WO2018086289A1 - 显示面板自动光学检测中的背景抑制方法及检测装置 - Google Patents

显示面板自动光学检测中的背景抑制方法及检测装置 Download PDF

Info

Publication number
WO2018086289A1
WO2018086289A1 PCT/CN2017/077817 CN2017077817W WO2018086289A1 WO 2018086289 A1 WO2018086289 A1 WO 2018086289A1 CN 2017077817 W CN2017077817 W CN 2017077817W WO 2018086289 A1 WO2018086289 A1 WO 2018086289A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
band
sub
wavelet
frequency sub
Prior art date
Application number
PCT/CN2017/077817
Other languages
English (en)
French (fr)
Inventor
鲁方波
姚峰
邓标华
陈凯
沈亚非
Original Assignee
武汉精测电子集团股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 武汉精测电子集团股份有限公司 filed Critical 武汉精测电子集团股份有限公司
Priority to JP2019515974A priority Critical patent/JP6735417B2/ja
Priority to KR1020197004865A priority patent/KR102047855B1/ko
Priority to US16/342,964 priority patent/US10621701B2/en
Publication of WO2018086289A1 publication Critical patent/WO2018086289A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G02—OPTICS
    • G02F—OPTICAL DEVICES OR ARRANGEMENTS FOR THE CONTROL OF LIGHT BY MODIFICATION OF THE OPTICAL PROPERTIES OF THE MEDIA OF THE ELEMENTS INVOLVED THEREIN; NON-LINEAR OPTICS; FREQUENCY-CHANGING OF LIGHT; OPTICAL LOGIC ELEMENTS; OPTICAL ANALOGUE/DIGITAL CONVERTERS
    • G02F1/00—Devices or arrangements for the control of the intensity, colour, phase, polarisation or direction of light arriving from an independent light source, e.g. switching, gating or modulating; Non-linear optics
    • G02F1/01—Devices or arrangements for the control of the intensity, colour, phase, polarisation or direction of light arriving from an independent light source, e.g. switching, gating or modulating; Non-linear optics for the control of the intensity, phase, polarisation or colour 
    • G02F1/13—Devices or arrangements for the control of the intensity, colour, phase, polarisation or direction of light arriving from an independent light source, e.g. switching, gating or modulating; Non-linear optics for the control of the intensity, phase, polarisation or colour  based on liquid crystals, e.g. single liquid crystal display cells
    • G02F1/1306—Details
    • G02F1/1309—Repairing; Testing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/10—Image enhancement or restoration using non-spatial domain filtering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/40—Image enhancement or restoration using histogram techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/73—Deblurring; Sharpening
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/90—Dynamic range modification of images or parts thereof
    • G06T5/92—Dynamic range modification of images or parts thereof based on global image properties
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10004—Still image; Photographic image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20048—Transform domain processing
    • G06T2207/20064—Wavelet transform [DWT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30121—CRT, LCD or plasma display

Definitions

  • the invention belongs to the technical field of automatic optical defect detection, and more particularly to a background suppression method and a detection device in automatic optical detection of a display panel.
  • TFT-LCD Thin Film Transistor Liquid Crystal Display
  • the Mura defect has low contrast and blurred borders. Moreover, due to the physical structure of the display panel itself, the defect image collected by a high-definition device such as a CCD camera generates regularly arranged mutually perpendicular texture background information, which is usually associated with Mura. The integration of defects further increases the difficulty of machine vision detection. How to suppress the texture background without affecting the original features of the Mura defect has become the key to the success of detecting the Mura defect.
  • the invention name is: a filtering method in automatic optical detection of a display panel
  • the publication number is CN201310004940.3
  • the texture background is suppressed by a Gabor filter method, and the method uniformly views the texture background.
  • the noise is filtered, and the image is subjected to multiple frequencies and multiple directions of filtering convolution, thereby filtering out the texture background in each direction, thereby achieving the purpose of enhancing defects.
  • the method is not removing Evening the background also weakens the Mura defect contrast itself, which seriously affects the segmentation and classification of defects in image post-processing.
  • the object of the present invention is to provide a background suppression method and a detection device for automatic optical detection of a display panel, which aim to solve the problem that the prior art suppresses the background contrast and weakens the defect contrast, resulting in poor texture suppression.
  • Technical problem is to provide a background suppression method and a detection device for automatic optical detection of a display panel, which aim to solve the problem that the prior art suppresses the background contrast and weakens the defect contrast, resulting in poor texture suppression.
  • the invention provides a background suppression method in automatic optical detection of a display panel, comprising the following steps:
  • S3 performing coefficient smoothing processing on the high frequency sub-bands of each direction in multiple directions, and performing contrast enhancement processing on each low frequency sub-band;
  • S4 Perform wavelet reconstruction on the processed high frequency sub-band and the processed low frequency sub-band to obtain a defect image after background suppression.
  • the image is a solid color image.
  • the solid color image is a pure white image, or a pure gray image, or a pure red image, or a pure green image, or a pure blue image.
  • the J-level wavelet decomposition of the image is performed to obtain a high-frequency sub-band and a low-frequency sub-band total of 3*J+1;
  • the high-frequency sub-band includes: a horizontal sub-band H, a vertical sub-band V and
  • the low frequency subband includes: an approximate subband A, wherein the jth wavelet subband is: ⁇ and ⁇ are scale function and wavelet function, respectively, f is the wavelet approximation subband of j-1 level, and the approximate subband is also called low frequency subband.
  • step S3 the high frequency sub-band is subjected to coefficient smoothing processing by Gaussian low-pass filtering; wherein the high frequency sub-bands of different stages adopt different Gaussian filtering kernel parameters, the Gaussian filtering
  • imgW and imgH are the width and height of the original image
  • sizeW (j) and sizeH (j) are the width and height parameters of the Gaussian filter kernel corresponding to the j-th wavelet subband
  • sigmaW (j) and sigmaH (j) are The standard deviation parameter of the Gaussian filter kernel corresponding to the j-th wavelet subband in the horizontal and vertical directions.
  • Gaussian low-pass filtered images are used. among them, Gaussian (j) is a Gaussian low-pass filter corresponding to the j-th wavelet subband, and ** is a filter convolution operation.
  • step S3 contrast enhancement processing is performed on each of the low frequency sub-bands by a histogram equalization method.
  • step S4 the defect image is:
  • m and n are the width and height of the jth wavelet subband image.
  • the invention further provides a display panel automatic optical detecting device, comprising a light source, a camera group and an image collecting and processing unit interacting with the camera group, the image collecting and processing unit collecting the image data of the display panel, and extracting Before the defect information of the image data, the background data suppression processing is performed on the image data by using the above technical solution.
  • a display panel automatic optical detecting device comprising a light source, a camera group and an image collecting and processing unit interacting with the camera group, the image collecting and processing unit collecting the image data of the display panel, and extracting Before the defect information of the image data, the background data suppression processing is performed on the image data by using the above technical solution.
  • the image acquisition and processing unit includes:
  • the wavelet decomposition and processing module is configured to obtain a series of high frequency sub-bands and low frequency sub-bands by performing multi-level wavelet decomposition on the image data; and perform coefficient smoothing processing on the high frequency sub-bands of each direction in multiple directions, Each level of low frequency sub-bands is subjected to contrast enhancement processing;
  • the wavelet reconstruction module is configured to perform wavelet reconstruction on the processed high frequency sub-band and the processed low frequency sub-band to obtain defect image data after background suppression.
  • the present invention has the following technical advantages compared with the prior art:
  • the present invention can be applied to the detection of various specifications and sizes of various types of Mura defects in the field of liquid crystal display, and has high versatility.
  • the present invention does not require any reference samples, and the method parameters can be adaptively adjusted, and the adaptation and robustness are strong.
  • the present invention performs multi-scale multi-resolution decomposition on images, and performs texture suppression and image enhancement on the decomposed high-frequency and low-frequency sub-bands respectively, and can maintain the contrast of the original defects while suppressing the background texture, and texture suppression. The effect is good.
  • FIG. 1 is a flow chart showing an implementation of a background suppression method in automatic optical detection of a display panel according to the present invention
  • FIG. 2 is a schematic diagram of wavelet decomposition in a background suppression method for automatic optical detection of a display panel according to the present invention
  • Figure 3 (a) is an original image with a drop of Mura defect
  • Figure 3 (b) is the background suppression of Figure 3 (a) after the background suppression method provided by the present invention Image;
  • Figure 4 (a) is an original image with horizontal light line defects
  • Fig. 4(b) is an image of Fig. 4(a) after background suppression by the background suppression method provided by the present invention.
  • the invention provides a background suppression method for automatic optical detection of a display panel; the method performs multi-scale multi-directional wavelet decomposition on the display panel image, and suppresses and enhances the high-frequency and low-frequency sub-bands respectively after wavelet decomposition, and finally
  • the defect image after texture background suppression is obtained by wavelet reconstruction, which solves the problem that the traditional filtering method reduces the defect contrast while suppressing the background.
  • Embodiments of the present invention provide an automatic optical detection device for a display panel, including a light source, a camera group, and an image acquisition and processing unit that interacts with the camera group, and the image acquisition and processing unit further extracts image defect information of the display panel before
  • the process of performing background suppression processing on the collected image includes the following steps:
  • the image acquisition and processing unit includes a wavelet decomposition and processing module and a wavelet reconstruction module.
  • the wavelet decomposition and processing module is configured to perform multi-level wavelet decomposition on the image data to obtain a series of high frequency sub-bands and low frequency sub-bands; and perform coefficient smoothing processing on the high frequency sub-bands of each direction in multiple directions, Each level of the low frequency sub-band performs contrast enhancement processing;
  • the wavelet reconstruction module is configured to perform wavelet reconstruction on the processed high frequency sub-band and the processed low frequency sub-band to obtain defect image data after background suppression.
  • the invention provides a simple and efficient display panel background suppression method, which can maintain the contrast of the original defect while suppressing the background texture, and can overcome the traditional filtering method or the background fitting method in terms of background suppression and defect retention.
  • multi-scale multi-resolution processing can overcome the shortcomings of the traditional method that only a single-scale single-resolution processing results in poor texture suppression.
  • FIG. 1 is a flow chart showing a background suppression method in automatic optical detection of a display panel according to an embodiment of the present invention, the method comprising the following steps:
  • Step S101 collecting an image of the display panel in a dot screen mode, such as white, gray, red, green, blue, etc., as shown in FIG. 3(a) and FIG. 4(a), collecting The display panel dot screen image presents a regularly arranged texture background with a substantially constant texture interval period.
  • a dot screen mode such as white, gray, red, green, blue, etc.
  • Step S102 Calculate the wavelet decomposition level J, the Gaussian filter kernel standard deviation parameter sigma, and the filter kernel size parameter size.
  • the specific calculation method is as follows:
  • Gaussian filter kernel variance and size parameters adopt adaptive calculation method. Wavelet subbands of different series can adopt different Gaussian filter kernel parameters. Specifically:
  • sizeW (j) (imgW/200)/2 j
  • sizeH (j) (imgH/200)/2 j
  • sigmaW (j) TW
  • imgW and imgH are the width and height of the original image
  • sizeW (j) and sizeH (j) are the width and height parameters of the Gaussian filter kernel corresponding to the j-th wavelet subband
  • sigmaW (j) and sigmaH (j) are The standard deviation parameter of the Gaussian filter kernel corresponding to the j-th wavelet subband in the horizontal and vertical directions.
  • Step S103 performing J-level wavelet decomposition on the image to obtain 3*J+1 wavelet low frequency and low frequency sub-bands.
  • Each wavelet image is composed of four sub-bands: approximate sub-band A, horizontal sub-band H, vertical sub-band V and diagonal sub-band D.
  • the approximate sub-band represents the basic information of the image, reflecting the overall trend of image brightness.
  • the three sub-bands H, V and D represent the high-frequency information of the image, reflecting the abrupt and detailed information of the image brightness, while the image texture background usually appears as high-frequency information in the wavelet frequency domain, and is mainly distributed in the wavelet high-frequency.
  • the Gaussian low-pass filter (filter including but not limited to Gaussian low-pass filter) can be used to suppress the wavelet coefficients reflecting the texture information in the wavelet high-frequency sub-band. In order to achieve the purpose of removing the texture background.
  • the j-th wavelet sub-band can be expressed as:
  • Step S104 The high frequency subband coefficient is suppressed.
  • Class j wavelet detail subband Perform Gaussian low-pass filtering, and the filtered image is
  • Gaussian (j) is a Gaussian low-pass filter corresponding to the j-th wavelet sub-band
  • ** is a filter convolution operation
  • Step S105 the low frequency subband coefficient is enhanced.
  • the histogram equalization enhancement is performed on the last-order low-frequency sub-band after wavelet decomposition and the sub-band after each-wavelet reconstruction.
  • the image enhancement methods include but are not limited to the histogram equalization method.
  • the wavelet low-frequency sub-band reflects the overall brightness trend of the image, while the Mura defect itself has a low contrast. If the direct reconstruction method is adopted, it will inevitably reduce the contrast of the defect while greatly reducing the contrast of the defect. Segmentation and recognition. Enhancing the wavelet low-frequency sub-band can further improve the contrast of the image, thereby facilitating segmentation and recognition of defects.
  • Step S106 performing wavelet reconstruction on the high frequency sub-band after coefficient suppression and the low frequency sub-band after coefficient enhancement to obtain a defect image after background suppression.
  • m and n are the width and height of the jth wavelet subband image, and other parameters are the same as above.
  • the wavelet decomposition diagram in FIG. 2 further illustrates the effectiveness of the present invention for multi-scale multi-resolution multi-directional wavelet transform for background suppression, which can separate the background texture from the image. Thereby, the texture information in the high frequency sub-band is suppressed while the basic information of the image in the low frequency sub-band is not affected.
  • the background suppression method in the automatic optical detection of the display panel provided by the embodiment of the present invention can extract many defects such as points, lines, Mura defects and the like from a complex texture background, and obtain an image with high defect contrast and uniform background distribution.
  • the display panel automatically lays a good foundation for optical inspection.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Nonlinear Science (AREA)
  • Chemical & Material Sciences (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Optics & Photonics (AREA)
  • Quality & Reliability (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Image Processing (AREA)
  • Liquid Crystal (AREA)

Abstract

一种显示面板自动光学检测中的背景抑制方法及检测装置，其中，该背景抑制方法包括下述步骤：S1：采集显示面板的纯色图像；S2：对纯色图像进行多级小波分解后获得一系列的高频子带和低频子带；S3：对每一级多个方向的高频子带进行系数平滑处理，对每一级低频子带进行对比度增强处理；S4：对处理后的高频子带和处理后的低频子带进行小波重构，获得背景抑制后的缺陷图像。通过对图像进行多尺度多分辨率分解，并对分解后的高频和低频子带分别进行纹理抑制和图像增强，在抑制背景纹理的同时能够保持原有缺陷的对比度，纹理抑制效果好。

Description

显示面板自动光学检测中的背景抑制方法及检测装置 [技术领域]
本发明属于自动光学缺陷检测技术领域，更具体地，涉及一种显示面板自动光学检测中的背景抑制方法及检测装置。
[背景技术]
由于TFT-LCD(薄膜晶体管液晶显示器)等新型显示面板的功耗低、清晰度高、辐射小等特点使其成为当前主流的显示设备，在二维以及三维信息传递过程中发挥着不可替代的作用。
TFT-LCD等新型显示面板的生产工序非常复杂，使得在实际制备过程中不可避免会产生各种各样的显示缺陷，其中以Mura缺陷最难检测。传统的依靠人眼的检测方法存在诸多不足，具有成本高、不稳定、效率低的问题。近年来，机器视觉缺陷检测方法因具有自动化程度高、鲁棒性好等特性受到广泛关注。
Mura缺陷对比度低，边界较为模糊，而且，受显示面板本身物理结构的影响，通过CCD相机等高清设备采集到的缺陷图像会产生规则排列的相互垂直的纹理背景信息，该纹理背景信息通常与Mura缺陷融为一体，从而进一步增加了机器视觉的检测难度。如何在不影响Mura缺陷原始特征的情况下对纹理背景进行抑制已成为对Mura缺陷进行检测的成功的关键。
在专利文献(发明名称为：一种显示面板自动光学检测中的滤波方法，公开号为CN201310004940.3)中，采用Gabor滤波器的方法对纹理背景进行抑制，该方法将均匀分布的纹理背景视为噪声进行滤波处理，对图像进行多个频率，多个方向滤波卷积，从而滤除各个方向上的纹理背景，进而达到增强缺陷的目的。然而，该方法在去除不 均匀背景的同时也减弱了Mura缺陷对比度本身，从而严重影响图像后处理中对缺陷的分割和分类判断。
[发明内容]
针对现有技术的缺陷，本发明的目的在于提供一种显示面板自动光学检测中的背景抑制方法及检测装置，旨在解决现有技术在背景抑制的同时减弱了缺陷对比度导致纹理抑制效果较差的技术问题。
本发明提供了一种显示面板自动光学检测中的背景抑制方法，包括下述步骤：
S1：采集显示面板的图像；
S2：对所述图像进行多级小波分解后获得一系列的高频子带和低频子带；
S3：对每一级多个方向的高频子带进行系数平滑处理，对每一级低频子带进行对比度增强处理；
S4：对处理后的高频子带和处理后的低频子带进行小波重构，获得背景抑制后的缺陷图像。
进一步地，所述图像为纯色图像。
更进一步地，所述纯色图像为纯白色的图像，或者为纯灰色的图像，或者为纯红色的图像，或者为纯绿色的图像，或者为纯蓝色的图像。
更进一步地，在步骤S1之后且在步骤S2之前，还包括下述步骤：获得小波分解级数J；其中，小波分解级数J＝ceil(log2(TW+TH))或J＝ceil(log2((TW+TH)/2))，TW为点屏图像的水平纹理周期，TH为垂直纹理周期，ceil为取大于或等于某个数的最小整数。
更进一步地，在步骤S2中，对图像进行J级小波分解后得到高频子带和低频子带共3*J+1个；高频子带包括：水平子带H、垂直子 带V和对角子带D，低频子带包括：近似子带A，其中第j级小波子带为：
Figure PCTCN2017077817-appb-000001
Figure PCTCN2017077817-appb-000002
φ和ψ分别为尺度函数和小波函数，f为j-1级小波近似子带，近似子带也称为低频子带，
Figure PCTCN2017077817-appb-000003
分别为第j级小波分解的近似子带、水平子带、垂直子带以及对角子带，j＝1,2,…,J。
更进一步地，在步骤S3中，采用高斯低通滤波的方法对所述高频子带进行系数平滑处理；其中，不同级数的高频子带采用不同的高斯滤波核参数，所述高斯滤波核参数为：sizeW(j)＝(imgW/200)/2j，sizeH(j)＝(imgH/200)/2j，sigmaW(j)＝TW，sigmaH(j)＝TH，j＝1…J。
其中imgW和imgH为原始图像的宽和高，sizeW(j)和sizeH(j)为第j级小波子带对应的高斯滤波核的宽和高尺寸参数，sigmaW(j)和sigmaH(j)为第j级小波子带对应的高斯滤波核在水平和垂直方向上的标准差参数。
更进一步地，采用高斯低通滤波后的图像
Figure PCTCN2017077817-appb-000004
其中，
Figure PCTCN2017077817-appb-000005
Gaussian(j)为第j级小波子带对应的高斯低通滤波器，**为滤波卷积操作。
更进一步地，在步骤S3中，采用直方图均衡方法对每一级低频子带进行对比度增强处理。
更进一步地，在步骤S4中，缺陷图像为：
Figure PCTCN2017077817-appb-000006
其中
Figure PCTCN2017077817-appb-000007
为低频子带增强后的图像，
Figure PCTCN2017077817-appb-000008
为高频子带系数抑制后的图像，m、n为第j级小波子带图像的宽和高。
本发明另外提供一种显示面板自动光学检测装置，包括光源、相机组以及与该相机组进行交互的图像采集与处理单元，该图像采集与处理单元采集到显示面板的图像数据后，并且在提取该图像数据的缺陷信息之前，还采用上述技术方案对该图像数据进行背景抑制处理。
进一步地，该图像采集与处理单元包括：
小波分解与处理模块，用于对该图像数据进行多级小波分解后获得一系列的高频子带和低频子带；并对每一级多个方向的高频子带进行系数平滑处理，对每一级低频子带进行对比度增强处理；
小波重构模块，用于对处理后的高频子带和处理后的低频子带进行小波重构，获得背景抑制后的缺陷图像数据。
通过本发明所构思的以上技术方案，与现有技术相比，本发明具有如下技术优点：
(1)本发明可以适用于液晶显示领域目前市面上所有规格尺寸，各种形态Mura缺陷的检测，通用性较强。
(2)本发明不需要任何参考样本，方法参数可以自适应调整，自适应和鲁棒性较强。
(3)本发明对图像进行多尺度多分辨率分解，并对分解后的高频和低频子带分别进行纹理抑制和图像增强，在抑制背景纹理的同时能够保持原有缺陷的对比度，纹理抑制效果好。
[附图说明]
图1为本发明提出的显示面板自动光学检测中的背景抑制方法的实现流程图；
图2为本发明提出的显示面板自动光学检测中的背景抑制方法中小波分解示意图；
图3(a)为带有滴落Mura缺陷的原始图像；
图3(b)为图3(a)经过本发明提供的背景抑制方法进行背景抑制后 的图像；
图4(a)为带有水平淡线缺陷的原始图像；
图4(b)为图4(a)经过本发明提供的背景抑制方法进行背景抑制后的图像。
[具体实施方式]
为了使本发明的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
本发明提供了一种显示面板自动光学检测中的背景抑制方法；该方法对显示面板图像进行多尺度多方向小波分解，并对小波分解后的高频和低频子带分别进行抑制和增强，最终经小波重构得到纹理背景抑制后的缺陷图像，从而解决传统滤波方法在背景抑制的同时减弱了缺陷对比度的问题。
本发明实施例提供一种显示面板自动光学检测装置，包括光源、相机组以及与该相机组进行交互的图像采集与处理单元，该图像采集与处理单元在提取显示面板的图像缺陷信息之前，还包括对采集到的图像进行背景抑制处理的过程，包括下述步骤：
(1)采集显示面板不同点屏模式下的图像(即一种点屏模式下的图像)。
(2)对图像进行多级小波分解得到一系列小波高频和低频子带。
(3)高频小波系数抑制：对每一级多个方向的高频子带运用高斯低通滤波方法进行系数平滑。
(4)低频子带系数增强：对每一级低频子带进行对比度增强。
(5)对系数抑制后的高频子带和系数增强后的低频子带进行小波重构得到背景抑制后的缺陷图像。
上述实施例中，该图像采集与处理单元包括小波分解与处理模块和小波重构模块。该小波分解与处理模块用于对该图像数据进行多级小波分解后获得一系列的高频子带和低频子带；并对每一级多个方向的高频子带进行系数平滑处理，对每一级低频子带进行对比度增强处理；该小波重构模块用于对处理后的高频子带和处理后的低频子带进行小波重构，获得背景抑制后的缺陷图像数据。
本发明提供了一种简单高效的显示面板背景抑制方法，在抑制背景纹理的同时能够保持原有缺陷的对比度，能够克服传统的滤波方法或者背景拟合方法在背景抑制和缺陷保持方面不能兼顾的缺点，同时，多尺度多分辨率处理能够克服传统方法仅在单一尺度单一分辨率处理而导致纹理抑制效果较差的缺点。
图1示出了本发明实施例提供的一种显示面板自动光学检测中的背景抑制方法流程图，该方法包括以下步骤：
步骤S101：采集显示面板一种点屏模式下的图像，如白、灰、红、绿、蓝等模式下的点屏图像，如图3(a)和图4(a)所示，采集到的显示面板点屏图像呈现规则排列的纹理背景，且纹理间隔周期基本恒定。
步骤S102：计算小波分解级数J，高斯滤波核标准差参数sigma以及滤波核尺寸参数size，具体计算方法如下：
(1)小波分解级数J的计算：设点屏图像的水平纹理周期为TW，垂直纹理周期为TH，水平或者垂直纹理周期可以直接从点屏图像中根据纹理周期来确定，则小波分解级数J表示为J＝ceil(log2((TW+TH)/2))，其中ceil为取大于或等于某个数的最小整数。比如，图3(a)的水平和垂直纹理周期为9，则小波分解级数J＝4，图4(a)的水平和垂直纹理周期为6，则小波分解级数J＝3。需要说明的是，小波分解级数J也可以用公式J＝ceil(log2(TW+TH))计算。
(2)高斯滤波核方差和尺寸参数采用自适应计算方法，不同级数的小波子带可以采用不同的高斯滤波核参数。具体为：
sizeW(j)＝(imgW/200)/2j，sizeH(j)＝(imgH/200)/2j，sigmaW(j)＝TW，sigmaH(j)＝TH，j＝1…J。
其中imgW和imgH为原始图像的宽和高，sizeW(j)和sizeH(j)为第j级小波子带对应的高斯滤波核的宽和高尺寸参数，sigmaW(j)和sigmaH(j)为第j级小波子带对应的高斯滤波核在水平和垂直方向上的标准差参数。
步骤S103：对图像进行J级小波分解得到小波高频和低频子带共3*J+1个。每一级图像经过小波变换后由近似子带A、水平子带H、垂直子带V以及对角子带D这四个子带组成，近似子带表示图像的基本信息，反映了图像亮度整体变化趋势；H、V和D三个子带表示了图像的高频信息，反映了图像亮度的突变和细节信息，而图像纹理背景在小波频率域中通常表现为高频信息，并主要分布在小波高频子带中，且对应小波系数值较大，因此，可以在小波高频子带中运用高斯低通滤波器(滤波器包括但不限于高斯低通滤波器)对反映纹理信息的小波系数进行抑制，从而达到去除纹理背景的目的。
其中第j级小波子带可表示为：
Figure PCTCN2017077817-appb-000009
Figure PCTCN2017077817-appb-000010
Figure PCTCN2017077817-appb-000011
Figure PCTCN2017077817-appb-000012
φ和ψ分别为尺度函数和小波函数，f为j-1级小波近似子带，近似子带也称为低频子带，
Figure PCTCN2017077817-appb-000013
分别为第j级小波分解的近似子带、水平子带、垂直子带以及对角子带，j＝1,2,…,J。
步骤S104：高频子带系数抑制。
对第j级小波细节子带
Figure PCTCN2017077817-appb-000014
进行高斯低通滤波，滤波后的 图像为
Figure PCTCN2017077817-appb-000015
Figure PCTCN2017077817-appb-000016
Figure PCTCN2017077817-appb-000017
其中Gaussian(j)为第j级小波子带对应的高斯低通滤波器，**为滤波卷积操作。
步骤S105：低频子带系数增强。
对小波分解后的最后一级低频子带以及每一级小波重建后的子带进行直方图均衡增强，图像增强方法包括但不限于直方图均衡方法。小波低频子带反映了图像整体亮度变化趋势，而Mura缺陷本身对比度又比较低，如果采用直接重建的方法势必会造成在纹理抑制的同时也降低了缺陷的对比度，极大的影响后续方法对缺陷的分割和识别。对小波低频子带进行增强可以进一步提高图像的对比度，从而有利于缺陷的分割和识别。
步骤S106：对系数抑制后的高频子带和系数增强后的低频子带进行小波重构得到背景抑制后的缺陷图像。
Figure PCTCN2017077817-appb-000018
其中
Figure PCTCN2017077817-appb-000019
为低频子带增强后的图像，
Figure PCTCN2017077817-appb-000020
为高频子带系数抑制后的图像，m、n为第j级小波子带图像的宽和高，其它参数同上。
结合图3和图4可以得出，经过本发明的背景抑制方法，图像纹理背景基本得到较好的去除，同时缺陷部分也得以增强。图2中小波分解示意图也进一步说明了本发明运用多尺度多分辨率多方向的小波变换进行背景抑制的有效性，它可以将背景纹理从图像中分离出来， 从而在抑制高频子带中纹理信息的同时不影响低频子带中图像的基本信息。
本发明实施例提供的显示面板自动光学检测中的背景抑制方法能够将诸多缺陷如点、线、Mura缺陷等从复杂的纹理背景中提取出来，得到缺陷对比度高，背景分布均匀的图像，为实现显示面板自动光学检测打下良好的基础。
本领域的技术人员容易理解，以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明的保护范围之内。

Claims (11)

  1. 一种显示面板自动光学检测中的背景抑制方法，其特征在于，包括下述步骤：
    S1：采集显示面板的图像；
    S2：对所述图像进行多级小波分解后获得一系列的高频子带和低频子带；
    S3：对每一级多个方向的高频子带进行系数平滑处理，对每一级低频子带进行对比度增强处理；
    S4：对处理后的高频子带和处理后的低频子带进行小波重构，获得背景抑制后的缺陷图像。
  2. 如权利要求1所述的背景抑制方法，其特征在于，所述图像为纯色图像。
  3. 如权利要求2所述的背景抑制方法，其特征在于，所述纯色图像为纯白色的图像，或者为纯灰色的图像，或者为纯红色的图像，或者为纯绿色的图像，或者为纯蓝色的图像。
  4. 如权利要求1所述的背景抑制方法，其特征在于，在步骤S1之后且在步骤S2之前，还包括下述步骤：获得小波分解级数J；
    其中，小波分解级数J＝ceil(log2(TW+TH))或J＝ceil(log2((TW+TH)/2))，TW为图像的水平纹理周期，TH为垂直纹理周期，ceil为取大于或等于某个数的最小整数。
  5. 如权利要求4所述的背景抑制方法，其特征在于，在步骤S2中，对图像进行J级小波分解后得到高频子带和低频子带共3*J+1个；
    高频子带包括：水平子带H、垂直子带V和对角子带D，低频子带包括：近似子带A，其中第j级小波子带为：
    Figure PCTCN2017077817-appb-100001
    Figure PCTCN2017077817-appb-100002
    Figure PCTCN2017077817-appb-100003
    Figure PCTCN2017077817-appb-100004
    φ和ψ分别为尺度函数和小波函数，f为j-1级小波近似子带，近似子带也称为低频子带，
    Figure PCTCN2017077817-appb-100005
    分别为第j级小波分解的近似子带、水平子带、垂直子带以及对角子带，j＝1,2,…,J。
  6. 如权利要求1-5任一项所述的背景抑制方法，其特征在于，在步骤S3中，采用高斯低通滤波的方法对所述高频子带进行系数平滑处理；其中，不同级数的高频子带采用不同的高斯滤波核参数，所述高斯滤波核参数为：sizeW(j)＝(imgw/200)/2j，sizeH(j)＝(imgH/200)/2j，sigmaW(j)＝TW，sigmaH(j)＝TH，j＝1…J；imgW和imgH为原始图像的宽和高，sizeW(j)和sizeH(j)为第j级小波子带对应的高斯滤波核的宽和高尺寸参数，sigmaW(j)和sigmaH(j)为第j级小波子带对应的高斯滤波核在水平和垂直方向上的标准差参数。
  7. 如权利要求6所述的背景抑制方法，其特征在于，采用高斯低通滤波后的图像
    Figure PCTCN2017077817-appb-100006
    其中，
    Figure PCTCN2017077817-appb-100007
    Gaussian(j)为第j级小波子带对应的高斯低通滤波器，**为滤波卷积操作。
  8. 如权利要求1-5任一项所述的背景抑制方法，其特征在于，在步骤S3中，采用直方图均衡方法对每一级低频子带进行对比度增强处理。
  9. 如权利要求8所述的背景抑制方法，其特征在于，在步骤S4中，缺陷图像为：
    Figure PCTCN2017077817-appb-100008
    其中
    Figure PCTCN2017077817-appb-100009
    为低频子带增强后的图像，
    Figure PCTCN2017077817-appb-100010
    为高频子带系数抑制后的图像，m、n为第j级小波子带图像的宽和高。
  10. 一种显示面板自动光学检测装置，其特征在于，该检测装置包括光源、相机组以及与该相机组进行交互的图像采集与处理单元，其特征在于，该图像采集与处理单元采用如权利要求1-5任一项所述的方法对从显示面板上采集的图像数据进行背景抑制处理。
  11. 如权利要求10所述的检测装置，其特征在于，该图像采集与处理单元包括：
    小波分解与处理模块，用于对该图像数据进行多级小波分解后获得一系列的高频子带和低频子带；并对每一级多个方向的高频子带进行系数平滑处理，对每一级低频子带进行对比度增强处理；
    小波重构模块，用于对处理后的高频子带和处理后的低频子带进行小波重构，获得背景抑制后的缺陷图像数据。
PCT/CN2017/077817 2016-11-10 2017-03-23 显示面板自动光学检测中的背景抑制方法及检测装置 WO2018086289A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
JP2019515974A JP6735417B2 (ja) 2016-11-10 2017-03-23 表示パネルの自動光学検出におけるバックグラウンド抑制方法及び検出装置
KR1020197004865A KR102047855B1 (ko) 2016-11-10 2017-03-23 디스플레이 패널 자동광학 탐지 중의 배경 억제 방법 및 탐지 장치
US16/342,964 US10621701B2 (en) 2016-11-10 2017-03-23 Background suppression method and detecting device in automatic optical detection of display panel

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610990996.4A CN106556940B (zh) 2016-11-10 2016-11-10 一种tft-lcd屏自动光学检测中的背景抑制方法
CN2016109909964 2016-11-10

Publications (1)

Publication Number Publication Date
WO2018086289A1 true WO2018086289A1 (zh) 2018-05-17

Family

ID=58444075

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/077817 WO2018086289A1 (zh) 2016-11-10 2017-03-23 显示面板自动光学检测中的背景抑制方法及检测装置

Country Status (5)

Country Link
US (1) US10621701B2 (zh)
JP (1) JP6735417B2 (zh)
KR (1) KR102047855B1 (zh)
CN (1) CN106556940B (zh)
WO (1) WO2018086289A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109949292A (zh) * 2019-03-20 2019-06-28 湖南镭目科技有限公司 一种带钢表面缺陷的检测方法及装置
CN111462065A (zh) * 2020-03-27 2020-07-28 浙江杜比医疗科技有限公司 超声、红外序列图像融合的乳腺癌检测方法及其***

Families Citing this family (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017034976A1 (en) * 2015-08-21 2017-03-02 Cummins Filtration Ip, Inc. High speed rotating crankcase ventilation filter media and media pack
DE112016003302T5 (de) 2015-08-28 2018-05-24 Cummins Filtration Ip, Inc. Rotierendes Koaleszerelement mit gerichteter Flüssigkeitsdrainage und gerichtetem Gasauslass
CN107742291B (zh) * 2017-10-20 2020-06-05 凌云光技术集团有限责任公司 一种光伏玻璃的缺陷检测方法及装置
US11113796B2 (en) * 2018-02-09 2021-09-07 Delta Electronics, Inc. Image enhancement circuit and method thereof
CN109801231B (zh) * 2018-12-25 2022-11-29 电子科技大学中山学院 一种电泳电子纸检测设备的图像处理方法
CN109828394A (zh) * 2019-03-22 2019-05-31 惠科股份有限公司 一种显示面板的检测方法和检测机台
CN110227968B (zh) * 2019-05-16 2020-08-21 中国工程物理研究院激光聚变研究中心 基于小波变换的磁流变去除函数寄生条纹消除方法及装置
CN111507974B (zh) * 2020-04-22 2023-10-24 广州柔视智能科技有限公司 缺陷检测方法、装置、缺陷检测设备及计算机存储介质
CN111797939A (zh) * 2020-07-20 2020-10-20 天津中德应用技术大学 基于小波分析的无人图书馆深度学习智能识别***与方法
CN112053299B (zh) * 2020-08-27 2022-07-19 杭州安旭生物科技股份有限公司 一种基于胶体金免疫层析法的图像处理方法
WO2022075825A1 (en) * 2020-10-07 2022-04-14 Samsung Electronics Co., Ltd. Methods and systems for denoising media frames captured in low-light environment
CN113034400A (zh) * 2021-04-07 2021-06-25 深圳鱼亮科技有限公司 一种基于无线图像传感器阵列的图像降噪方法
CN113822884B (zh) * 2021-11-22 2022-06-28 中导光电设备股份有限公司 基于小波变换的mura缺陷检测方法和***
CN115908207A (zh) * 2022-09-28 2023-04-04 南京颖图电子技术有限公司 一种金属表面下复杂背景的图像特征增强处理方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006133196A (ja) * 2004-11-09 2006-05-25 Seiko Epson Corp 画素ムラ欠陥検出方法、画素ムラ欠陥検出装置、画素ムラ欠陥検出プログラムおよびこのプログラムを記録した記録媒体。
CN1928535A (zh) * 2006-09-07 2007-03-14 哈尔滨工业大学 一种基于机器视觉的液晶显示器斑痕缺陷检测方法与***
CN103674975A (zh) * 2013-09-09 2014-03-26 中华人民共和国四川出入境检验检疫局 基于Gabor滤波与C-V模型分割的LCD Mura缺陷机器视觉检测方法
CN103913858A (zh) * 2013-01-07 2014-07-09 北京兆维电子(集团)有限责任公司 一种tft-lcd屏自动光学检测中的滤波方法

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2002174564A (ja) * 2000-12-06 2002-06-21 Hitachi Ltd ディスプレイ装置の画質定量評価方法並びにディスプレイ装置
JP2004212392A (ja) * 2002-12-17 2004-07-29 Okayama Prefecture 表面検査方法
US11197651B2 (en) * 2007-03-08 2021-12-14 Sync-Rx, Ltd. Identification and presentation of device-to-vessel relative motion
KR101702887B1 (ko) * 2007-04-18 2017-02-06 마이크로닉 마이데이타 에이비 무라 검출 및 계측을 위한 방법 및 장치
CN102158701B (zh) * 2011-04-19 2012-07-25 湖南大学 一种基于压缩感知理论的分类量化图像编码方法
CN104200452B (zh) * 2014-09-05 2017-02-15 西安电子科技大学 基于谱图小波变换的红外与可见光图像融合方法及其装置
CN106067164B (zh) * 2016-05-26 2018-11-20 哈尔滨工业大学 基于自适应小波域处理的彩色图像对比度增强算法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006133196A (ja) * 2004-11-09 2006-05-25 Seiko Epson Corp 画素ムラ欠陥検出方法、画素ムラ欠陥検出装置、画素ムラ欠陥検出プログラムおよびこのプログラムを記録した記録媒体。
CN1928535A (zh) * 2006-09-07 2007-03-14 哈尔滨工业大学 一种基于机器视觉的液晶显示器斑痕缺陷检测方法与***
CN103913858A (zh) * 2013-01-07 2014-07-09 北京兆维电子(集团)有限责任公司 一种tft-lcd屏自动光学检测中的滤波方法
CN103674975A (zh) * 2013-09-09 2014-03-26 中华人民共和国四川出入境检验检疫局 基于Gabor滤波与C-V模型分割的LCD Mura缺陷机器视觉检测方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
BI, XIN ET AL.: "Machine Vision Inspection Method of Mura Defect for TFT-LCD", JOURNAL OF MECHANICAL ENGINEERING, vol. 46, no. 12, 30 June 2010 (2010-06-30) *
LIU, XINGMIAO ET AL.: "A Nonlinear Algorithm of Infrared Image Enhancement Based on Wavelet Transform", INFRARED TECHNOLOGY, vol. 31, no. 12, 31 December 2009 (2009-12-31), pages 708 and 709 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109949292A (zh) * 2019-03-20 2019-06-28 湖南镭目科技有限公司 一种带钢表面缺陷的检测方法及装置
CN109949292B (zh) * 2019-03-20 2023-11-14 湖南镭目科技有限公司 一种带钢表面缺陷的检测方法及装置
CN111462065A (zh) * 2020-03-27 2020-07-28 浙江杜比医疗科技有限公司 超声、红外序列图像融合的乳腺癌检测方法及其***
CN111462065B (zh) * 2020-03-27 2023-06-27 浙江杜比医疗科技有限公司 超声、红外序列图像融合的乳腺癌检测方法及其***

Also Published As

Publication number Publication date
CN106556940B (zh) 2019-11-19
KR102047855B1 (ko) 2019-11-22
KR20190031295A (ko) 2019-03-25
US20200058102A1 (en) 2020-02-20
JP6735417B2 (ja) 2020-08-05
JP2019536128A (ja) 2019-12-12
US10621701B2 (en) 2020-04-14
CN106556940A (zh) 2017-04-05

Similar Documents

Publication Publication Date Title
WO2018086289A1 (zh) 显示面板自动光学检测中的背景抑制方法及检测装置
JP2019536128A5 (zh)
CN105654436B (zh) 一种基于前景背景分离的背光图像增强去噪方法
US10168526B2 (en) Cell contour formation apparatus and method of the same, and non-transitory computer readable storage medium storing a cell contour formation program
WO2020124873A1 (zh) 图像处理方法
CN104616255B (zh) 基于乳腺x线图像的自适应增强方法
CN103295225B (zh) 弱光照条件下的列车转向架边缘检测方法
CN109801231A (zh) 一种电泳电子纸检测设备的图像处理方法
CN112991199A (zh) 一种基于残差密集网络的图像高低频分解噪声去除方法
CN109993703A (zh) 多尺度图像降噪方法及装置
Liu et al. Spectral bayesian uncertainty for image super-resolution
CN101957912B (zh) 基于多尺度各向异性扩散的人脸光照不变图像的获取方法
CN111242876B (zh) 低对比度图像增强方法、装置及计算机可读存储介质
CN110717869B (zh) 一种水下浑浊图像清晰化方法
CN110706180A (zh) 一种极暗图像视觉质量提升方法、***、设备及介质
TWI460667B (zh) 模糊指紋影像重建方法
Zhang et al. A LCD screen Mura defect detection method based on machine vision
CN106447616A (zh) 一种实现小波去噪的方法和装置
CN105894478B (zh) 基于***部秩特征的图像去噪方法
CN110992294B (zh) 一种类纸显示器的图像对比度提升方法
CN108133467B (zh) 基于粒计算的水下图像增强***及增强方法
Lin et al. Automatic noise removal in MR images using bilateral filtering associated with artificial neural networks
Gupta Medical Image Denoising Expending A Hybrid Recurrent Neural Network Through Support Vector Machine
Zhu et al. Three-dimensional bilateral filtering applying to DE-noise of microscopic image stack
Sorwar et al. Orthopedic X-ray Image Enhancement Based on Wavelet Transform and Histogram Matching

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17870554

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 20197004865

Country of ref document: KR

Kind code of ref document: A

ENP Entry into the national phase

Ref document number: 2019515974

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17870554

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 17870554

Country of ref document: EP

Kind code of ref document: A1