image, and reference image ) May be a uniform background, and ordinary people may receive an uneven background image (step 55).

The test image and the reference image generate a recognition flow by the generation system of the difference recognition, and this process includes the front processing, the estimation of the image information, the combination of each parameter, and the like. This recognition flow can introduce viewpoint conditions, the inspector adds adjustments as appropriate, and the output inspection values can respond to the results of a moderately different test environment, or can introduce model calibration, and overall, model the system. By changing the parameters therein to suit various test environments (step 56).

Front processing includes searching for image edges, capturing display ranges, adjusting images, compensating edges, simulating and calibrating ambient light in a test environment, and so forth (steps 561 and 564).

Incidentally, the estimation of the image information is a plurality of flows, for example, a parameter of a human element is introduced to the following process (for details, refer to FIG. 8 and the description is omitted here).

a, light adaptation correction

b, channel decomposition

c, contrast computation

d, contrast sensitivity filter

e, correction of oblique effect

f, correction of aperture

By correcting each of the above parameters, more accurate inspection data is obtained for the purpose of estimating and planning image information from various different reference values (steps 562 and 565).

The combination of the process of the image information captures the image information consisting of the test image and the reference image in the processes 562 and 565, taking into account the interaction of different test information (channels), and each of the channels generated in the process of estimating the image information. Further correction is made as a result of (step 563, 566).

The combination (channel) of each test information may generate the detection degree of a JND map, for example by the difference recognition generation system shown in FIG. 4, and this detection degree may be each channel, and luminance may be sufficient as it. The detection degree of luminance and chromaticity may be sufficient, and the detection degree of each test parameter mentioned above may be combined (process 57).

By means of arithmetic calculations and mathematics, the preferred embodiment uses the Minkowski pooling averaging method (step 58), whereby each of the detection degrees is coupled to, for example, a detection value of a JND map (e.g., a JND value). 59)

It is possible to judge the image quality of the flat panel display by the detected value and the detection degree, to judge whether the panel is good or bad, and to obtain the maximum difference area received by the naked eye from the detection degree.

FIG. 6 is a detailed flow of step 52 of the image capture adjustment shown in FIG. 5, and there are the following steps.

The process of adjusting the image capture is started to prepare a panel of the test display device and various testers (step 601).

Next, the captured image is corrected and, for example, adjustment of the captured image deviation, noise cancellation, adjustment of the angle, ratio, and size of the image (step 605).

The image is sampled through conversion, for example, digital processing converts the image of the display device to the image of the inspection equipment independently, and the luminance or color space (CIELAB) employed by the CIE International Commissioner The results of the inspection are obtained objectively and do not affect the defects of the environment and equipment (process 607).

FIG. 7 generates the flow of the front processing step in the recognition flow in step 56 shown in FIG. 5 and includes the following steps.

Start of front treatment process (step 701)

The processing step further includes detection of the image edge, and an accurate image edge can be obtained, thereby reducing the error of detection (step 703).

Making the edge compensation, that is, correcting the width and image brightness of the edge as appropriate, describes the effect by the edge (step 705).

Correct the image appropriately, and adjust it as needed if the image which requires each inspection apparatus differs (step 707).

Correction of environmental parameters results in, for example, correction of environmental light sources, cancellation of environmental influences, and more accurate inspection results (step 709).

8 is a flow of the estimation process of the image information in the generation of the recognition flow shown in FIG. 5 according to the present invention.

Start of estimation process of image information (step 801)

The visual adaptation of the light adaptation of the naked eye is used to adjust the light adaptation to reflect the response of different luminance of the naked eye (step 803).

Channel Decomposition means that the image of the channel having different image analysis is generated by test information (channel) having different image analysis. For example, pyramid decomposition, cortex transform, etc. Or using only one channel (step 805).

Contrast Computation finds the contrast of the various channel images (step 807).

The Contrast Sensitivity Filter has different contrast sensitivity in different environments and spatial frequencies, so that the result is corrected by contrast in step 807 (step 809).

The correction of the oblique effect is designed by the difference in the visual angle of the inclination angle of the people, and corrects the result of the step 809 with the inclination in different directions. (Step 811)

Aperture correction relates to a method and apparatus for detecting a flat panel display by a vision model. The imaging system generates a detection level and a numerical value through a recognition flow, and thus the quality of the display device. In consideration of the entire panel, it is possible to reliably know that all panel quality is good or bad by the direct detection method.

While the present invention is certainly a rare invention, it has industrial applicability, novelty and inventiveness, and the patented elements of the invention are completely corrected by the difference between the inclination of the image center (visual focus) and the edge of the image. (Step 813)

By correcting each of the above parameters, the object of estimating image information from different reference values is achieved, and a more accurate detection index is obtained.

As mentioned above, since this invention conforms to a visual model case, it applies according to a patent law. Examined in detail, fast patents are granted, thereby protecting the rights of inventors. The foregoing descriptions are merely preferred embodiments of the present invention, and are not limited to the claims of the present invention. Therefore, it is stated that all changes in the equivalent structure according to the specification and illustration of the present invention are included in the patent scope of the present invention.

The present invention relates to a method and apparatus for detecting a flat panel display by a visual model, wherein an image is taken by a photographing system, and a detection degree (JND map) and an index map (JND value) are generated by the visual model. By judging the quality of the display device, it is possible to consider the whole panel and to know whether the quality of the entire panel is good or bad by the direct inspection method.

Claims (22)

A method of detecting a flat panel display by a visual model, An image generation step of the test panel for capturing an image of the test panel by the imaging system, A generation step of simulating the reference image by the background, Generating, by the difference recognition generation system, a recognition flow between the image of the test panel and the reference image; Combining the test information generated by the recognition flow to generate a degree of detection; Obtaining a detection value by mathematical means from test information generated by the recognition flow; And determining the quality of the display device according to the detection degree and the detection value. The method of claim 1, And the imaging system is a CCD element or a photosensitive element of a CMOS. The method of claim 1, And after the image generating step of the test panel, adjusting the captured image, wherein the adjusting step includes at least a step of adjusting a test panel image correcting the image deviation of the test panel and removing noise. A method of detecting a flat panel display by a visual model. The method of claim 1, wherein the reference image is a background of a defect. The method of claim 1, The generation of the recognition flow may also obtain an edge of the image of the test panel and the reference image, capture the adjustment of the test panel image, the compensation of the edge, and also simulate and correct the brightness of the detection environment. Shearing the panel image, An estimation step of the test panel image information which introduces flow parameters of light adaptation, channel analysis, contrast calculation, contrast sensitivity filter, tilt effect correction, and aperture correction; And combining the image information of the test panel to be corrected in consideration of the influence of each parameter by capturing the image information of the test panel image and the reference image in the front end processing and estimating the image information. A method of detecting a flat panel display by a visual model. The method of claim 5, Detecting an edge of the image of the test panel by the flow of the shearing step of the test panel image; Correcting an image of the test panel; And compensating for environmental parameters, such as the effect of an environmental light source on the detecting step. delete delete The method of claim 1, And correcting by introducing a viewpoint condition and a visual model into the recognition flow. The method of claim 1, A method for detecting a flat panel display by a visual model, wherein the degree of detection is a JND map. The method of claim 1, A method for detecting a flat panel display by a visual model, wherein the detection value is a JND value. The method of claim 1, And the mathematical means is obtained by obtaining an average value. The method of claim 12, And obtaining the average value by Minkowski pooling. The method for detecting a flat panel display device using a visual model according to claim 1, wherein a defect in the flat panel display device can be visually detected by the detection degree. An apparatus for detecting a flat panel display by a visual model, comprising: a generation system of difference recognition; A photographing system connected to the difference recognition generation system to capture a test panel image of a display device and transmit the image to the difference recognition generation system; A background simulation system connected to the difference recognition generation system and generating a reference image; A visual model system for introducing a visual model into the generation system of the difference recognition; And a detection information generation system connected to the difference recognition generation system and generating reference information for detection. The method of claim 15, An apparatus for detecting a flat panel display by a visual model, wherein a degree of detection is generated by a generation system of detection information. The method of claim 15, An apparatus for detecting a flat panel display by a visual model, characterized in that a detection value is generated by a generation system of detection information. The method of claim 15, And said imaging system is a CCD device or a photosensitive device of CMOS. The method of claim 15, And an output system to which the generation system of the detection information is connected. The apparatus for detecting a flat panel display by a visual model according to claim 15, wherein the generation system of the difference recognition is performed by hardware or software. The method of claim 15, And a generation system of the difference recognition is further connected to a viewpoint condition. The method of claim 15, And the generation system of the difference recognition is also connected to a model calibration apparatus.

Images