[go: up one dir, main page]

CN115131348B - Method and system for detecting textile surface defects - Google Patents

Method and system for detecting textile surface defects Download PDF

Info

Publication number
CN115131348B
CN115131348B CN202211044200.8A CN202211044200A CN115131348B CN 115131348 B CN115131348 B CN 115131348B CN 202211044200 A CN202211044200 A CN 202211044200A CN 115131348 B CN115131348 B CN 115131348B
Authority
CN
China
Prior art keywords
sliding window
image
length
point
point pair
Prior art date
Legal status (The legal status 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 status listed.)
Active
Application number
CN202211044200.8A
Other languages
Chinese (zh)
Other versions
CN115131348A (en
Inventor
唐琴
华真珍
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Haimen Ximanting Textile Co ltd
Original Assignee
Haimen Ximanting Textile Co ltd
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 Haimen Ximanting Textile Co ltd filed Critical Haimen Ximanting Textile Co ltd
Priority to CN202211044200.8A priority Critical patent/CN115131348B/en
Publication of CN115131348A publication Critical patent/CN115131348A/en
Application granted granted Critical
Publication of CN115131348B publication Critical patent/CN115131348B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • 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
    • G06T5/90—Dynamic range modification of images or parts thereof
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • 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
    • G06T2207/20081—Training; Learning
    • 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
    • G06T2207/20084—Artificial neural networks [ANN]
    • 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
    • G06T2207/30124—Fabrics; Textile; Paper

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Quality & Reliability (AREA)
  • Treatment Of Fiber Materials (AREA)
  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

The invention discloses a method for detecting surface defects of textiles, relates to the field of artificial intelligence, and is mainly used for detecting broken threads of the textiles. The method comprises the following steps: acquiring a gray level image of the surface of the textile and preprocessing the gray level image; acquiring the maximum frequency point pair length in each direction in the gray level image, and calculating the point pair period length probability in each direction; acquiring the period length in the period extending direction, setting sliding window parameters to slide the gray level image, acquiring a frequency domain space image of each sliding window image, acquiring an intersection frequency value of the sliding window images, filtering each sliding window image, calculating the contrast value of each pixel point in each sliding window, and acquiring high-contrast pixel points in each sliding window; and calculating the defect probability of each sliding window, and judging whether the sliding window image has defects according to the defect probability. According to the technical means provided by the invention, the interference of the printing texture of the textile can be overcome, the defect area can be accurately positioned, and the detection efficiency is effectively improved.

Description

一种纺织品表面缺陷的检测方法及系统A method and system for detecting surface defects of textiles

技术领域technical field

本发明涉及人工智能领域，具体涉及一种纺织品表面缺陷的检测方法及系统。The invention relates to the field of artificial intelligence, in particular to a method and system for detecting surface defects of textiles.

背景技术Background technique

在纺织品表面缺陷的检测过程中，由于纺织品自身纹理的干扰，导致检测出的缺陷不够精确，无法精确的定位出纺织品表面缺陷，本发明设计一种纺织品表面缺陷的检测方法，获取精确的缺陷位置，便于后续的纺织品修复处理。In the detection process of textile surface defects, due to the interference of the textile's own texture, the detected defects are not accurate enough, and the textile surface defects cannot be accurately located. The present invention designs a detection method for textile surface defects to obtain accurate defect locations , to facilitate the subsequent textile repair treatment.

由于纹理和颜色的干扰导致常规的阈值分割和边缘检测方法不能检测出纺织品缺陷位置。常规的滤波检测方法也无法简单的获取背景图案的频段，导致需花费大量时间来进行频段判定。Due to the interference of texture and color, conventional threshold segmentation and edge detection methods cannot detect textile defect locations. The conventional filter detection method cannot simply obtain the frequency band of the background pattern, resulting in a large amount of time needed to determine the frequency band.

发明内容Contents of the invention

本发明提供一种纺织品表面缺陷的检测方法及系统，以解决现有的问题，包括：获取纺织品表面灰度图像并进行预处理；获取灰度图像中各个方向上最大频率点对长度，计算每个方向的点对周期长度概率；获取周期延伸方向上的周期长度，设置滑窗参数对所述灰度图像进行滑窗，获取各个滑窗图像的频域空间图像，获取滑窗图像的交集频率值对每个滑窗图像进行滤波处理，计算各个滑窗内每个像素点的对比度值，获取每个滑窗中的高对比度像素点；计算每个滑窗的缺陷概率，根据缺陷概率判断滑窗图像是否存在缺陷。The present invention provides a method and system for detecting surface defects of textiles to solve the existing problems, including: acquiring the grayscale image of the surface of the textile and performing preprocessing; acquiring the length of the maximum frequency point pair in each direction in the grayscale image, and calculating each The point-to-period length probability of each direction; obtain the period length in the period extension direction, set the sliding window parameter to perform sliding window on the grayscale image, obtain the frequency domain spatial image of each sliding window image, and obtain the intersection frequency of the sliding window image Filter each sliding window image, calculate the contrast value of each pixel in each sliding window, and obtain high-contrast pixels in each sliding window; calculate the defect probability of each sliding window, and judge the sliding window according to the defect probability Window image for defects.

根据本发明提出的技术手段，通过分析图像中的点对纹理信息得到纺织品的纹理周期信息，利用周期信息设置滑窗参数，从而进行进一步的滤波处理，能够克服纺织品自身印花纹理的干扰，精确的定位出缺陷区域，从而有效提升了检测效率和生产质量。According to the technical means proposed by the present invention, the texture cycle information of the textile is obtained by analyzing the point-to-texture information in the image, and the sliding window parameters are set by using the cycle information, so as to perform further filtering processing, which can overcome the interference of the textile's own printing texture, and accurately The defect area is located, thereby effectively improving the detection efficiency and production quality.

本发明采用如下技术方案，一种纺织品表面缺陷的检测方法，包括：The present invention adopts the following technical scheme, a method for detecting surface defects of textiles, comprising:

获取纺织品表面灰度图像，并对所述灰度图像进行预处理获得梯度图像。The grayscale image of the textile surface is obtained, and the grayscale image is preprocessed to obtain a gradient image.

获取梯度图像中每一个像素点在每个方向上两两像素点形成的点对，利用各个方向上最大频率点对所对应的点对长度计算所述灰度图像每个方向的点对周期长度概率。Obtain the point pair formed by two or two pixels in each direction of each pixel in the gradient image, and calculate the point pair cycle length in each direction of the grayscale image by using the point pair length corresponding to the maximum frequency point pair in each direction probability.

将所有点对得到的周期长度概率最大值所对应的方向作为周期的延申方向，获取周期延申方向上的周期长度，根据所述周期长度以及周期延申方向设置滑窗参数。The direction corresponding to the maximum value of the cycle length probability obtained by all point pairs is used as the extension direction of the cycle, and the cycle length in the cycle extension direction is obtained, and the sliding window parameter is set according to the cycle length and the cycle extension direction.

利用设定参数的窗口对所述灰度图像进行滑窗，获取各个滑窗图像的频域空间图像，根据两两滑窗的频域空间交集得到对应两个滑窗图像的交集频率值。Sliding the grayscale image by using a window with set parameters to obtain a frequency-domain spatial image of each sliding-window image, and obtaining an intersection frequency value corresponding to two sliding-window images according to the frequency-domain spatial intersection of any pair of sliding windows.

利用每一个滑窗图像与其它滑窗图像的交集频率值分别对该滑窗图像进行滤波处理获得该滑窗图像在不同交集频率值滤波处理后的滤波图像。Filtering is performed on the sliding window image by using the intersection frequency values of each sliding window image and other sliding window images to obtain filtered images of the sliding window image after filtering at different intersection frequency values.

利用每一个滑窗图像对应的所有滤波图像中每个像素点的对比度值，计算该滑窗图像的缺陷概率，根据所述缺陷概率判断该滑窗图像是否存在缺陷。Using the contrast value of each pixel in all the filtered images corresponding to each sliding window image, calculate the defect probability of the sliding window image, and judge whether the sliding window image has a defect according to the defect probability.

进一步的，一种纺织品表面缺陷的检测方法，计算所述灰度图像每个方向的点对周期长度概率的方法如下：Further, a method for detecting surface defects of textiles, the method for calculating the probability of point-to-period length in each direction of the grayscale image is as follows:

利用起始点与终止点之间的欧氏距离计算所述灰度图像中第k行点对在每个方向上的点对长度，获取第s个方向上最大频率对应的行点对长度

Figure DEST_PATH_IMAGE001
；Use the Euclidean distance between the start point and the end point to calculate the point pair length of the k-th row point pair in each direction in the grayscale image, and obtain the row point pair length corresponding to the maximum frequency in the s-th direction
Figure DEST_PATH_IMAGE001
;

同理，计算从第k行出发与第s个方向垂直方向上的所有点对长度得到列点对长度，获取对应方向上频率最大的列点对长度

Figure 100002_DEST_PATH_IMAGE002
；Similarly, calculate the length of all point pairs starting from the k-th row and perpendicular to the s-th direction to obtain the length of the column point pair, and obtain the length of the column point pair with the highest frequency in the corresponding direction
Figure 100002_DEST_PATH_IMAGE002
;

计算第s个方向的周期长度概率的表达式为：The expression for calculating the cycle length probability in the sth direction is:

Figure DEST_PATH_IMAGE003
Figure DEST_PATH_IMAGE003

其中，

Figure 100002_DEST_PATH_IMAGE004
表示从第k行出发垂直于第s个角度方向上频率最大的列点对长度，
Figure DEST_PATH_IMAGE005
从第k行点对出发的第s个角度方向为第一行，平行向下i行的点对在第s个方向上的最大频率行点对长度，
Figure 829510DEST_PATH_IMAGE001
表示第k行点对在第s个方向上的最大频率行点对长度，N表示共有N行点对，
Figure 100002_DEST_PATH_IMAGE006
表示所述灰度图像第s个方向上的点对周期长度概率。in,
Figure 100002_DEST_PATH_IMAGE004
Indicates the length of the column point pair with the highest frequency in the direction perpendicular to the sth angle starting from the kth row,
Figure DEST_PATH_IMAGE005
The sth angular direction starting from the point pair of the kth row is the first row, and the point pair of the i row parallel to the downward direction is the maximum frequency row point pair length in the sth direction,
Figure 829510DEST_PATH_IMAGE001
Indicates the maximum frequency row point pair length of the k-th row point pair in the s-th direction, N means that there are N row point pairs in total,
Figure 100002_DEST_PATH_IMAGE006
Indicates the point-to-period length probability in the s-th direction of the grayscale image.

进一步的，一种纺织品表面缺陷的检测方法，根据所述周期长度以及周期延申方向设置滑窗参数的方法如下：Further, a method for detecting surface defects of textiles, the method of setting sliding window parameters according to the cycle length and cycle extension direction is as follows:

获取周期长度概率最大的角度方向

Figure DEST_PATH_IMAGE007
作为周期的延申方向，获取周期长度概率最大的行
Figure 100002_DEST_PATH_IMAGE008
作为周期始点，选取从
Figure 792656DEST_PATH_IMAGE008
行出发的第
Figure 775655DEST_PATH_IMAGE007
个角度方向直线的最大频率点对长度
Figure DEST_PATH_IMAGE009
为周期行长度，将垂直于从第
Figure 829586DEST_PATH_IMAGE008
行出发的第
Figure 971855DEST_PATH_IMAGE007
个角度方向的角度方向直线上的最大频率点对长度
Figure 100002_DEST_PATH_IMAGE010
作为周期列长度；Get the angular direction with the greatest probability of cycle length
Figure DEST_PATH_IMAGE007
As the extension direction of the cycle, get the row with the highest probability of the cycle length
Figure 100002_DEST_PATH_IMAGE008
As the starting point of the cycle, choose from
Figure 792656DEST_PATH_IMAGE008
line departure
Figure 775655DEST_PATH_IMAGE007
The maximum frequency point-pair length of a straight line in the direction of an angle
Figure DEST_PATH_IMAGE009
is the periodic row length, which will be perpendicular to the
Figure 829586DEST_PATH_IMAGE008
line departure
Figure 971855DEST_PATH_IMAGE007
The length of the maximum frequency point pair on the straight line in the angle direction of the angle direction
Figure 100002_DEST_PATH_IMAGE010
as period column length;

根据周期始点、周期长度和周期延申方向获取布匹中滑窗的初始位置

Figure DEST_PATH_IMAGE011
,滑窗尺寸为
Figure 100002_DEST_PATH_IMAGE012
,滑窗的滑动方向为
Figure DEST_PATH_IMAGE013
方向，滑窗的滑动步长为
Figure 100002_DEST_PATH_IMAGE014
。Obtain the initial position of the sliding window in the cloth according to the cycle start point, cycle length and cycle extension direction
Figure DEST_PATH_IMAGE011
, the sliding window size is
Figure 100002_DEST_PATH_IMAGE012
, the sliding direction of the sliding window is
Figure DEST_PATH_IMAGE013
direction, the sliding step of the sliding window is
Figure 100002_DEST_PATH_IMAGE014
.

进一步的，一种纺织品表面缺陷的检测方法，获取对应两个滑窗图像的交集频率值的方法为：Further, a method for detecting surface defects of textiles, the method for obtaining the intersection frequency value corresponding to two sliding window images is as follows:

对各个滑窗图像进行傅里叶变化得到对应滑窗图像的频域空间图像，将任意两两滑窗的频域空间进行交集处理，得到对应两个滑窗图像的交集频率值。The Fourier transformation is performed on each sliding window image to obtain the frequency domain space image corresponding to the sliding window image, and the frequency domain space of any pair of sliding windows is intersected to obtain the intersection frequency value corresponding to the two sliding window images.

进一步的，一种纺织品表面缺陷的检测方法，每一个滑窗图像对应的所有滤波图像中每个像素点的对比度值包括：Further, a method for detecting surface defects of textiles, the contrast value of each pixel in all filtered images corresponding to each sliding window image includes:

通过8邻域像素计算每个滑窗中各个像素点的对比度值，获取所有对比度值大于第一阈值

Figure DEST_PATH_IMAGE015
的像素点，将所述对比度大于第一阈值的像素点作为高对比度像素点。Calculate the contrast value of each pixel in each sliding window through 8 neighboring pixels, and obtain all contrast values greater than the first threshold
Figure DEST_PATH_IMAGE015
Pixels whose contrast is greater than the first threshold are regarded as high-contrast pixels.

进一步的，一种纺织品表面缺陷的检测方法，根据每个滑窗中高对比度像素点的个数计算对应每个滑窗的缺陷概率，表达式为：Further, a method for detecting surface defects of textiles, calculates the defect probability corresponding to each sliding window according to the number of high-contrast pixels in each sliding window, and the expression is:

Figure 100002_DEST_PATH_IMAGE016
Figure 100002_DEST_PATH_IMAGE016

其中，

Figure DEST_PATH_IMAGE017
表示利用第i个滑窗和第j个滑窗的频率交集对第i个滑窗的滤波后的滑窗图像内的第e个高对比度像素8邻域内高对比度像素个数，
Figure 100002_DEST_PATH_IMAGE018
表示该滑窗内高对比度像素的总个数，
Figure DEST_PATH_IMAGE019
表示利用第i个滑窗和第j个滑窗的频率交集对第i个滑窗的滤波后的滑窗图像的缺陷概率。in,
Figure DEST_PATH_IMAGE017
Represents the number of high-contrast pixels in the e-th high-contrast pixel 8 neighborhood of the e-th high-contrast pixel in the filtered sliding window image of the i-th sliding window using the frequency intersection of the i-th sliding window and the j-th sliding window,
Figure 100002_DEST_PATH_IMAGE018
Indicates the total number of high-contrast pixels in the sliding window,
Figure DEST_PATH_IMAGE019
Denotes the defect probability of the filtered sliding window image of the i sliding window using the frequency intersection of the i sliding window and the j sliding window.

进一步的，一种纺织品表面缺陷的检测方法，计算对应每个滑窗的缺陷概率之后，还包括：Further, a method for detecting surface defects of textiles, after calculating the defect probability corresponding to each sliding window, further includes:

根据第i个滑窗和每个滑窗的频率交集对第i个滑窗的滤波后的滑窗缺陷概率计算第i个滑窗的综合缺陷概率，表达式为：Calculate the comprehensive defect probability of the i-th sliding window according to the frequency intersection of the i-th sliding window and each sliding window to the filtered sliding window defect probability of the i-th sliding window, the expression is:

Figure 100002_DEST_PATH_IMAGE020
Figure 100002_DEST_PATH_IMAGE020

其中，

Figure DEST_PATH_IMAGE021
表示利用第i个与第j个滑窗交集对第i个滑窗滤波后的滑窗图像的缺陷概率，
Figure 100002_DEST_PATH_IMAGE022
表示利用第i个与第k个滑窗的频率交集对第k个滑窗的滤波得到滤波后滑窗图像的缺陷概率，Q表示滑窗的个数，
Figure DEST_PATH_IMAGE023
表示第i个滑窗的综合缺陷概率。in,
Figure DEST_PATH_IMAGE021
Indicates the defect probability of the sliding window image filtered by the i-th sliding window using the intersection of the i-th sliding window and the j-th sliding window,
Figure 100002_DEST_PATH_IMAGE022
Indicates that the frequency intersection of the i-th and k-th sliding windows is used to filter the k-th sliding window to obtain the defect probability of the filtered sliding window image, and Q represents the number of sliding windows,
Figure DEST_PATH_IMAGE023
Indicates the comprehensive defect probability of the i-th sliding window.

一种纺织品表面缺陷的检测系统，包括图像预处理单元、第一计算单元、第二计算单元、第三计算单元、第四计算单元以及缺陷检测单元；A detection system for surface defects of textiles, comprising an image preprocessing unit, a first calculation unit, a second calculation unit, a third calculation unit, a fourth calculation unit and a defect detection unit;

图像预处理单元，用于获取纺织品表面灰度图像，并对所述灰度图像进行预处理获得梯度图像；An image preprocessing unit, configured to obtain a grayscale image of the textile surface, and preprocess the grayscale image to obtain a gradient image;

第一计算单元，用于获取梯度图像中每一个像素点在每个方向上两两像素点形成的点对，利用各个方向上最大频率点对所对应的点对长度计算所述灰度图像每个方向的点对周期长度概率；The first calculation unit is used to obtain a point pair formed by two or two pixels in each direction of each pixel in the gradient image, and use the length of the point pair corresponding to the maximum frequency point pair in each direction to calculate each grayscale image. point-to-period length probability in each direction;

第二计算单元，用于将所有点对得到的周期长度概率最大值所对应的方向作为周期的延申方向，获取周期延申方向上的周期长度，根据所述周期长度以及周期延申方向设置滑窗参数；The second calculation unit is used to use the direction corresponding to the maximum value of the cycle length probability obtained by all point pairs as the extension direction of the cycle, obtain the cycle length in the cycle extension direction, and set according to the cycle length and the cycle extension direction Sliding window parameters;

第三计算单元，用于利用设定参数的窗口对所述灰度图像进行滑窗，获取各个滑窗图像的频域空间图像，根据两两滑窗的频域空间交集得到对应两个滑窗图像的交集频率值；The third calculation unit is used to perform sliding windowing on the grayscale image by using the window with set parameters to obtain the frequency domain space image of each sliding window image, and obtain two corresponding sliding windows according to the frequency domain space intersection of any pair of sliding windows The intersection frequency value of the image;

第四计算单元，用于利用每一个滑窗图像与其它滑窗图像的交集频率值分别对该滑窗图像进行滤波处理获得该滑窗图像在不同交集频率值滤波处理后的滤波图像；The fourth calculation unit is used to use the intersection frequency values of each sliding window image and other sliding window images to filter the sliding window image respectively to obtain the filtered image of the sliding window image after filtering at different intersection frequency values;

缺陷检测单元，用于利用每一个滑窗图像对应的所有滤波图像中每个像素点的对比度值，计算该滑窗图像的缺陷概率，根据所述缺陷概率判断该滑窗图像是否存在缺陷。The defect detection unit is configured to use the contrast value of each pixel in all the filtered images corresponding to each sliding window image to calculate the defect probability of the sliding window image, and judge whether the sliding window image has defects according to the defect probability.

本发明的有益效果是：根据本发明提出的技术手段，通过分析图像中的点对纹理信息得到纺织品的纹理周期信息，利用周期信息设置滑窗参数，从而进行进一步的滤波处理，能够克服纺织品自身印花纹理的干扰，精确的定位出缺陷区域，从而有效提升了检测效率和生产质量。The beneficial effects of the present invention are: according to the technical means proposed by the present invention, the texture cycle information of the textile is obtained by analyzing the point-to-texture information in the image, and the sliding window parameters are set by using the cycle information, thereby performing further filtering processing, which can overcome the textile itself. The interference of the printed texture can accurately locate the defect area, thus effectively improving the detection efficiency and production quality.

附图说明Description of drawings

为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动性的前提下，还可以根据这些附图获得其他的附图。In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those skilled in the art, other drawings can also be obtained according to these drawings without any creative effort.

图1为本发明实施例的一种纺织品表面缺陷检测方法结构示意图；Fig. 1 is a schematic structural diagram of a textile surface defect detection method according to an embodiment of the present invention;

图2为本发明实施例的另一种纺织品表面缺陷检测方法结构示意图；Fig. 2 is a schematic structural diagram of another textile surface defect detection method according to an embodiment of the present invention;

图3为本发明实施例的一种纺织品表面缺陷检测系统流程示意图。Fig. 3 is a schematic flowchart of a textile surface defect detection system according to an embodiment of the present invention.

具体实施方式Detailed ways

下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

实施例1Example 1

如图1所示，给出了本发明实施例的一种纺织品表面缺陷检测方法及系统结构示意图，包括：As shown in Figure 1, a method for detecting surface defects of textiles and a schematic structural diagram of the system according to an embodiment of the present invention are provided, including:

101.获取纺织品表面灰度图像，并对所述灰度图像进行预处理。101. Acquire a grayscale image of a textile surface, and perform preprocessing on the grayscale image.

本实施例所针对的情景为：在纹理周期性相同（注：颜色周期可能不同）的布匹生产线上方设置相机，当布匹运行至该相机下侧时，采集布匹图片，通过处理布匹图片来实现布匹的经纬线的断裂现象。The scenario for this embodiment is: set up a camera above the cloth production line with the same texture periodicity (note: the color cycle may be different), when the cloth runs to the lower side of the camera, collect the cloth picture, and realize the cloth picture by processing the cloth picture. The breaking phenomenon of latitude and longitude.

将采集的图像从RGB颜色空间转化为灰度化图像，对灰度图像预处理的操作包括：Convert the collected image from the RGB color space to a grayscale image, and the preprocessing operations on the grayscale image include:

为了获取布匹印花纹理的周期性信息，防止布匹经纬线纹理的干扰，需要利用低通滤波器把布匹经纬线纹理去除，利用sober算子处理滤波图像得到梯度图像。In order to obtain the periodic information of the cloth printing texture and prevent the interference of the cloth latitude and longitude texture, it is necessary to use a low-pass filter to remove the cloth latitude and longitude texture, and use the sober operator to process the filtered image to obtain a gradient image.

102.获取所述灰度图像中每个方向的点对，根据各个方向上最大频率点对长度计算所述灰度图像每个方向的点对周期长度概率。102. Acquire point pairs in each direction in the grayscale image, and calculate a point pair period length probability in each direction of the grayscale image according to the maximum frequency point pair length in each direction.

从第k行第一个非零梯度坐标点出发，在第

Figure 100002_DEST_PATH_IMAGE024
个角度方向上，将同梯度值的像素组成一个点对，将该点对记作
Figure DEST_PATH_IMAGE025
,
Figure 100002_DEST_PATH_IMAGE026
分别表示该点对的起始位置和终止位置。Starting from the first non-zero gradient coordinate point in row k, at
Figure 100002_DEST_PATH_IMAGE024
In the angle direction, the pixels with the same gradient value form a point pair, and the point pair is denoted as
Figure DEST_PATH_IMAGE025
,
Figure 100002_DEST_PATH_IMAGE026
represent the starting and ending positions of the point pair, respectively.

需要说明的是，一个像素可以具有多个点对。例如在作为

Figure DEST_PATH_IMAGE027
处梯度角度值为23度，在该像素
Figure 463491DEST_PATH_IMAGE024
方向上还有6个像素具有23梯度角度，因而从坐标
Figure 58421DEST_PATH_IMAGE027
处出发的点对个数为7。It should be noted that one pixel may have multiple point pairs. For example in as
Figure DEST_PATH_IMAGE027
The gradient angle value at is 23 degrees, at this pixel
Figure 463491DEST_PATH_IMAGE024
There are 6 more pixels in the direction with a gradient angle of 23, so from the coordinate
Figure 58421DEST_PATH_IMAGE027
The number of point pairs starting from is 7.

统计第k行的第

Figure 100002_DEST_PATH_IMAGE028
角度方向的点对长度：统计第k行的第s角度方向的各点对长度的点对频率，获取最大频率对应的行点对长度
Figure DEST_PATH_IMAGE029
。类比该方式得到第s角度方向上其余行的最大频率点对长度。Count the k-th row's
Figure 100002_DEST_PATH_IMAGE028
Point pair length in the angular direction: count the point pair frequency of each point pair length in the sth angular direction of the kth row, and obtain the row point pair length corresponding to the maximum frequency
Figure DEST_PATH_IMAGE029
. By analogy to this method, the maximum frequency point pair lengths of the remaining rows in the s-th angle direction are obtained.

获取垂直于从第k行出发的第s个角度的角度方向，类比第k行的第s个角度方向点对长度求解方法得到该方向的频率最高的列点对长度

Figure 585217DEST_PATH_IMAGE002
。Obtain the angle direction perpendicular to the s-th angle starting from the k-th row, analogous to the point-pair length solution method for the s-th angle direction of the k-th row to obtain the length of the column point pair with the highest frequency in this direction
Figure 585217DEST_PATH_IMAGE002
.

103.将点对周期长度概率最大值对应的方向作为周期的延申方向，获取周期延伸方向上的周期长度，根据所述周期长度以及周期延申方向设置滑窗参数。103. Use the direction corresponding to the maximum probability of the point-to-period length as the extension direction of the cycle, obtain the cycle length in the cycle extension direction, and set the sliding window parameters according to the cycle length and the cycle extension direction.

选取周期长度概率最大的角度方向

Figure 100002_DEST_PATH_IMAGE030
作为周期的延申方向，选取该方向上周期长度概率最大的行
Figure DEST_PATH_IMAGE031
作为周期始点，选取从
Figure 972205DEST_PATH_IMAGE031
行出发的第
Figure 290535DEST_PATH_IMAGE030
个角度方向直线的点对长度
Figure 100002_DEST_PATH_IMAGE032
为周期行长度，选取垂直于从第
Figure 576023DEST_PATH_IMAGE031
行出发的第
Figure 640931DEST_PATH_IMAGE030
个角度方向的角度方向直线上的点对长度
Figure DEST_PATH_IMAGE033
作为列点对长度。Select the angular direction with the greatest probability of period length
Figure 100002_DEST_PATH_IMAGE030
As the extension direction of the cycle, select the row with the highest cycle length probability in this direction
Figure DEST_PATH_IMAGE031
As the starting point of the cycle, choose from
Figure 972205DEST_PATH_IMAGE031
line departure
Figure 290535DEST_PATH_IMAGE030
The point-to-point length of a straight line in the direction of an angle
Figure 100002_DEST_PATH_IMAGE032
For the periodic row length, choose perpendicular to the
Figure 576023DEST_PATH_IMAGE031
line departure
Figure 640931DEST_PATH_IMAGE030
The point-pair length on the straight line in the angle direction of the angle direction
Figure DEST_PATH_IMAGE033
as column-point-pair lengths.

由于布匹的纹理呈现周期性变化，因而需要根据周期参数设置滑窗参数。Since the texture of the cloth changes periodically, it is necessary to set the sliding window parameters according to the period parameters.

104.对所述灰度图像进行滑窗，获取各个滑窗图像的频域空间图像，根据两两滑窗的频域空间交集得到对应两个滑窗图像的交集频率值。104. Perform a sliding window on the grayscale image, acquire frequency-domain space images of each sliding-window image, and obtain an intersection frequency value corresponding to two sliding-window images according to the frequency-domain space intersection of any pair of sliding windows.

将各滑窗图像进行傅里叶变化得到频域空间图像，将任意两两滑窗的频域空间进行交集集处理得到两图像的交集频率值集合，利用交集频率集合对对应两个滑窗图像分别进行滤波处理得到滤波后图像。Perform Fourier transformation of each sliding window image to obtain a frequency domain space image, and perform intersection set processing on the frequency domain space of any pair of sliding windows to obtain the intersection frequency value set of the two images, and use the intersection frequency set to correspond to the two sliding window images Filtering is performed respectively to obtain a filtered image.

通过该方式即可得到基于滑窗的其余滑窗滤波图像。In this way, other sliding window filtered images based on the sliding window can be obtained.

105.根据所述交集频率值对每个滑窗图像进行滤波处理，计算滤波处理后的各个滑窗内每个像素点的对比度值，获取每个滑窗中对比度值大于第一阈值的像素点作为高对比度像素点。105. Perform filtering processing on each sliding window image according to the intersection frequency value, calculate the contrast value of each pixel in each sliding window after filtering, and obtain the pixels whose contrast value is greater than the first threshold in each sliding window as high-contrast pixels.

通过8邻域像素计算出各像素的对比度值，通过对比度值分割出可能缺陷区域，即对比度大于设定阈值

Figure 100002_DEST_PATH_IMAGE034
的像素集合。Calculate the contrast value of each pixel through 8 neighboring pixels, and segment possible defect areas through the contrast value, that is, the contrast is greater than the set threshold
Figure 100002_DEST_PATH_IMAGE034
collection of pixels.

106.根据每个滑窗中高对比度像素点的个数计算对应每个滑窗的缺陷概率，根据所述缺陷概率判断对应滑窗图像是否存在缺陷。106. Calculate the defect probability corresponding to each sliding window according to the number of high-contrast pixels in each sliding window, and judge whether there is a defect in the corresponding sliding window image according to the defect probability.

由于每个滑窗内包含完整的单周期印花图案，因而有过不存在缺陷滤波后的图像较为平滑。在存在缺陷时，缺陷区域存在一些纹理信息，其他区域的纹理信息较少，较为平滑，因而基于该特征来评估各滑窗内缺陷概率。Since each sliding window contains a complete single-cycle printing pattern, the image after filtering without defects is relatively smooth. When there is a defect, there is some texture information in the defect area, and the texture information in other areas is less and smoother, so the defect probability in each sliding window is estimated based on this feature.

为了防止单个窗口评估印刷缺陷的精度低的问题，因而需再结合滑窗集合的滤波效果来综合评估各滑窗的综合缺陷概率。In order to prevent the problem of low accuracy of evaluating printing defects by a single window, it is necessary to combine the filtering effect of the sliding window set to comprehensively evaluate the comprehensive defect probability of each sliding window.

通过缺陷概率筛选出可能存在缺陷滑窗，当滑窗的缺陷概率

Figure DEST_PATH_IMAGE035
时认为该滑窗存在缺陷。Sliding windows that may have defects are screened out through the defect probability, when the defect probability of the sliding window
Figure DEST_PATH_IMAGE035
The sliding window is considered to be defective.

根据本发明提出的技术手段，通过分析图像中的点对纹理信息得到纺织品的纹理周期信息，利用周期信息设置滑窗参数，从而进行进一步的滤波处理，能够克服纺织品自身印花纹理的干扰，精确的定位出缺陷区域，从而有效提升了检测效率和生产质量。According to the technical means proposed by the present invention, the texture cycle information of the textile is obtained by analyzing the point-to-texture information in the image, and the sliding window parameters are set by using the cycle information, so as to perform further filtering processing, which can overcome the interference of the textile's own printing texture, and accurately The defect area is located, thereby effectively improving the detection efficiency and production quality.

实施例2Example 2

如图2所示，给出了本发明实施例另一种纺织品表面缺陷检测方法及系统，包括：As shown in Figure 2, another textile surface defect detection method and system according to the embodiment of the present invention is provided, including:

201.获取纺织品表面灰度图像，并对所述灰度图像进行预处理。201. Acquire a grayscale image of a textile surface, and perform preprocessing on the grayscale image.

本实施例需要根据采集的布匹图像来实现布匹的缺陷检测，所以需要先采集布匹图像，分割出布匹区域。In this embodiment, cloth defect detection needs to be realized based on the collected cloth images, so it is necessary to first collect the cloth images and segment the cloth areas.

在纺织品生产线正上方设置相机，相机每间隔一段时间拍摄一张图片，相机的间隔时间可以根据相机视角宽度和生产线运行速度来设置。Set up the camera directly above the textile production line, and the camera takes a picture at intervals. The interval of the camera can be set according to the width of the camera's viewing angle and the running speed of the production line.

将采集的图像从RGB颜色空间转化为灰度化图像。Convert the acquired image from RGB color space to grayscale image.

纺织品的花式各式各样，为了让系统能够使用与各种情况，增强其泛化能力，所以本发明采用DNN语义分割的方式来识别分割图像中的布匹区域。There are various styles of textiles. In order to allow the system to be used in various situations and enhance its generalization ability, the present invention uses DNN semantic segmentation to identify the cloth area in the segmented image.

该DNN网络的相关内容如下：The relevant content of the DNN network is as follows:

使用的数据集为俯视采集的纺织品图像数据集。The dataset used is a dataset of textile images collected from above.

需要分割的像素，共分为两类，即训练集对应标签标注过程为：单通道的语义标签，对应位置像素属于背景类的标注为0，属于布匹的标注为1。The pixels that need to be segmented are divided into two categories, that is, the corresponding label labeling process of the training set is: single-channel semantic label, the corresponding position pixel belongs to the background class is marked as 0, and the label belonging to the cloth is marked as 1.

网络的任务是分类，所以使用的loss函数为交叉熵损失函数。The task of the network is classification, so the loss function used is the cross-entropy loss function.

为了获取布匹印花纹理的周期性信息，防止布匹经纬线纹理的干扰，需要利用低通滤波器把布匹经纬线纹理去除。In order to obtain the periodic information of the cloth printing texture and prevent the interference of the cloth latitude and longitude texture, it is necessary to use a low-pass filter to remove the cloth latitude and longitude texture.

利用尺度为

Figure 100002_DEST_PATH_IMAGE036
的高斯滤波器对布匹灰度图像
Figure DEST_PATH_IMAGE037
进行滤波处理，得到处理后图片
Figure 100002_DEST_PATH_IMAGE038
。The scale of use is
Figure 100002_DEST_PATH_IMAGE036
The Gaussian filter on the cloth grayscale image
Figure DEST_PATH_IMAGE037
Perform filtering processing to obtain the processed image
Figure 100002_DEST_PATH_IMAGE038
.

利用sober算子处理图像

Figure DEST_PATH_IMAGE039
得到梯度图像
Figure 100002_DEST_PATH_IMAGE040
.Image processing using sober operator
Figure DEST_PATH_IMAGE039
get the gradient image
Figure 100002_DEST_PATH_IMAGE040
.

202.获取所述灰度图像中每个方向的点对，根据各个方向上最大频率点对长度计算所述灰度图像每个方向的点对周期长度概率。202. Acquire point pairs in each direction in the grayscale image, and calculate a point pair period length probability in each direction of the grayscale image according to the maximum frequency point pair length in each direction.

以水平向右为0度角度方向集合，以1为角度间隔，得到360个角度方向集合。Take the horizontal right as the set of 0-degree angle directions, and take 1 as the angle interval to obtain 360 sets of angle directions.

从第k行第一个非零梯度坐标点出发，在第

Figure 665781DEST_PATH_IMAGE024
个角度方向上，将同梯度值的像素组成一个点对，将该点对记作
Figure 157942DEST_PATH_IMAGE025
,
Figure 727464DEST_PATH_IMAGE026
分别表示该点对的起始位置和终止位置。注：一个像素可以具有多个点对。例如在作为
Figure 205850DEST_PATH_IMAGE027
处梯度角度值为23度，在该像素
Figure 911637DEST_PATH_IMAGE024
方向上还有6个像素具有23梯度角度，因而从坐标
Figure 574700DEST_PATH_IMAGE027
处出发的点对个数为7。通过该方式得到点对集合
Figure DEST_PATH_IMAGE041
。Starting from the first non-zero gradient coordinate point in row k, at
Figure 665781DEST_PATH_IMAGE024
In the angle direction, the pixels with the same gradient value form a point pair, and the point pair is denoted as
Figure 157942DEST_PATH_IMAGE025
,
Figure 727464DEST_PATH_IMAGE026
represent the starting and ending positions of the point pair, respectively. NOTE: A pixel can have multiple point pairs. For example in as
Figure 205850DEST_PATH_IMAGE027
The gradient angle value at is 23 degrees, at this pixel
Figure 911637DEST_PATH_IMAGE024
There are 6 more pixels in the direction with a gradient angle of 23, so from the coordinate
Figure 574700DEST_PATH_IMAGE027
The number of point pairs starting from is 7. In this way, the set of point pairs is obtained
Figure DEST_PATH_IMAGE041
.

类比该方式得到其余行的第s角度方向上的点对。By analogy to this method, the point pairs in the s-th angle direction of the remaining rows are obtained.

统计第k行的第

Figure 448763DEST_PATH_IMAGE028
角度方向的点对长度：统计第k行的第s角度方向的各点对长度的点对频率，获取最大频率对应的行点对长度
Figure 855474DEST_PATH_IMAGE029
。类比该方式得到第s角度方向上其余行的最大频率点对长度。Count the k-th row's
Figure 448763DEST_PATH_IMAGE028
Point pair length in the angular direction: count the point pair frequency of each point pair length in the sth angular direction of the kth row, and obtain the row point pair length corresponding to the maximum frequency
Figure 855474DEST_PATH_IMAGE029
. By analogy to this method, the maximum frequency point pair lengths of the remaining rows in the s-th angle direction are obtained.

计算垂直从k行出发的第s个角度的角度方向得到列点对长度：获取垂直于从第k行出发的第s个角度的角度方向，类比第k行的第s个角度方向点对长度求解方法得到该方向的频率最高的列点对长度

Figure 415768DEST_PATH_IMAGE002
。Calculate the angle direction perpendicular to the sth angle starting from row k to obtain the column point pair length: Obtain the angle direction perpendicular to the sth angle starting from row k, analogous to the point pair length of the sth angle direction of row k The solution method obtains the length of the column point pair with the highest frequency in this direction
Figure 415768DEST_PATH_IMAGE002
.

计算所述灰度图像每个方向的点对周期长度概率的方法如下：The method for calculating the point-to-period length probability in each direction of the grayscale image is as follows:

利用起始点与终止点之间的欧氏距离计算所述灰度图像中第k行点对在每个方向上的点对长度，获取第s个方向上最大频率对应的行点对长度

Figure 515311DEST_PATH_IMAGE001
；Use the Euclidean distance between the start point and the end point to calculate the point pair length of the k-th row point pair in each direction in the grayscale image, and obtain the row point pair length corresponding to the maximum frequency in the s-th direction
Figure 515311DEST_PATH_IMAGE001
;

同理，计算从第k行出发与第s个方向垂直方向上的所有点对长度得到列点对长度，获取每个方向上频率最大的列点对长度

Figure 528266DEST_PATH_IMAGE002
；Similarly, calculate the length of all point pairs starting from the k-th row and perpendicular to the s-th direction to obtain the length of the column point pair, and obtain the length of the column point pair with the highest frequency in each direction
Figure 528266DEST_PATH_IMAGE002
;

计算第s个方向的周期长度概率的表达式为：The expression for calculating the cycle length probability in the sth direction is:

Figure 473089DEST_PATH_IMAGE003
Figure 473089DEST_PATH_IMAGE003

其中，

Figure 28835DEST_PATH_IMAGE004
表示从第k行出发垂直于第s个角度方向上频率最大的列点对长度，
Figure 33700DEST_PATH_IMAGE005
从第k行点对出发的第s个角度方向为第一行，平行向下i行的点对在第s个方向上的最大频率行点对长度，
Figure 799531DEST_PATH_IMAGE001
表示第k行点对在第s个方向上的最大频率行点对长度，N表示共有N行点对，
Figure 285394DEST_PATH_IMAGE006
表示所述灰度图像第s个方向上的点对周期长度概率。in,
Figure 28835DEST_PATH_IMAGE004
Indicates the length of the column point pair with the highest frequency in the direction perpendicular to the sth angle starting from the kth row,
Figure 33700DEST_PATH_IMAGE005
The sth angular direction starting from the point pair of the kth row is the first row, and the point pair of the i row parallel to the downward direction is the maximum frequency row point pair length in the sth direction,
Figure 799531DEST_PATH_IMAGE001
Indicates the maximum frequency row point pair length of the k-th row point pair in the s-th direction, N means that there are N row point pairs in total,
Figure 285394DEST_PATH_IMAGE006
Indicates the point-to-period length probability in the s-th direction of the grayscale image.

203.将点对周期长度概率最大值对应的方向作为周期的延申方向，获取周期延伸方向上的周期长度，根据所述周期长度以及周期延申方向设置滑窗参数。203. Use the direction corresponding to the maximum value of the point-to-cycle length probability as the extension direction of the cycle, obtain the cycle length in the cycle extension direction, and set the sliding window parameters according to the cycle length and the cycle extension direction.

根据所述周期长度以及周期延申方向设置滑窗参数的方法如下：The method of setting the sliding window parameters according to the cycle length and the cycle extension direction is as follows:

获取周期长度概率最大的角度方向

Figure 85860DEST_PATH_IMAGE007
作为周期的延申方向，获取周期长度概率最大的行
Figure 261627DEST_PATH_IMAGE008
作为周期始点，选取从
Figure 249174DEST_PATH_IMAGE008
行出发的第
Figure 535799DEST_PATH_IMAGE007
个角度方向直线的最大频率点对长度
Figure 925192DEST_PATH_IMAGE009
为周期行长度，将垂直于从第
Figure 271860DEST_PATH_IMAGE008
行出发的第
Figure 887649DEST_PATH_IMAGE007
个角度方向的角度方向直线上的最大频率点对长度
Figure 243544DEST_PATH_IMAGE010
作为周期列长度；Get the angular direction with the greatest probability of cycle length
Figure 85860DEST_PATH_IMAGE007
As the extension direction of the cycle, get the row with the highest probability of the cycle length
Figure 261627DEST_PATH_IMAGE008
As the starting point of the cycle, choose from
Figure 249174DEST_PATH_IMAGE008
line departure
Figure 535799DEST_PATH_IMAGE007
The maximum frequency point-pair length of a straight line in an angular direction
Figure 925192DEST_PATH_IMAGE009
is the periodic row length, which will be perpendicular to the
Figure 271860DEST_PATH_IMAGE008
line departure
Figure 887649DEST_PATH_IMAGE007
The length of the maximum frequency point pair on the straight line in the angle direction of the angle direction
Figure 243544DEST_PATH_IMAGE010
as period column length;

根据周期始点、周期长度和周期延申方向获取布匹中滑窗的初始位置

Figure 484514DEST_PATH_IMAGE011
,滑窗尺寸为
Figure 2083DEST_PATH_IMAGE012
,滑窗的滑动方向为
Figure 964222DEST_PATH_IMAGE013
方向，滑窗的滑动步长为
Figure 592650DEST_PATH_IMAGE014
。Obtain the initial position of the sliding window in the cloth according to the cycle start point, cycle length and cycle extension direction
Figure 484514DEST_PATH_IMAGE011
, the sliding window size is
Figure 2083DEST_PATH_IMAGE012
, the sliding direction of the sliding window is
Figure 964222DEST_PATH_IMAGE013
direction, the sliding step of the sliding window is
Figure 592650DEST_PATH_IMAGE014
.

2041.对所述灰度图像进行滑窗，获取各个滑窗图像的频域空间图像，根据两两滑窗的频域空间交集得到对应两个滑窗图像的交集频率值。2041. Perform a sliding window on the grayscale image, acquire frequency domain spatial images of each sliding window image, and obtain an intersection frequency value corresponding to two sliding window images according to the frequency domain space intersection of any pair of sliding windows.

获取对应两个滑窗图像的交集频率值的方法为：The method to obtain the intersection frequency value corresponding to two sliding window images is:

对各个滑窗图像进行傅里叶变化得到对应滑窗图像的频域空间图像，将任意两两滑窗的频域空间进行交集处理，得到对应两个滑窗图像的交集频率值。The Fourier transformation is performed on each sliding window image to obtain the frequency domain space image corresponding to the sliding window image, and the frequency domain space of any pair of sliding windows is intersected to obtain the intersection frequency value corresponding to the two sliding window images.

2042.根据所述交集频率值对每个滑窗图像进行滤波处理，计算滤波处理后的各个滑窗内每个像素点的对比度值，获取每个滑窗中对比度值大于第一阈值的像素点作为高对比度像素点。2042. Perform filtering processing on each sliding window image according to the intersection frequency value, calculate the contrast value of each pixel in each sliding window after filtering, and obtain the pixel points in each sliding window whose contrast value is greater than the first threshold as high-contrast pixels.

获取每个滑窗中对比度值大于第一阈值的像素点作为高对比度像素点的方法为：The method of obtaining pixels with a contrast value greater than the first threshold in each sliding window as high-contrast pixels is:

通过8邻域像素计算每个滑窗中各个像素点的对比度值，获取所有对比度值大于第一阈值

Figure 222214DEST_PATH_IMAGE015
的像素点，将所述对比度大于第一阈值的像素点作为高对比度像素点。Calculate the contrast value of each pixel in each sliding window through 8 neighboring pixels, and obtain all contrast values greater than the first threshold
Figure 222214DEST_PATH_IMAGE015
Pixels whose contrast is greater than the first threshold are regarded as high-contrast pixels.

2043.根据每个滑窗中高对比度像素点的个数计算对应每个滑窗的缺陷概率，根据所述缺陷概率判断对应滑窗图像是否存在缺陷。2043. Calculate the defect probability corresponding to each sliding window according to the number of high-contrast pixels in each sliding window, and judge whether there is a defect in the corresponding sliding window image according to the defect probability.

由于每个滑窗内包含完整的单周期印花图案，因而有过不存在缺陷滤波后的图像较为平滑。在存在缺陷时，缺陷区域存在一些纹理信息，其他区域的纹理信息较少，较为平滑，因而基于该特征来评估各滑窗内缺陷概率。Since each sliding window contains a complete single-cycle printing pattern, the image after filtering without defects is relatively smooth. When there is a defect, there is some texture information in the defect area, and the texture information in other areas is less and smoother, so the defect probability in each sliding window is estimated based on this feature.

根据每个滑窗中高对比度像素点的个数计算对应每个滑窗的缺陷概率，表达式为：Calculate the defect probability corresponding to each sliding window according to the number of high-contrast pixels in each sliding window, the expression is:

Figure 645105DEST_PATH_IMAGE016
Figure 645105DEST_PATH_IMAGE016

其中，

Figure 235487DEST_PATH_IMAGE017
表示利用第i个滑窗和第j个滑窗的频率交集对第i个滑窗的滤波后的滑窗图像内的第e个高对比度像素8邻域内高对比度像素个数，
Figure 667605DEST_PATH_IMAGE018
表示该滑窗内高对比度像素的总个数，
Figure 620518DEST_PATH_IMAGE019
表示利用第i个滑窗和第j个滑窗的频率交集对第i个滑窗的滤波后的滑窗图像的缺陷概率，
Figure 100002_DEST_PATH_IMAGE042
表示利用第i个与第j个滑窗的频率交集对第i个滑窗滤波得到的滤波后滑窗图像中的缺陷概率值，该值越大说明该滑窗内纹理区域集中分布，因而也说明该滑窗内存在缺陷的概率较大。in,
Figure 235487DEST_PATH_IMAGE017
Represents the number of high-contrast pixels in the e-th high-contrast pixel 8 neighborhood of the e-th high-contrast pixel in the filtered sliding window image of the i-th sliding window using the frequency intersection of the i-th sliding window and the j-th sliding window,
Figure 667605DEST_PATH_IMAGE018
Indicates the total number of high-contrast pixels in the sliding window,
Figure 620518DEST_PATH_IMAGE019
Represents the defect probability of the filtered sliding window image of the i sliding window using the frequency intersection of the i sliding window and the j sliding window,
Figure 100002_DEST_PATH_IMAGE042
Indicates the defect probability value in the filtered sliding window image obtained by filtering the i-th sliding window by using the frequency intersection of the i-th and j-th sliding windows. It shows that there is a high probability of defects in the sliding window.

为了防止单个窗口评估印刷缺陷的精度低的问题，因而需再结合滑窗集合的滤波效果来综合评估各滑窗的综合缺陷概率，计算对应每个滑窗的缺陷概率之后，还包括：In order to prevent the problem of low accuracy of evaluating printing defects by a single window, it is necessary to combine the filtering effect of the sliding window set to comprehensively evaluate the comprehensive defect probability of each sliding window. After calculating the defect probability corresponding to each sliding window, it also includes:

根据第i个滑窗和每个滑窗的频率交集对第i个滑窗的滤波后的滑窗缺陷概率计算第i个滑窗的综合缺陷概率，表达式为：Calculate the comprehensive defect probability of the i-th sliding window according to the frequency intersection of the i-th sliding window and each sliding window to the filtered sliding window defect probability of the i-th sliding window, the expression is:

Figure 810715DEST_PATH_IMAGE020
Figure 810715DEST_PATH_IMAGE020

其中，

Figure 481868DEST_PATH_IMAGE021
表示利用第i个与第j个滑窗交集对第i个滑窗滤波后的滑窗图像的缺陷概率，
Figure 717677DEST_PATH_IMAGE022
表示利用第i个与第k个滑窗的频率交集对第k个滑窗的得到滤波后滑窗图像的缺陷概率，Q表示滑窗的个数，
Figure 56255DEST_PATH_IMAGE023
表示第i个滑窗的综合缺陷概率，当第i个滑窗的存在缺陷时，该滑窗与其他滑窗存在较大的频域差异，因而利用其余滑窗与该滑窗交集对该滑窗进行滤波后就会将缺陷区域暴漏出来，因而滤波后滑窗内缺陷概率较大，
Figure DEST_PATH_IMAGE043
表示滤波后第i个滑窗的缺陷概率均值，该值越大，说明第i个滑窗区域存在缺陷的概率越大，
Figure 148844DEST_PATH_IMAGE022
表示利用第i个与第k个滑窗的频率交集对第k个滑窗的滤波得到滤波后滑窗图像的缺陷概率，该值越大，说明第i个滑窗图像由于缺陷导致部分频率丢失导致该滑窗与其他滑窗的存在一定的频率差异，因而通过频率交集滤波导致部分布匹本身纹理信息没有被滤波去除，因而会被误识为缺陷，其缺陷概率提高。
Figure 41714DEST_PATH_IMAGE023
表示第i个滑窗的缺陷概率。in,
Figure 481868DEST_PATH_IMAGE021
Indicates the defect probability of the sliding window image filtered by the i-th sliding window using the intersection of the i-th sliding window and the j-th sliding window,
Figure 717677DEST_PATH_IMAGE022
Represents the defect probability of the filtered sliding window image obtained by using the frequency intersection of the i-th and k-th sliding windows for the k-th sliding window, Q represents the number of sliding windows,
Figure 56255DEST_PATH_IMAGE023
Indicates the comprehensive defect probability of the i-th sliding window. When there is a defect in the i-th sliding window, there is a large frequency domain difference between this sliding window and other sliding windows, so the intersection of other sliding windows and this sliding window is used to determine the After the window is filtered, the defect area will be exposed, so the probability of defects in the sliding window after filtering is relatively high.
Figure DEST_PATH_IMAGE043
Indicates the average value of the defect probability of the i-th sliding window after filtering, the larger the value, the greater the probability of defects in the i-th sliding window area,
Figure 148844DEST_PATH_IMAGE022
Indicates the defect probability of the filtered sliding window image obtained by filtering the k-th sliding window by the frequency intersection of the i-th and k-th sliding windows. The larger the value, the i-th sliding window image loses part of the frequency due to defects As a result, there is a certain frequency difference between the sliding window and other sliding windows, so the texture information of part of the cloth itself is not filtered and removed through frequency intersection filtering, so it will be misidentified as a defect, and its defect probability will increase.
Figure 41714DEST_PATH_IMAGE023
Indicates the defect probability of the i-th sliding window.

通过缺陷概率筛选出可能存在缺陷滑窗，当滑窗的缺陷概率

Figure 78285DEST_PATH_IMAGE035
时认为该滑窗存在缺陷，根据经验该阈值
Figure 100002_DEST_PATH_IMAGE044
通常取0.7。Sliding windows that may have defects are screened out through the defect probability, when the defect probability of the sliding window
Figure 78285DEST_PATH_IMAGE035
When the sliding window is considered to be defective, according to experience the threshold
Figure 100002_DEST_PATH_IMAGE044
Usually take 0.7.

该滑窗内高对比度像素区域即为缺陷区域。The high-contrast pixel area within the sliding window is the defect area.

如图3所示，公开了本实施例的一种纺织品表面缺陷的检测系统，包括图像预处理单元、第一计算单元、第二计算单元、第三计算单元、第四计算单元以及缺陷检测单元；As shown in Figure 3, a textile surface defect detection system of this embodiment is disclosed, including an image preprocessing unit, a first calculation unit, a second calculation unit, a third calculation unit, a fourth calculation unit and a defect detection unit ;

图像预处理单元，用于获取纺织品表面灰度图像，并对所述灰度图像进行预处理获得梯度图像；An image preprocessing unit, configured to obtain a grayscale image of the textile surface, and preprocess the grayscale image to obtain a gradient image;

第一计算单元，用于获取梯度图像中每一个像素点在每个方向上两两像素点形成的点对，利用各个方向上最大频率点对所对应的点对长度计算所述灰度图像每个方向的点对周期长度概率；The first calculation unit is used to obtain a point pair formed by two or two pixels in each direction of each pixel in the gradient image, and use the length of the point pair corresponding to the maximum frequency point pair in each direction to calculate each grayscale image. point-to-period length probability in each direction;

第二计算单元，用于将所有点对得到的周期长度概率最大值所对应的方向作为周期的延申方向，获取周期延申方向上的周期长度，根据所述周期长度以及周期延申方向设置滑窗参数；The second calculation unit is used to use the direction corresponding to the maximum value of the cycle length probability obtained by all point pairs as the extension direction of the cycle, obtain the cycle length in the cycle extension direction, and set according to the cycle length and the cycle extension direction Sliding window parameters;

第三计算单元，用于利用设定参数的窗口对所述灰度图像进行滑窗，获取各个滑窗图像的频域空间图像，根据两两滑窗的频域空间交集得到对应两个滑窗图像的交集频率值；The third calculation unit is used to perform sliding windowing on the grayscale image by using the window with set parameters to obtain the frequency domain space image of each sliding window image, and obtain two corresponding sliding windows according to the frequency domain space intersection of any pair of sliding windows The intersection frequency value of the image;

第四计算单元，用于利用每一个滑窗图像与其它滑窗图像的交集频率值分别对该滑窗图像进行滤波处理获得该滑窗图像在不同交集频率值滤波处理后的滤波图像；The fourth calculation unit is used to use the intersection frequency values of each sliding window image and other sliding window images to filter the sliding window image respectively to obtain the filtered image of the sliding window image after filtering at different intersection frequency values;

缺陷检测单元，用于利用每一个滑窗图像对应的所有滤波图像中每个像素点的对比度值，计算该滑窗图像的缺陷概率，根据所述缺陷概率判断该滑窗图像是否存在缺陷。The defect detection unit is configured to use the contrast value of each pixel in all the filtered images corresponding to each sliding window image to calculate the defect probability of the sliding window image, and judge whether the sliding window image has defects according to the defect probability.

根据本发明提出的技术手段，通过分析图像中的点对纹理信息得到纺织品的纹理周期信息，利用周期信息设置滑窗参数，从而进行进一步的滤波处理，能够克服纺织品自身印花纹理的干扰，精确的定位出缺陷区域，从而有效提升了检测效率和生产质量。According to the technical means proposed by the present invention, the texture cycle information of the textile is obtained by analyzing the point-to-texture information in the image, and the sliding window parameters are set by using the cycle information, so as to perform further filtering processing, which can overcome the interference of the textile's own printing texture, and accurately The defect area is located, thereby effectively improving the detection efficiency and production quality.

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、改进等，均应包含在本发滤波明的保护范围之内。The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention should be included in the filter of the present invention. within the scope of protection.

Claims (7)

1. A method for detecting textile surface defects, comprising:
acquiring a gray level image of the surface of the textile, and preprocessing the gray level image to obtain a gradient image;
acquiring a point pair formed by every two pixel points in each direction of each pixel point in the gradient image, and calculating the point pair period length probability of each direction of the gray image by using the point pair length corresponding to the maximum frequency point pair in each direction;
the method for calculating the probability of the point-to-period length of each direction of the gray level image comprises the following steps:
calculating the point pair length of the kth line point pair in each direction in the gray scale image by using the Euclidean distance between the starting point and the ending point, and acquiring the line point pair length corresponding to the maximum frequency in the s direction
Figure DEST_PATH_IMAGE002
；
Similarly, the direction perpendicular to the s direction from the k line is calculatedObtaining the length of the column point pair by the lengths of all the upward point pairs, and obtaining the length of the column point pair with the maximum frequency in the corresponding direction
Figure DEST_PATH_IMAGE004
；
The expression for calculating the cycle length probability in the s-th direction is:
Figure DEST_PATH_IMAGE006
wherein,
Figure DEST_PATH_IMAGE008
represents the maximum frequency column point pair length in the direction perpendicular to the s-th angle from the k-th row,
Figure DEST_PATH_IMAGE010
the s angular direction from the point pair of the k row is the first row, the point pair parallel to the lower i row has the maximum frequency row point pair length in the s direction,
Figure 956824DEST_PATH_IMAGE002
represents the maximum frequency row point pair length of the k-th row point pair in the s-th direction, N represents the total N row point pairs,
Figure DEST_PATH_IMAGE012
representing the probability of the point-to-period length in the s direction of the gray level image;
taking the direction corresponding to the maximum value of the period length probability obtained by all the point pairs as the extension direction of the period, obtaining the period length in the period extension direction, and setting a sliding window parameter according to the period length and the period extension direction;
sliding the gray level images by using the windows with set parameters to obtain frequency domain space images of each sliding window image, and obtaining intersection frequency values of two corresponding sliding window images according to the intersection of the frequency domain spaces of every two sliding windows;
respectively filtering the sliding window image by using the intersection frequency value of each sliding window image and other sliding window images to obtain a filtered image of the sliding window image after filtering at different intersection frequency values;
and calculating the defect probability of each sliding window image by using the contrast value of each pixel point in all the filtering images corresponding to each sliding window image, and judging whether the sliding window image has defects or not according to the defect probability.
2. A method of detecting textile surface defects according to claim 1 in which the method of setting the parameters of the sliding window in dependence on the cycle length and cycle extension direction is as follows:
obtaining the angular direction with the largest period length probability
Figure DEST_PATH_IMAGE014
As the extension direction of the period, the row with the largest period length probability is obtained
Figure DEST_PATH_IMAGE016
As the beginning of the cycle, choose from
Figure 79369DEST_PATH_IMAGE016
Go out of
Figure 459535DEST_PATH_IMAGE014
Maximum frequency point pair length of angle direction straight line
Figure DEST_PATH_IMAGE018
For the length of the periodic line, will be perpendicular to the second
Figure 227027DEST_PATH_IMAGE016
Go out of
Figure 9039DEST_PATH_IMAGE014
Maximum frequency point pair length on angle direction straight line of angle direction
Figure DEST_PATH_IMAGE020
As the periodic column length;
acquiring the initial position of a sliding window in the cloth according to the period starting point, the period length and the period extending direction
Figure DEST_PATH_IMAGE022
The sliding window has the size of
Figure DEST_PATH_IMAGE024
The sliding direction of the sliding window is
Figure DEST_PATH_IMAGE026
Direction, sliding step length of the sliding window
Figure DEST_PATH_IMAGE028
。
3. The method for detecting textile surface defects of claim 1, wherein the method for obtaining the intersection frequency value corresponding to the two sliding window images comprises:
fourier transformation is carried out on each sliding window image to obtain a frequency domain space image corresponding to the sliding window image, intersection processing is carried out on the frequency domain space of any two sliding windows, and an intersection frequency value corresponding to the two sliding window images is obtained.
4. The method of claim 1, wherein the contrast value of each pixel point in all the filtered images corresponding to each sliding window image comprises:
calculating the contrast value of each pixel point in each sliding window through 8 neighborhood pixels, and acquiring all the contrast values larger than a first threshold value
Figure DEST_PATH_IMAGE030
And taking the pixel points with the contrast ratio larger than the first threshold value as high-contrast pixel points.
5. The method for detecting the textile surface defects of claim 4, wherein the defect probability corresponding to each sliding window is calculated according to the number of high-contrast pixel points in each sliding window, and the expression is as follows:
Figure DEST_PATH_IMAGE032
wherein,
Figure DEST_PATH_IMAGE034
indicating the number of high contrast pixels in the neighborhood of the e-th high contrast pixel 8 in the filtered sliding window image of the ith sliding window by using the frequency intersection of the ith sliding window and the jth sliding window,
Figure DEST_PATH_IMAGE036
representing the total number of high contrast pixels within the sliding window,
Figure DEST_PATH_IMAGE038
indicating the defect probability of the filtered sliding window image of the ith sliding window by using the frequency intersection of the ith sliding window and the jth sliding window.
6. A method of detecting defects on a textile surface according to claim 5, wherein after calculating the probability of defects for each sliding window, further comprising:
calculating the comprehensive defect probability of the ith sliding window according to the filtered defect probability of the ith sliding window and the frequency intersection of each sliding window, wherein the expression is as follows:
Figure DEST_PATH_IMAGE040
wherein,
Figure DEST_PATH_IMAGE042
indicating the defect probability of the ith sliding window image after filtering the ith sliding window by using the ith and jth intersection of the sliding window,
Figure DEST_PATH_IMAGE044
the defect probability of the filtered sliding window image obtained by filtering the kth sliding window by using the frequency intersection of the ith sliding window and the kth sliding window is shown, Q represents the number of the sliding windows,
Figure DEST_PATH_IMAGE046
representing the integrated defect probability of the ith sliding window.
7. The textile surface defect detection system is characterized by comprising an image preprocessing unit, a first calculating unit, a second calculating unit, a third calculating unit, a fourth calculating unit and a defect detecting unit;
the image preprocessing unit is used for acquiring a gray level image of the surface of the textile and preprocessing the gray level image to obtain a gradient image;
the first calculation unit is used for acquiring a point pair formed by every two pixel points in each direction of each pixel point in the gradient image, and calculating the point pair period length probability of each direction of the gray image by using the point pair length corresponding to the maximum frequency point pair in each direction;
calculating the point pair length of the kth line point pair in each direction in the gray scale image by using the Euclidean distance between the starting point and the ending point, and acquiring the line point pair length corresponding to the maximum frequency in the s direction
Figure 375255DEST_PATH_IMAGE002
；
Similarly, calculating all the point pair lengths in the direction perpendicular to the s-th direction from the k-th row to obtain the column point pair length, and obtaining the column point pair length with the maximum frequency in the corresponding direction
Figure 632930DEST_PATH_IMAGE004
；
The expression for calculating the cycle length probability in the s-th direction is:
Figure DEST_PATH_IMAGE006A
wherein,
Figure 135936DEST_PATH_IMAGE008
represents the maximum frequency column point pair length in the direction perpendicular to the s-th angle from the k-th row,
Figure 772454DEST_PATH_IMAGE010
the s angular direction from the k row of point pairs is the first row, the point pairs parallel to the lower i row have the maximum frequency row point pair length in the s direction,
Figure 925218DEST_PATH_IMAGE002
represents the maximum frequency row point pair length of the k-th row point pair in the s-th direction, N represents the total N row point pairs,
Figure 404610DEST_PATH_IMAGE012
representing the probability of the period length of the point pairs in the s direction of the gray level image;
the second calculation unit is used for taking the direction corresponding to the maximum value of the period length probability obtained by all the point pairs as the extension direction of the period, obtaining the period length in the period extension direction, and setting a sliding window parameter according to the period length and the period extension direction;
the third calculation unit is used for performing sliding window on the gray level image by using a window with set parameters, acquiring a frequency domain space image of each sliding window image, and obtaining an intersection frequency value corresponding to two sliding window images according to the frequency domain space intersection of every two sliding windows;
the fourth calculation unit is used for respectively carrying out filtering processing on the sliding window image by using the intersection frequency value of each sliding window image and other sliding window images to obtain a filtering image of the sliding window image after filtering processing at different intersection frequency values;
and the defect detection unit is used for calculating the defect probability of each sliding window image by using the contrast value of each pixel point in all the filtering images corresponding to each sliding window image, and judging whether the sliding window image has defects or not according to the defect probability.
CN202211044200.8A 2022-08-30 2022-08-30 Method and system for detecting textile surface defects Active CN115131348B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211044200.8A CN115131348B (en) 2022-08-30 2022-08-30 Method and system for detecting textile surface defects

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211044200.8A CN115131348B (en) 2022-08-30 2022-08-30 Method and system for detecting textile surface defects

Publications (2)

Publication Number Publication Date
CN115131348A CN115131348A (en) 2022-09-30
CN115131348B true CN115131348B (en) 2022-11-15

Family

ID=83387648

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211044200.8A Active CN115131348B (en) 2022-08-30 2022-08-30 Method and system for detecting textile surface defects

Country Status (1)

Country Link
CN (1) CN115131348B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115861310B (en) * 2023-02-23 2023-05-16 深圳市富安娜艺术家居有限公司 Method for detecting textile defects on surface of bed sheet
CN116721101B (en) * 2023-08-09 2023-10-20 山东华航高分子材料有限公司 Defect detection method for surface wave flow marks of plastic part
CN116949732B (en) * 2023-09-20 2023-12-08 济宁华晟服装股份有限公司 Efficient ironing method for clothing production
CN117132593B (en) * 2023-10-25 2024-03-26 济宁华晟服装股份有限公司 Cloth roughness detection method for resisting periodic texture influence

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114820631A (en) * 2022-07-04 2022-07-29 南通中豪超纤制品有限公司 Fabric defect detection method capable of resisting texture interference

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114445387A (en) * 2022-01-29 2022-05-06 泗阳富艺木业股份有限公司 Fiberboard quality classification method based on machine vision
CN114723705B (en) * 2022-03-31 2023-08-22 深圳市启灵图像科技有限公司 Cloth flaw detection method based on image processing
CN114627111B (en) * 2022-05-12 2022-07-29 南通英伦家纺有限公司 Textile defect detection and identification device
CN114820610B (en) * 2022-06-29 2022-09-06 数聚(山东)医疗科技有限公司 Image processing-based defect detection method for new material medical instrument
CN114842007B (en) * 2022-07-04 2022-09-06 南通东德纺织科技有限公司 Textile wear defect detection method based on image processing

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114820631A (en) * 2022-07-04 2022-07-29 南通中豪超纤制品有限公司 Fabric defect detection method capable of resisting texture interference

Also Published As

Publication number Publication date
CN115131348A (en) 2022-09-30

Similar Documents

Publication Publication Date Title
CN115131348B (en) Method and system for detecting textile surface defects
CN117541588B (en) Printing defect detection method for paper product
CN111553929B (en) Mobile phone screen defect segmentation method, device and equipment based on converged network
CN111179225B (en) Test paper surface texture defect detection method based on gray gradient clustering
CN109816644A (en) Bearing defect automatic detection system based on multi-angle light source image
CN103793708B (en) A kind of multiple dimensioned car plate precise positioning method based on motion correction
CN115100206B (en) Printing defect identification method for textile with periodic pattern
CN103413288A (en) LCD general defect detecting method
CN108564577A (en) Solar cell segment grid defect inspection method based on convolutional neural networks
CN114972356B (en) Plastic product surface defect detection and identification method and system
CN107945200A (en) Image binaryzation dividing method
CN109540917B (en) A method for extraction and analysis of yarn appearance feature parameters in multi-angle mode
CN115311507B (en) Building board classification method based on data processing
CN106780464A (en) A kind of fabric defect detection method based on improvement Threshold segmentation
CN116883333B (en) A deep learning-based textile surface defect detection method
CN107480607A (en) A kind of method that standing Face datection positions in intelligent recording and broadcasting system
CN103866551A (en) Fabric weft inclination rapid-detection method based on machine vision
CN113838038A (en) Carbon fiber cloth cover defect detection method and device, electronic equipment and storage medium
CN104392203A (en) A Method of Spot-Stripe Noise Detection Based on Video Analysis
CN113192061A (en) LED package appearance detection image extraction method and device, electronic equipment and storage medium
CN105699386B (en) An automatic cloth inspection and marking method using a contact image sensor
CN105738376B (en) A kind of automatic cloth inspecting machine using contact-type image sensor
CN115330821B (en) Image segmentation algorithm based on watershed constraint and edge connection
CN111161264A (en) Method for segmenting TFT circuit image with defects
CN117710873A (en) Novel turning plate state detection method based on binocular vision

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant