WO2021227287A1 - 生产线上蜂窝产品规整度的检测方法及检测装置 - Google Patents
生产线上蜂窝产品规整度的检测方法及检测装置 Download PDFInfo
- Publication number
- WO2021227287A1 WO2021227287A1 PCT/CN2020/109729 CN2020109729W WO2021227287A1 WO 2021227287 A1 WO2021227287 A1 WO 2021227287A1 CN 2020109729 W CN2020109729 W CN 2020109729W WO 2021227287 A1 WO2021227287 A1 WO 2021227287A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- honeycomb
- image
- cell
- camera
- product
- Prior art date
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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
Definitions
- the invention relates to the fields of design, manufacture and application of lightweight structural products for transportation, machinery, aerospace, ship and other equipment, and in particular to a method and device for detecting regularity of honeycomb products on a production line.
- the lightweight honeycomb structure is widely used in various engineering fields due to its excellent load-bearing and energy-absorbing properties.
- various structural defects such as the bend, warpage, and cell deformity of the honeycomb core block inevitably appear, and these defects have been proved to have a greater bearing and energy absorption performance. Influence. Therefore, the inspection and evaluation of the regularity of cellular products to avoid the risk of using inferior products and to further improve the regularity of cellular products urgently need to be carried out.
- the existing related technologies mainly include:
- Chinese Patent Application Nos. 201610585321.1 and 201610585419.7 (both filing dates are July 22, 2016), respectively disclose a method and device for measuring the shape of a honeycomb core, including the following steps: covering the surface to be measured with a reflector on the honeycomb core
- the thin film adopts the vacuum adsorption method to make the reflective film close to the honeycomb core to be tested surface, and the reflective film at the honeycomb cells is recessed downward; the reflective film on the tested surface is scanned and measured to obtain honeycomb cores in different spatial positions
- the wall height can analyze the cell deformation of the honeycomb core.
- This method is based on the idea of physical length measurement and uses negative pressure adsorption film to perform detection. It can initially obtain the approximate position of the edge, but the accuracy is low and the efficiency is low. Especially for the extraction of the characteristic edge of the pore size and thin-walled honeycomb structure, it is difficult to achieve .
- the Chinese patent application number is 201710203081.9 (application date is September 1, 2017), which discloses an automatic detection method for honeycomb defects of remote sensor hoods based on machine vision, including the following steps: acquiring a honeycomb image of the hood; Preprocess the mask honeycomb image to reduce noise; perform feature extraction on the preprocessed mask honeycomb image to obtain the features of the straight line segment of the mask honeycomb edge; screen the feature vectors of normal and defective cells as positive and negative samples, and establish an artificial neural network ,train.
- Chinese patent application number is 201510740221.7 (application date November 04, 2015), which discloses a checkerboard corner automatic screening method for corner detection;
- China patent application number is 200710194135.6 (application date December 2007) 05) discloses a surface shape measuring device;
- Chinese patent application number is 200810166508.3 (application date October 08, 2008) discloses a three-dimensional shape measurement method,
- Chinese patent application number is 201010557356.7 (application date November 2010 22nd) disclosed a detection and recognition method for the triple X combination mark.
- the corresponding surface shape detection technology was reported, and this type of technology was mainly for the recognition and measurement of the shape and surface contour.
- the purpose of the present invention is to provide a method and device for detecting regularity of honeycomb products on a production line to solve the problems of complicated honeycomb quality inspection operations and inaccurate judgment results in the prior art, which are not suitable for detecting honeycomb products on the production line.
- the first aspect of the present invention provides a method for detecting regularity of honeycomb products on a production line, including:
- the adjusting the camera so that the camera can collect at least a complete row of cells of the honeycomb product on the production line perpendicular to the moving direction of the product includes:
- the camera is adjusted according to the acquisition frequency.
- the calculation of the camera acquisition frequency according to the relationship between the moving speed, the side length of the cell and the size of the camera field of view satisfies the following relationship:
- f is the sampling frequency of the camera
- V is the moving speed of the cellular product
- A is the side length of the cell
- B is the length of the camera's field of view along the moving direction of the cellular product.
- the performing binarization processing on the image to obtain a binarized image includes:
- the denoising image is subjected to binarization processing to obtain a binarized image.
- the performing binarization processing on the denoising image to obtain a binarized image includes:
- the initial binarized image is subjected to morphological filtering processing to obtain a binarized image.
- the extracting the vertices of the honeycomb cells in the binarized image includes:
- the processing of taking the largest circle center of the honeycomb wall is performed on the smooth honeycomb apex image to obtain the apex of the honeycomb cell.
- the extracting the vertices of the honeycomb cells in the binarized image includes:
- the processing of taking the largest circle center of the honeycomb wall is performed on the image of the intersection of only the honeycomb walls to obtain the apex of the honeycomb cell.
- it also includes: issuing a warning when the current evaluation value of the cellular quality is higher than the preset value.
- the calculating the average deviation of all the cell angles in the honeycomb cell image, and evaluating the quality of the honeycomb according to the average value includes: filtering out n pieces of the honeycomb cell images taken continuously in the past The maximum value among the average deviations of all cell angles. When the maximum value is greater than the preset value, the honeycomb product is unqualified; and/or all cells in n images of the honeycomb cells taken continuously in the past are calculated The ratio of the number of angle deviations that exceed the preset value to the number of all cell angles. When the ratio is greater than the preset value, the honeycomb product is unqualified.
- the method further includes: calculating the average value of the internal angle deviations of the three cell internal angles of the common vertices in the honeycomb cell image, and evaluating the honeycomb quality according to the average value of the internal angles.
- a honeycomb quality detection device on a production line including:
- the camera adjustment module is used to adjust the camera so that the camera can at least collect a complete list of cells of the honeycomb product on the production line perpendicular to the moving direction of the product;
- the binarization processing module is configured to perform binarization processing on the honeycomb image to obtain a binarized image
- a vertex extraction module for extracting the vertices of honeycomb cells in the binarized image
- a cellular cell image reconstruction module configured to reconstruct a cellular cell image according to the mapping relationship between the vertices and the cells
- the honeycomb quality detection module is used to calculate the average deviation of all cell angles in the cell image of the honeycomb, and evaluate the quality of the honeycomb according to the average.
- the camera adjustment module includes:
- the parameter acquisition unit is used to acquire the moving speed of the cellular product on the production line, the side length of the cell and the size of the camera's field of view;
- a frequency calculation unit configured to calculate the camera acquisition frequency according to the relationship between the moving speed, the side length of the cell and the size of the camera field of view;
- the camera adjustment unit is used to adjust the camera according to the acquisition frequency.
- the frequency calculation unit calculates that the camera acquisition frequency satisfies the following relationship:
- f is the sampling frequency of the camera
- V is the moving speed of the cellular product
- A is the side length of the cell
- B is the length of the camera's field of view along the moving direction of the cellular product.
- the binarization processing module includes:
- An image denoising unit configured to perform filtering processing on the image to remove noise to obtain a denoised image
- the binarization processing unit is configured to perform binarization processing on the denoising image to obtain a binarized image.
- the binarization processing module further includes: a filtering unit
- binarization processing unit After the binarization processing unit performs binarization processing on the denoised image to obtain an initial binarized image
- the filtering unit performs morphological filtering processing on the initial binarized image to obtain a binarized image.
- the vertex extraction module includes:
- a closed operation processing unit configured to perform closed operation processing on the binarized image to obtain a smooth honeycomb vertex image
- the vertex extraction unit is used to perform processing of taking the largest circle center of the honeycomb wall on the smooth honeycomb vertex image to obtain the apex of the honeycomb cell.
- the vertex extraction module includes:
- a closed operation processing unit configured to perform closed operation processing on the binarized image to obtain a smooth honeycomb vertex image
- the honeycomb wall intersection extraction unit is configured to sequentially subject the smooth honeycomb apex image to expansion processing and corrosion processing to obtain only the honeycomb wall intersection image;
- the vertex extraction unit is used to perform processing of taking the largest circle center of the honeycomb wall on the image of the intersection of the honeycomb walls to obtain the apex of the honeycomb cell.
- a shelf and a gantry also includes: a shelf and a gantry
- the storage table is used to hold the honeycomb to be tested, and a levelness indicator board is arranged on it;
- the gantry is set on the ground, and the gantry is provided with the image acquisition module, and the image acquisition module is composed of one or more high-speed cameras depending on the width of the honeycomb product, so that the image acquisition module can obtain the honeycomb product Complete image in width;
- the height of the gantry is adjustable to ensure that for honeycomb products of different heights, the distance between the camera and the upper end surface of the honeycomb product is consistent.
- a calibration module Also includes: a calibration module
- the calibration module is used in conjunction with the stage to check the accuracy of the detection device.
- honeycomb quality detection module is also used to calculate the average value of the internal angle deviations of the three cell internal angles of the common vertices in the honeycomb cell image, and evaluate the honeycomb quality according to the average value of the internal angles.
- a storage medium is provided, and a computer program is stored on the storage medium, and when the program is executed by a processor, the steps of the method described in any one of the above technical solutions are implemented.
- an electronic device including a memory, a display, a processor, and a computer program stored on the memory and running on the processor, and when the processor executes the program.
- the average deviation of the cell angle is obtained.
- the smaller the deviation average the more regular the cells are.
- a concept of regularity can be introduced, that is, the closer the honeycomb is to a full hexagon, The higher the regularity; experiments have also proved that the higher the regularity of the honeycomb product, the better the stiffness and strength, that is, the better the quality of the honeycomb product. Therefore, the present invention can determine the honeycomb product through simple operation and processing. Quality, suitable for testing the quality of honeycomb products on the production line.
- Fig. 1 is a flow chart of a method for detecting regularity of honeycomb products on a production line according to a first embodiment of the present invention
- FIG. 2 is a schematic diagram of camera sampling according to an embodiment of the present invention.
- FIG. 3 is a flowchart of a method for detecting regularity of honeycomb products on a production line according to a specific embodiment of the present invention
- FIG. 4 is a top view of a honeycomb product quality inspection device according to an alternative embodiment of the present invention.
- Fig. 5 is a front view of a honeycomb product quality inspection device according to an alternative embodiment of the present invention.
- 1 storage table
- 2 digital camera
- 3 mobile device
- 4 gantry
- 5 control system.
- a method for detecting regularity of honeycomb products on a production line including:
- adjusting the camera so that the camera can collect at least a complete row of cells of the honeycomb product on the production line perpendicular to the moving direction of the product includes:
- the calculation of the camera acquisition frequency according to the relationship between the moving speed, the side length of the cell and the size of the camera's field of view satisfies the following relationship:
- f is the sampling frequency of the camera
- V is the moving speed of the cellular product
- A is the side length of the cell
- B is the length of the camera's field of view along the moving direction of the cellular product.
- performing binarization processing on an image to obtain a binarized image includes:
- performing binarization processing on the denoising image to obtain the binarized image includes:
- Morphological filtering is performed on the initial binarized image to obtain a binarized image.
- extracting the vertices of the honeycomb cells in the binarized image includes:
- the processing of taking the largest circle center of the honeycomb wall on the smooth honeycomb apex image is performed to obtain the apex of the honeycomb cell.
- extracting the vertices of the honeycomb cells in the binarized image includes:
- the smooth honeycomb apex image is sequentially subjected to expansion processing and corrosion processing to obtain only the image of the intersection of the honeycomb walls;
- the processing of taking the largest circle center of the honeycomb wall is performed to obtain the apex of the honeycomb cell.
- a method for real-time detection of cell regularity of a honeycomb cell includes the following steps: camera setting, image acquisition, image processing, vertex extraction, cell reconfiguration Structure, quality assessment;
- S1 Camera setting is to set the sampling frequency of the camera according to the moving speed of cellular products on the production line and the size of the camera's field of view.
- f The length of a single cell due to the moving direction of the sample is Where A is the side length of the cell.
- the length of the repeated area of two adjacent photos should be within with In between, considering the accuracy of the camera's shooting field of view, the actual length of the repeated area should be between 2A and 2.5A, so the range of the camera sampling frequency f is: V/(B-2A) ⁇ f ⁇ V/(B -2.5A), where V is the moving speed of the cellular product, and B is the length of the camera's field of view along the moving direction of the cellular product;
- S2 Obtaining images includes shooting images and computer reading images
- S3 Image processing; the sequence of image processing includes: noise reduction filtering, binarization, morphological filtering to obtain morphological images; among them, noise reduction filtering uses median filtering to filter out the noise of the image; binarization: the product outline The pixel value of the image is set to 1, and the pixel value of the background image of the product is set to 0; Morphological filtering is to eliminate pixels with an area smaller than a given threshold and reduce the error caused by binarization;
- the skeletonization is based on the morphological image, and the pixel value is 1
- the line uses a line segment with a line width of 1 pixel to draw a skeleton diagram
- the second step uses a 5 ⁇ 5 pixel window to calculate the corner response function value R of each pixel, for which the R value is greater than the maximum R value of all pixels 1% of the pixel point and the maximum value of the 3 ⁇ 3 neighborhood with it as the center, extract its coordinates and record it as a vertex
- the third method of vertex extraction is the skeletonization process in the first step
- the second step is in On the basis of the skeleton diagram, for all pixels with a pixel value of 1, count the number of changes in the eight-neighbor pixel value in turn clockwise or counterclockwise. If the number of changes encountered is 6, or the number of changes is 4 and the pixel is For pixels that are not on the same straight line as the other two points of the eight neighbors, their coordinates are extracted and recorded as vertices;
- Cell reconstruction is to connect the extracted vertices according to the mapping relationship between cells and vertices to obtain a cell reconstruction graph; the first method of cell reconstruction: traverse the image and encounter a pixel with a pixel value of 0 Point, the Moore neighborhood tracking algorithm is used for boundary tracking, and a window is made with each boundary point as the center to determine whether there are vertices in the window.
- stop tracking and set the pixel value of the cell to 1 repeat the above process again until there is no more point with the pixel value of 0, and connect the vertices of each cell in sequence to complete the sequence.
- the quality evaluation is based on the current cell reconstruction map.
- the first step is to calculate: that is, calculate the angular deviation value and total average value of all cells, and the linear deviation value and total average value;
- Two-step judgment that is, compared with the set tolerance zone, the one that falls within the tolerance zone is judged as qualified, otherwise it is judged as unqualified.
- the above method is novel and efficient, and can realize the real-time detection of the geometric regularity of the honeycomb product on the production line, and realize the real-time quality monitoring in the production process of the honeycomb product.
- This step is the cellular quality warning function on the production line, that is, if the current cellular quality evaluation value is higher than the preset value, a warning will be issued and the machine will be shut down, and production will resume after inspection and troubleshooting.
- the calculating the average value of the deviation of all the cell angles in the cell image of the honeycomb, and evaluating the quality of the honeycomb according to the average value includes: filtering out n pieces of the honeycomb that have been continuously photographed in the past.
- the maximum value among the average deviations of all cell angles in the cell image, and when the maximum value is greater than the preset value, the honeycomb product is unqualified; and/or the n honeycomb cell images taken continuously in the past are calculated The ratio of the number of deviations of all cell angles in the average value exceeding the preset value to the number of all cell angles. When the ratio is greater than the preset value, the honeycomb product is unqualified.
- the average value of the deviations of the three cell internal angles of the common vertices in the honeycomb cell image is calculated, and the honeycomb quality is evaluated according to the average value.
- a honeycomb quality detection device on a production line including:
- the camera adjustment module is used to adjust the camera so that the camera can at least collect a complete list of cells of the honeycomb product on the production line perpendicular to the moving direction of the product;
- the binarization processing module is used to perform binarization processing on the honeycomb image to obtain a binarized image
- the vertex extraction module is used to extract the vertices of the honeycomb cells in the binarized image
- the cellular cell image reconstruction module is used to reconstruct the cellular cell image according to the mapping relationship between vertices and cells;
- the honeycomb quality detection module is used to calculate the average deviation of all cell angles in the image of the honeycomb cell, and evaluate the quality of the honeycomb according to the average value.
- the camera adjustment module includes:
- the parameter acquisition unit is used to acquire the moving speed of the cellular product on the production line, the side length of the cell and the size of the camera's field of view;
- the frequency calculation unit is used to calculate the camera acquisition frequency according to the relationship between the moving speed, the side length of the cell and the size of the camera's field of view;
- the camera adjustment unit is used to adjust the camera according to the acquisition frequency.
- the frequency calculation unit calculates that the camera acquisition frequency satisfies the following relationship:
- f is the sampling frequency of the camera
- V is the moving speed of the cellular product
- A is the side length of the cell
- B is the length of the camera's field of view along the moving direction of the cellular product.
- the binarization processing module includes: an image denoising unit for filtering the image to remove noise to obtain a denoised image; a binarization processing unit for binarizing the denoised image Processing to obtain a binarized image.
- the binarization processing module further includes: a filtering unit; the denoising image is binarized in the binarization processing unit to obtain the initial binarized image; the filtering unit is used to binarize the initial The image undergoes morphological filtering to obtain a binary image.
- the vertex extraction module includes: a closed operation processing unit, configured to perform closed operation processing on the binarized image to obtain a smooth honeycomb vertex image;
- the vertex extraction unit is used for processing the honeycomb wall to take the largest circle center on the smooth honeycomb vertex image to obtain the apex of the honeycomb cell.
- the vertex extraction module includes: a closed operation processing unit, configured to perform closed operation processing on the binarized image to obtain a smooth honeycomb vertex image;
- the honeycomb wall intersection extraction unit is used to sequentially subject the smooth honeycomb apex image to expansion and corrosion processing to obtain only the honeycomb wall intersection image;
- the vertex extraction unit is used to perform processing of taking the largest circle center of the honeycomb wall on the image of the intersection of the honeycomb wall to obtain the apex of the honeycomb cell.
- the present invention also includes: a shelf and a gantry; the shelf is used to hold the honeycombs to be inspected, on which a levelness indicator board is set; the gantry is set on the ground, and the gantry is provided with image acquisition Module, the image acquisition module is composed of one or more high-speed cameras depending on the width of the honeycomb product, so that the image acquisition module can obtain a complete image of the width of the honeycomb product;
- the height of the gantry is adjustable to ensure that for honeycomb products of different heights, the distance between the camera and the upper end surface of the honeycomb product is consistent; in the embodiment of the present invention, it also includes: a calibration module; Check the accuracy of the detection device.
- the honeycomb quality detection module is also used to calculate the average value of the internal angle deviations of the three internal angles of the common vertices in the honeycomb cell image, and evaluate the honeycomb quality according to the average value of the internal angles.
- a storage medium is provided, and a computer program is stored on the storage medium, and when the program is executed by a processor, the steps of any one of the methods in the foregoing embodiments are implemented.
- an electronic device including a memory, a display, a processor, and a computer program stored on the memory and running on the processor, and the processor executes the The program implements the steps of any one of the above-mentioned embodiments.
- a device for real-time detection of cellular cell regularity in a production line which includes a shelf, a gantry, a digital camera and a control system, a mobile device, and a calibration module; the digital camera is connected to the control system;
- Level adjustment device and level indicator board are set on the storage table
- At least one digital camera with a resolution of not less than 1080P and equipped with a telecentric lens to obtain high-resolution cellular product photos and reduce its distortion within the depth of field;
- the control system includes a system control module, a calculation analysis module and a result display module;
- the control module controls the start and stop of the system and the movement of the camera moving device
- the calculation analysis module uses the corresponding analysis software to analyze the photos collected by the digital camera, calculates the geometric regularity of the honeycomb sample, and evaluates the geometric regularity of the honeycomb sample according to the selected evaluation standard and threshold, and transmits the evaluation result to the result Indicating value module, the result is displayed by the result indicating module;
- the result display module can display according to the product quality evaluation result, the green light is displayed for qualified quality, and the red light is displayed for unqualified.
- the calibration board is a display board with an electronic ink screen that can display standard honeycombs with adjustable side lengths and wall thicknesses. The outside of the screen displays a color that contrasts with the honeycombs. Place the calibration board on the stage, adjust the digital camera to a suitable position, obtain the photo of the calibration board, and pass it to the software of the control system for calibration, and check the detection accuracy of the system
- the present invention aims to protect a method for detecting the regularity of honeycomb products on a production line, including: adjusting the camera so that the camera can at least collect a complete list of cells of the honeycomb product on the production line perpendicular to the moving direction of the product; acquiring the honeycomb image; The image is binarized to obtain a binarized image; the vertices of the honeycomb cell in the binarized image are extracted; the honeycomb cell image is reconstructed according to the mapping relationship between the vertices and the cells; all the cells in the honeycomb cell image are calculated The average value of the deviation of the element angle, and the honeycomb quality is evaluated based on the average value.
- the method is novel and efficient, and can realize the real-time detection of the geometric regularity of the honeycomb product on the production line, and realize the real-time quality monitoring in the production process of the honeycomb product.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Length Measuring Devices By Optical Means (AREA)
- Image Analysis (AREA)
Abstract
一种生产线上蜂窝产品规整度的检测方法、装置及电子设备，其中，生产线上蜂窝产品规整度的检测方法，包括：调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元（S1）；获取蜂窝图像（S2）；将蜂窝图像进行二值化处理，得到二值化图像（S3）；提取二值化图像中蜂窝胞元的顶点（S4）；根据顶点与胞元的映射关系，重构得到蜂窝胞元图像（S5）；计算蜂窝胞元图像中所有胞元角的偏差平均值，并根据平均值评价蜂窝质量（S6）。该方法新颖、高效，可实现蜂窝产品几何规整性的生产线实时检测，实现蜂窝产品生产过程中的实时质量监控。
Description
本发明涉及交通、机械、航空航天、船舶等装备的轻质结构产品设计、制造及应用等领域，尤其涉及一种生产线上蜂窝产品规整度的检测方法及检测装置。
轻质蜂窝结构以其优异的承载与吸能特性而被广泛应用到各种工程领域。然而，在该产品的生产制造过程中，不可避免地出现蜂窝芯块拱弯、翘曲、胞孔畸形等各型结构性缺陷，而这些缺陷已被证实对其承载与吸能性能产生较大影响。因此，有关蜂窝产品规整性检测与评估以规避低劣产品的使用风险，进一步改进蜂窝规整度的工作亟待开展。
由于蜂窝产品为周期排列多孔结构，具有典型的多顶点、细薄壁、承载面宽等特征，传统超声检测技术无法获得其结构性缺陷的特征信息。现有相关技术主要包括：
中国专利申请号为201610585321.1和201610585419.7(申请日均为2016年07月22日)，分别公开了一种蜂窝芯面形的测量方法及实现装置，包括如下步骤：在蜂窝芯待测面覆上反射薄膜，采用真空吸附的方式使所述反射薄膜紧贴蜂窝芯待测面，且使蜂窝孔格处的反射薄膜向下凹陷；对待测面反射薄膜扫描测量，获得蜂窝芯在不同空间位置的蜂窝壁高度，能够分析蜂窝芯的孔格变形。该方法基于物理长度测定的思想，利用负压吸附薄膜实施检测，可初步获得棱边的大致位置，但精度差、效率低，尤其对细孔径、薄壁蜂窝结构的特征边提取，实现难度大。
中国专利申请号为201710203081.9(申请日为2017年9月1日)，公开了一种基于机器视觉的遥感器遮光罩蜂窝缺陷自动检测方法，包括以下步骤：获取遮光罩蜂窝图像；对获取的遮光罩蜂窝图像进行预处理，减少噪声；对经过预处理的遮光罩蜂窝图像进行特征提取，得到遮光罩蜂窝边缘直线段特征；筛选正常蜂窝和缺陷蜂窝的特征向量作为正负样本，人工神经网络建立、训练。中国优秀硕士论文全文数据库收录的2017年王薇所作的《基于机器视觉的蜂窝结构三维外形测量技术研究》，公开了采用正六边形和正四边形网格的图像化识别方法，提出了一种基于直线分段(LSD)的单元网格处理方法，获取单个网格边界信息，通过计算待评估直线区域内像素与该区域矩形包围盒夹角判定是否为目标直线段，从而实现单元网格边缘线段的提取，进一步定位网格交点。该类方法主要定位于规则几何六边形与四边形的线段提取，仅涉及了单一胞元蜂窝的线条特征提取。
除此以外，中国专利申请号为201510740221.7(申请日2015年11月04日)，公开了一种边角检测的棋盘格角点自动筛选方法；中国专利申请号为200710194135.6(申请日2007年12月05日)公开了一种表面形状测定装置；中国专利申请号为200810166508.3(申请日2008年10月08日)公开了一种三维形状测量方法，中国专利申请号为201010557356.7(申请日2010年11月22日)公开了一种三X组合标记的检测识别方法均报导了相应的表面形状检测技术，该类技术均只主要针对形状表面轮廓进行识别与测定。
发明内容
(一)发明目的
本发明的目的是提供一种生产线上蜂窝产品规整度的检测方法及检测装置以解决现有技术对蜂窝质量检测操作复杂及判断结果不准确，不适用生产线上蜂窝产品的检测问题。
(二)技术方案
为解决上述问题，本发明的第一方面提供了一种生产线上蜂窝产品规整 度的检测方法，包括：
调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元；
获取蜂窝图像；
将所述蜂窝图像进行二值化处理，得到二值化图像；
提取所述二值化图像中蜂窝胞元的顶点；
根据所述顶点与胞元的映射关系，重构得到蜂窝胞元图像；
计算所述蜂窝胞元图像中所有胞元角的偏差平均值，并根据所述平均值评价所述蜂窝质量。
进一步地，所述调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元包括：
获取生产线上蜂窝产品的移动速度、胞元边长及相机视场的大小；
根据所述移动速度、所述胞元边长及所述相机视场的大小之间的关系计算相机采集频率；
根据所述采集频率调节所述相机。
进一步地，所述根据所述移动速度、所述胞元边长及所述相机视场的大小之间的关系计算相机采集频率满足下列关系：
V/(B-2A)＜f＜V/(B-2.5A)
其中，f为相机的采样频率；V为蜂窝产品的移动速度；A为胞元边长；B为相机视场沿蜂窝产品移动方向的长度。
进一步地，所述将所述图像进行二值化处理，得到二值化图像包括：
将所述图像进行滤波处理去除噪声，得到去噪图像；
将所述去噪图像进行二值化处理，得到二值化图像。
进一步地，所述将所述去噪图像进行二值化处理，得到二值化图像包括：
将所述去噪图像进行二值化处理，得到初始二值化图像；
将所述初始二值化图像进行形态学滤波处理，得到二值化图像。
进一步地，所述提取所述二值化图像中蜂窝胞元的顶点包括：
将所述二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；
在所述平滑蜂窝顶点图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
进一步地，所述提取所述二值化图像中蜂窝胞元的顶点包括：
将所述二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；
将所述平滑蜂窝顶点图像依次经过膨胀处理和腐蚀处理，得到只有蜂窝壁交汇处图像；
在所述只有蜂窝壁交汇处图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
进一步地，还包括：当前蜂窝质量的评价值高于预设值时，发出警示。
进一步地，所述计算所述蜂窝胞元图像中所有胞元角的偏差平均值，并根据所述平均值评价所述蜂窝质量包括：筛选出过去连续拍摄的n张所述蜂窝胞元图像中所有胞元角的偏差平均值中最大值，当所述最大值大于预设值时，所述蜂窝产品不合格；和/或计算过去连续拍摄的n张所述蜂窝胞元图像中所有胞元角的偏差平均值中超过预设值的数量占所有胞元角的数量比，当比例大于预设值时，所述蜂窝产品不合格。
进一步地，首先计算过去连续拍摄的n张图像中各张图像的所有胞元角的偏差平均值，然后可以采取下述指标的一种或两种组合对蜂窝产品的质量进行评价：求出过去连续拍摄的n张图像中各张图像的胞元角的偏差平均值中的最大值，若该值大于预设值，则当前蜂窝产品质量不合格；过去连续拍摄的n张图像中胞元角的偏差平均值大于预设值的单张图像个数占n张图像的比例大于预设值，则当前蜂窝产品的质量不合格。
进一步地，还包括：计算蜂窝胞元图像中共顶点的三个胞元内角的内角偏差平均值，并根据所述内角平均值评价蜂窝质量。
根据本发明的另一个方面，提供一种生产线上蜂窝质量检测装置，包括：
相机调节模块，用于调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元；
相机，用于获取蜂窝图像；
二值化处理模块，用于将所述蜂窝图像进行二值化处理，得到二值化图像；
顶点提取模块，用于提取所述二值化图像中蜂窝胞元的顶点；
蜂窝胞元图像重构模块，用于根据所述顶点与胞元的映射关系，重构得到蜂窝胞元图像；
蜂窝质量检测模块，用于计算所述蜂窝胞元图像中所有胞元角的偏差平均值，并根据所述平均值评价所述蜂窝质量。
进一步地，所述相机调节模块包括：
参数获取单元，用于获取生产线上蜂窝产品的移动速度、胞元边长及相机视场的大小；
频率计算单元，用于根据所述移动速度、所述胞元边长及所述相机视场的大小之间的关系计算相机采集频率；
相机调节单元，用于根据所述采集频率调节所述相机。
进一步地，所述频率计算单元计算相机采集频率满足下列关系：
V/(B-2A)＜f＜V/(B-2.5A)
其中，f为相机的采样频率；V为蜂窝产品的移动速度；A为胞元边长；B为相机视场沿蜂窝产品移动方向的长度。
进一步地，所述二值化处理模块包括：
图像去噪单元，用于将所述图像进行滤波处理去除噪声，得到去噪图像；
二值化处理单元，用于将所述去噪图像进行二值化处理，得到二值化图像。
进一步地，所述二值化处理模块还包括：滤波单元
在所述二值化处理单元将所述去噪图像进行二值化处理，得到初始二值化图像后；
所述滤波单元，将所述初始二值化图像进行形态学滤波处理，得到二值化图像。
进一步地，所述顶点提取模块包括：
闭运算处理单元，用于将所述二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；
顶点提取单元，用于在所述平滑蜂窝顶点图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
进一步地，所述顶点提取模块包括：
闭运算处理单元，用于将所述二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；
蜂窝壁交汇处提取单元，用于将所述平滑蜂窝顶点图像依次经过膨胀处理和腐蚀处理，得到只有蜂窝壁交汇处图像；
顶点提取单元，用于在所述蜂窝壁交汇处图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
进一步地，还包括：置物台、龙门架；
所述置物台用于盛放待检测蜂窝，其上设置有水平度示值板；
所述龙门架设置地面上，且所述龙门架设置有所述图像获取模块，所述图像获取模块视蜂窝产品的宽度由一台或多台高速相机组成，使得该图像获取模块可获得蜂窝产品宽度上的完整图像；
所述龙门架高度可调，以保证对于不同高度的蜂窝产品，相机与蜂窝产品上端面的距离一致。
进一步地，还包括：标定模块；
所述标定模块与所述置物台配合使用，用于校核检测装置的准确性。
进一步地，所述蜂窝质量检测模块还用于计算蜂窝胞元图像中共顶点的三个胞元内角的内角偏差平均值，并根据所述内角平均值评价蜂窝质量。
根据本发明的又一方面，提供一种储存介质，所述存储介质上存储有计算机程序，所述程序被处理器执行时实现上述技术方案中任意一项所述方法的步骤。
根据本发明的又一方面，提供一种电子设备，包括存储器、显示器、处 理器及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述程序时实现上述技术方案中任意一项所述方法的步骤。
(三)有益效果
本发明的上述技术方案具有如下有益的技术效果：
通过本发明方法及装置对蜂窝图像处理得到胞元角的偏差平均值，偏差平均值越小，说明胞元越规整，在这里可以引入一个规整度的概念，就是蜂窝越接近整六边形，规整度就越高；通过实验也证明了，规整度越高的蜂窝产品刚度和强度都越好，即蜂窝产品的质量越好，因此，本发明通过简单的操作处理即可判断出蜂窝产品的质量，适于生产线上蜂窝产品质量的检测。
图1是根据本发明第一实施方式的生产线上蜂窝产品规整度的检测方法流程图；
图2是根据本发明一具体实施方式的相机采样示意图；
图3是根据本发明一具体实施方式的生产线上蜂窝产品规整度的检测方法流程图；
图4是根据本发明一可选实施方式的蜂窝产品质量检测装置的俯视图；
图5是根据本发明一可选实施方式的蜂窝产品质量检测装置的主视图。
附图标记：
1：置物台；2：数码相机；3：移动装置；4：龙门架；5：控制系统。
为使本发明的目的、技术方案和优点更加清楚明了，下面结合具体实施方式并参照附图，对本发明进一步详细说明。应该理解，这些描述只是示例性的，而并非要限制本发明的范围。此外，在以下说明中，省略了对公知结构和技术的描述，以避免不必要地混淆本发明的概念。
显然，所描述的实施例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。
此外，下面所描述的本发明不同实施方式中所涉及的技术特征只要彼此之间未构成冲突就可以相互结合。
如图1所示，在本发明实施例的第一方面，提供了一种生产线上蜂窝产品规整度的检测方法，包括：
S1：调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元，如图2所示；
S2：获取蜂窝图像；
S3：将蜂窝图像进行二值化处理，得到二值化图像；
S4：提取二值化图像中蜂窝胞元的顶点；
S5：根据顶点与胞元的映射关系，重构得到蜂窝胞元图像；
S6：计算蜂窝胞元图像中所有胞元角的偏差平均值，并根据平均值评价蜂窝质量。
在本发明实施例中，调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元包括：
获取生产线上蜂窝产品的移动速度、胞元边长及相机视场的大小；
根据移动速度、胞元边长及相机视场的大小之间的关系计算相机采集频率；
根据采集频率调节相机。
在本发明实施例中，根据移动速度、胞元边长及相机视场的大小之间的关系计算相机采集频率满足下列关系：
V/(B-2A)＜f＜V/(B-2.5A)
其中，f为相机的采样频率；V为蜂窝产品的移动速度；A为胞元边长；B为相机视场沿蜂窝产品移动方向的长度。
在本发明实施例中，将图像进行二值化处理，得到二值化图像包括：
将图像进行滤波处理去除噪声，得到去噪图像；
将去噪图像进行二值化处理，得到二值化图像。
在本发明实施例中，将去噪图像进行二值化处理，得到二值化图像包括：
将去噪图像进行二值化处理，得到初始二值化图像；
将初始二值化图像进行形态学滤波处理，得到二值化图像。
在本发明实施例中，提取二值化图像中蜂窝胞元的顶点包括：
将二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；
在平滑蜂窝顶点图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
在本发明实施例中，提取二值化图像中蜂窝胞元的顶点包括：
将二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；
将平滑蜂窝顶点图像依次经过膨胀处理和腐蚀处理，得到只有蜂窝壁交汇处图像；
在只有蜂窝壁交汇处图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
如图3所示，在本发明一具体实施例中，提供一种蜂窝胞元规整度的生产线实时检测方法，方法包括以下步骤：相机设定、获取图像、图像处理、顶点提取、胞元重构、质量评估；
S1：相机设定即根据生产线上蜂窝产品的移动速度及相机视场大小，设定相机的采样频率f：因样本移动方向单个胞元所占长度为
其中A为胞元边长，为保证当前照片边界处的胞元能完整地进入下一张照片，从而被检测到，且不过多重复检测，相邻两张照片的重复区域长度应在
和
之间，考虑相机拍摄视场精度问题，实际重复区域长度取值应在2A和2.5A之间，因此相机采样频率f取值范围为：V/(B-2A)＜f＜V/(B-2.5A)，其中V为蜂窝产品的移动速度，B为相机视场沿蜂窝产品移动方向的长度；
S2：获取图像包括拍摄图像和计算机读取图像；
S3：图像处理；图像处理顺序包括：降噪滤波、二值化、形态学滤波，获得形态图像；其中，降噪滤波是采用中值滤波法滤除图像的噪声；二值化： 将产品轮廓图像的像素值置为1，将产品的背景图像的像素值置为0；形态学滤波是消除面积小于给定阈值的像素，降低二值化带来的误差；
S4：提取顶点；顶点提取是在图像处理的基础上寻找胞元的顶点并记录；顶点提取的第一种方法：采用遍历图像后其内最小的像素值为0的像素数大于0的最小窗口作为统计窗口，采用该窗口再次遍历图像，将窗口内的像素值为1的像素数赋给窗口中心点上，提取像素值为1的像素数最大值的点，将其记录为顶点，采用边长等于胞元边长的湮灭窗口，将该点为中心的湮灭窗口内的像素值为1的像素数置0，再次提取像素值为1的像素数最大值的点并记录，不断重复这个步骤，直到像素值为1的像素数小于给定阈值，则顶点提取完毕；顶点提取的第二种方法的第一步进行骨架化处理，骨架化是以形态图像为基础，将像素值为1的线条采用线宽为1个像素的线段绘制骨架图；第二步采用5×5个像素的窗口计算每个像素点的角点响应函数值R，对于其R值大于所有像素点的最大R值的1％且是以其为中心的3×3邻域的最大值的像素点，提取其坐标并记录为顶点；顶点提取的第三种方法第一步进行骨架化处理；第二步是在骨架图的基础上，对所有像素值为1的像素点，顺时针或逆时针依次统计其八邻域像素值的变化次数，若遇到变化次数为6，或变化次数为4且该像素点与其八邻域的另两点不在同一条直线上的像素点，则提取其坐标并记录为顶点；
S5：胞元重构是将提取的顶点依据胞元与顶点的映射关系连线，得到胞元重构图；胞元重构的第一种方法：遍历图像，遇到像素值为0的像素点，则采用摩尔邻域追踪算法进行边界追踪，并以每个边界点为中心做一个窗口，判断该窗口内有无顶点，若有则记录其编号并依次为其标记序号，遇到该胞元的起始追踪像素时，停止追踪并将该胞元的像素值置为1，再次重复上述过程，直到不再存在像素值为0的点，将各胞元的顶点按序号依次连线完成胞元的重构；胞元重构的第二种方法：对于每个顶点，计算其余所有顶点与它的距离，选取距离最近的三点并记录，计算所有顶点与最近三点的距离并求和，除以顶点数的两倍，得到平均蜂窝胞元边长A，选取位于图像边缘向 内1A～2A宽度之外区域的顶点，对于其中的每个顶点，分别与最近的三点进行连线，从而得到蜂窝的重构图像；
S6：质量评估是以当前的胞元重构图为基础，第一步计算：即：计算出所有胞元的角偏差值及其总的平均值、线偏差值及其总的平均值；第二步判断：即：与设置的公差带相比较，落在公差带范围内的判定为合格，否则判定为不合格。
上述方法新颖、高效，可实现蜂窝产品几何规整性的生产线实时检测，实现蜂窝产品生产过程中的实时质量监控。
在本发明实施例中，还包括：
S7：当前蜂窝质量的评价值高于预设值时，发出警示。
该步骤是生产线上蜂窝质量警示功能，即当前蜂窝质量的评价值高于预设值，则发出警示，并停机，经检查排除故障后恢复生产。
在本发明实施例中，所述计算所述蜂窝胞元图像中所有胞元角的偏差平均值，并根据所述平均值评价所述蜂窝质量包括：筛选出过去连续拍摄的n张所述蜂窝胞元图像中所有胞元角的偏差平均值中最大值，当所述最大值大于预设值时，所述蜂窝产品不合格；和/或计算过去连续拍摄的n张所述蜂窝胞元图像中所有胞元角的偏差平均值中超过预设值的数量占所有胞元角的数量比，当比例大于预设值时，所述蜂窝产品不合格。
具体为：首先计算过去连续拍摄的n张图像中各张图像的所有胞元角的偏差平均值，然后可以采取下述指标的一种或两种组合对蜂窝产品的质量进行评价，1.求出过去连续拍摄的n张图像中各张图像的胞元角的偏差平均值中的最大值，若该值大于预设值，则当前蜂窝产品质量不合格；2.故去连续拍摄的n张图像中胞元角的偏差平均值大于预设值的单张图像个数占n张图像的比例大于预设值，则当前蜂窝产品的质量不合格，在实际检测中，n取值范围可为10～30。
在本发明实施例中，计算蜂窝胞元图像中共顶点的三个胞元内角的偏差平均值，并根据平均值评价蜂窝质量。
如图4-5所示，在本发明实施例的另一个方面，提供一种生产线上蜂窝质量检测装置，包括：
相机调节模块，用于调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元；
相机，用于获取蜂窝图像；
二值化处理模块，用于将蜂窝图像进行二值化处理，得到二值化图像；
顶点提取模块，用于提取二值化图像中蜂窝胞元的顶点；
蜂窝胞元图像重构模块，用于根据顶点与胞元的映射关系，重构得到蜂窝胞元图像；
蜂窝质量检测模块，用于计算蜂窝胞元图像中所有胞元角的偏差平均值，并根据平均值评价蜂窝质量。
在本发明实施例中，相机调节模块包括：
参数获取单元，用于获取生产线上蜂窝产品的移动速度、胞元边长及相机视场的大小；
频率计算单元，用于根据移动速度、胞元边长及相机视场的大小之间的关系计算相机采集频率；
相机调节单元，用于根据采集频率调节相机。
在本发明实施例中，频率计算单元计算相机采集频率满足下列关系：
V/(B-2A)＜f＜V/(B-2.5A)
其中，f为相机的采样频率；V为蜂窝产品的移动速度；A为胞元边长；B为相机视场沿蜂窝产品移动方向的长度。
在本发明实施例中，二值化处理模块包括：图像去噪单元，用于将图像进行滤波处理去除噪声，得到去噪图像；二值化处理单元，用于将去噪图像进行二值化处理，得到二值化图像。
在本发明实施例中，二值化处理模块还包括：滤波单元；在二值化处理单元将去噪图像进行二值化处理，得到初始二值化图像后；滤波单元，将初始二值化图像进行形态学滤波处理，得到二值化图像。
在本发明实施例中，顶点提取模块包括：闭运算处理单元，用于将二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；
顶点提取单元，用于在平滑蜂窝顶点图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
在本发明实施例中，顶点提取模块包括：闭运算处理单元，用于将二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；
蜂窝壁交汇处提取单元，用于将平滑蜂窝顶点图像依次经过膨胀处理和腐蚀处理，得到只有蜂窝壁交汇处图像；
顶点提取单元，用于在蜂窝壁交汇处图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
在本发明实施例中，还包括：置物台、龙门架；置物台用于盛放待检测蜂窝，其上设置有水平度示值板；龙门架设置在地面上，且龙门架设置有图像获取模块，图像获取模块视蜂窝产品的宽度由一台或多台高速相机组成，使得该图像获取模块可获得蜂窝产品宽度上的完整图像；
所述龙门架高度可调，以保证对于不同高度的蜂窝产品，相机与蜂窝产品上端面的距离一致；在本发明实施例中，还包括：标定模块；标定模块与置物台配合使用，用于校核检测装置的准确性。在本发明实施例中，蜂窝质量检测模块还用于计算蜂窝胞元图像中共顶点的三个胞元内角的内角偏差平均值，并根据内角平均值评价蜂窝质量。
在本发明实施例的又一方面，提供一种储存介质，存储介质上存储有计算机程序，程序被处理器执行时实现上述实施例中任意一项方法的步骤。
在本发明实施例的又一方面，提供一种电子设备，包括存储器、显示器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述程序时实现上述实施例中任意一项所述方法的步骤。
在本发明一具体实施例中，提供一种蜂窝胞元规整度的生产线实时检测装置，包括置物台、龙门架、数码相机和控制系统、移动装置、标定模块；数码相机和控制系统相连接；
置物台上设置水平调节装置和水平度示值板；
数码相机至少为一台，其分辨率不低于1080P，配置远心镜头，以得到高分辨率的蜂窝产品照片，且减小其在景深范围内的畸变；
当一台相机不足以捕捉蜂窝产品宽度方向上的完整图像时，则在宽度方向上按一定间隔布置多台相机，使得该相机阵列可以得到蜂窝产品宽度上的完整图像；
当采用多台相机阵列安装时，其可在移动装置的驱动下沿横梁横向移动，调节各相机之间的间距；
控制系统包括系统控制模块、计算分析模块及结果示值模块；
控制模块控制系统启停及相机移动装置的运动；
计算分析模块采用相应分析软件对数码相机采集的照片进行分析，计算出蜂窝样品的几何规整度，并根据选定的评估标准及阈值对蜂窝样品的几何规整度进行评估，将评估结果传递给结果示值模块，由结果示值模块进行结果显示；
结果示值模块可根据产品质量评定结果进行显示，质量合格显示绿灯，不合格显示红灯。
标定模块：标定板为一个采用电子墨水屏的显示板，可显示边长、壁厚可调的标准蜂窝，屏幕外侧显示与蜂窝成对比色的颜色。将该标定板放置于置物台上，调整数码相机至合适位置，获取该标定板的照片，传递给控制系统的软件进行标定，校核系统的检测准确性
本发明旨在保护一种生产线上蜂窝产品规整度的检测方法，包括：调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元；获取蜂窝图像；将蜂窝图像进行二值化处理，得到二值化图像；提取二值化图像中蜂窝胞元的顶点；根据顶点与胞元的映射关系，重构得到蜂窝胞元图像；计算蜂窝胞元图像中所有胞元角的偏差平均值，并根据平均值评价蜂窝质量。该方法新颖、高效，可实现蜂窝产品几何规整性的生产线实时检测，实现蜂窝产品生产过程中的实时质量监控。
应当理解的是，本发明的上述具体实施方式仅仅用于示例性说明或解释本发明的原理，而不构成对本发明的限制。因此，在不偏离本发明的精神和范围的情况下所做的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。此外，本发明所附权利要求旨在涵盖落入所附权利要求范围和边界、或者这种范围和边界的等同形式内的全部变化和修改例。
Claims (14)
- 一种生产线上蜂窝产品规整度的检测方法，其特征在于，包括：调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元；获取蜂窝图像；将所述蜂窝图像进行二值化处理，得到二值化图像；提取所述二值化图像中蜂窝胞元的顶点；根据所述顶点与胞元的映射关系，重构得到蜂窝胞元图像；计算所述蜂窝胞元图像中所有胞元角的偏差平均值，并根据所述平均值评价所述蜂窝质量。
- 根据权利要求1所述的评价方法，其特征在于，所述调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元包括：获取生产线上蜂窝产品的移动速度、胞元边长及相机视场的大小；根据所述移动速度、所述胞元边长及所述相机视场的大小之间的关系计算相机采集频率；根据所述采集频率调节所述相机。
- 根据权利要求2所述的评价方法，其特征在于，所述根据所述移动速度、所述胞元边长及所述相机视场的大小之间的关系计算相机采集频率满足下列关系：V/(B-2A)＜f＜V/(B-2.5A)其中，f为相机的采样频率；V为蜂窝产品的移动速度；A为胞元边长；B为相机视场沿蜂窝产品移动方向的长度。
- 根据权利要求1所述的评价方法，其特征在于，所述将所述图像进行二值化处理，得到二值化图像包括：将所述图像进行滤波处理去除噪声，得到去噪图像；将所述去噪图像进行二值化处理，得到二值化图像。
- 根据权利要求4所述的评价方法，其特征在于，所述将所述去噪图像进行二值化处理，得到二值化图像包括：将所述去噪图像进行二值化处理，得到初始二值化图像；将所述初始二值化图像进行形态学滤波处理，得到二值化图像。
- 根据权利要求1所述的评价方法，其特征在于，所述提取所述二值化图像中蜂窝胞元的顶点包括：将所述二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；在所述平滑蜂窝顶点图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
- 根据权利要求1所述的评价方法，其特征在于，所述提取所述二值化图像中蜂窝胞元的顶点包括：将所述二值化图像进行闭运算处理，得到平滑蜂窝顶点图像；将所述平滑蜂窝顶点图像依次经过膨胀处理和腐蚀处理，得到只有蜂窝壁交汇处图像；在所述只有蜂窝壁交汇处图像上进行蜂窝壁取最大圆圆心处理，得到蜂窝胞元的顶点。
- 根据权利要求1所述的评价方法，其特征在于，包括：当前蜂窝质量的评价值高于预设值时，发出警示。
- 根据权利要求1所述的评价方法，其特征在于，所述计算所述蜂窝胞元图像中所有胞元角的偏差平均值，并根据所述平均值评价所述蜂窝质量包括：筛选出过去连续拍摄的n张所述蜂窝胞元图像中所有胞元角的偏差平均值中最大值，当所述最大值大于预设值时，所述蜂窝产品不合格；和/或计算过去连续拍摄的n张所述蜂窝胞元图像中所有胞元角的偏差平均值中超过预设值的数量占所有胞元角的数量比，当比例大于预设值时，所 述蜂窝产品不合格。
- 根据权利要求1所述的评价方法，其特征在于，还包括：计算蜂窝胞元图像中共顶点的三个胞元内角的内角偏差平均值，并根据所述内角平均值评价蜂窝质量。
- 一种生产线上蜂窝质量检测装置，其特征在于，包括：相机调节模块，用于调节相机，使相机至少能采集到生产线上蜂窝产品垂直于产品推移方向的一列完整的胞元；相机，用于获取蜂窝图像；二值化处理模块，用于将所述蜂窝图像进行二值化处理，得到二值化图像；顶点提取模块，用于提取所述二值化图像中蜂窝胞元的顶点；蜂窝胞元图像重构模块，用于根据所述顶点与胞元的映射关系，重构得到蜂窝胞元图像；蜂窝质量检测模块，用于计算所述蜂窝胞元图像中所有胞元角的偏差平均值，并根据所述平均值评价所述蜂窝质量。
- 根据权利要求11所述的检测装置，其特征在于，还包括：置物台、龙门架；所述置物台用于盛放待检测蜂窝，其上设置有水平度示值板；所述龙门架设置在地面上，且所述龙门架设置有所述图像获取模块；所述图像获取模块视蜂窝产品的宽度由一台或多台高速相机组成，使得该图像获取模块可获得蜂窝产品宽度上的完整图像；所述龙门架高度可调，以保证对于不同高度的蜂窝产品，相机与蜂窝产品上端面的距离一致。
- 一种储存介质，其特征在于，所述存储介质上存储有计算机程序，所述程序被处理器执行时实现权利要求1-10中任意一项所述方法的步骤。
- 一种电子设备，其特征在于，包括存储器、显示器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执 行所述程序时实现权利要求1-10中任意一项所述方法的步骤。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010388965.8 | 2020-05-09 | ||
| CN202010388965.8A CN111524134B (zh) | 2020-05-09 | 2020-05-09 | 生产线上蜂窝产品规整度的检测方法及检测装置 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2021227287A1 true WO2021227287A1 (zh) | 2021-11-18 |
Family
ID=71907320
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2020/109729 WO2021227287A1 (zh) | 2020-05-09 | 2020-08-18 | 生产线上蜂窝产品规整度的检测方法及检测装置 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN111524134B (zh) |
| WO (1) | WO2021227287A1 (zh) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116448174A (zh) * | 2023-04-11 | 2023-07-18 | 深圳市森辉智能自控技术有限公司 | 用于洁净生产车间环境在线监测系统 |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111524134B (zh) * | 2020-05-09 | 2021-07-20 | 中南大学 | 生产线上蜂窝产品规整度的检测方法及检测装置 |
| CN116233614B (zh) * | 2023-04-24 | 2023-07-18 | 钛玛科（北京）工业科技有限公司 | 一种工业相机采集的处理方法 |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109978903A (zh) * | 2019-03-13 | 2019-07-05 | 浙江大华技术股份有限公司 | 一种标识点识别方法、装置、电子设备及存储介质 |
| CN110232388A (zh) * | 2019-06-11 | 2019-09-13 | 大连理工大学 | 一种从蜂窝芯表面测量数据中识别蜂窝边的方法 |
| CN110473174A (zh) * | 2019-07-26 | 2019-11-19 | 东南大学 | 一种基于图像计算铅笔精确数目的方法 |
| CN111054782A (zh) * | 2019-12-31 | 2020-04-24 | 太原科技大学 | 一种宽厚板板形检测装置及方法 |
| CN111524134A (zh) * | 2020-05-09 | 2020-08-11 | 中南大学 | 生产线上蜂窝产品规整度的检测方法及检测装置 |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100533484C (zh) * | 2007-05-25 | 2009-08-26 | 同济大学 | 基于自适应平坦度将无序点云转换为三角网格的系统及方法 |
| DE112016001723T5 (de) * | 2015-10-06 | 2018-01-04 | Ngk Insulators, Ltd. | Verfahren zum Kontrollieren einer Keramikkörperoberfläche |
| CN106530481A (zh) * | 2016-11-02 | 2017-03-22 | 深圳怡化电脑股份有限公司 | 一种检测纸币传输异常的方法及装置 |
| DE112017005711T5 (de) * | 2016-11-14 | 2019-08-14 | Ngk Insulators, Ltd. | Keramikkörper-Defektuntersuchungsvorrichtung und Defektuntersuchungsverfahren |
| CN108364311B (zh) * | 2018-01-29 | 2020-08-25 | 深圳市亿图视觉自动化技术有限公司 | 一种金属部件自动定位方法及终端设备 |
| CN110569907B (zh) * | 2019-09-10 | 2022-03-04 | 网易（杭州）网络有限公司 | 拼接图形的识别方法及装置、计算机存储介质、电子设备 |
-
2020
- 2020-05-09 CN CN202010388965.8A patent/CN111524134B/zh active Active
- 2020-08-18 WO PCT/CN2020/109729 patent/WO2021227287A1/zh active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109978903A (zh) * | 2019-03-13 | 2019-07-05 | 浙江大华技术股份有限公司 | 一种标识点识别方法、装置、电子设备及存储介质 |
| CN110232388A (zh) * | 2019-06-11 | 2019-09-13 | 大连理工大学 | 一种从蜂窝芯表面测量数据中识别蜂窝边的方法 |
| CN110473174A (zh) * | 2019-07-26 | 2019-11-19 | 东南大学 | 一种基于图像计算铅笔精确数目的方法 |
| CN111054782A (zh) * | 2019-12-31 | 2020-04-24 | 太原科技大学 | 一种宽厚板板形检测装置及方法 |
| CN111524134A (zh) * | 2020-05-09 | 2020-08-11 | 中南大学 | 生产线上蜂窝产品规整度的检测方法及检测装置 |
Non-Patent Citations (1)
| Title |
|---|
| WANG ZHONGGANG; LI ZHENDONG; ZHOU WEI; HUI DAVID: "On the influence of structural defects for honeycomb structure", COMPOSITES PART B, ELSEVIER, AMSTERDAM, NL, vol. 142, 2 February 2018 (2018-02-02), AMSTERDAM, NL, pages 183 - 192, XP085378705, ISSN: 1359-8368, DOI: 10.1016/j.compositesb.2018.01.015 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116448174A (zh) * | 2023-04-11 | 2023-07-18 | 深圳市森辉智能自控技术有限公司 | 用于洁净生产车间环境在线监测系统 |
| CN116448174B (zh) * | 2023-04-11 | 2023-10-31 | 深圳市森辉智能自控技术有限公司 | 用于洁净生产车间环境在线监测系统 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111524134A (zh) | 2020-08-11 |
| CN111524134B (zh) | 2021-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021227287A1 (zh) | 生产线上蜂窝产品规整度的检测方法及检测装置 | |
| US11551341B2 (en) | Method and device for automatically drawing structural cracks and precisely measuring widths thereof | |
| CN105133471B (zh) | 一种基于线结构光路面检测系统的路面深度图像生产方法 | |
| CN102171554B (zh) | 用于分析基底的表面的装置 | |
| CN102032875B (zh) | 一种基于图像处理的电缆护套厚度测量方法 | |
| JP6620477B2 (ja) | コンクリートのひび割れ検出方法及び検出プログラム | |
| CN104101600B (zh) | 连铸坯断面微小裂纹检测方法 | |
| CN102297660B (zh) | 一种盾构隧道衬砌管片接缝张开宽度的测量方法及装置 | |
| CN106447684B (zh) | 工业ct图像中弱边缘尺寸测量方法 | |
| CN109900711A (zh) | 基于机器视觉的工件缺陷检测方法 | |
| CN109580630A (zh) | 一种机械零部件缺陷的视觉检测方法 | |
| CN102183525A (zh) | 基于ccd阵列摄像技术的桥梁表观状态自动检测装置及其方法 | |
| CN104458764B (zh) | 基于大景深条带图像投影的弯曲粗糙表面缺陷鉴别方法 | |
| CN112233248A (zh) | 基于三维点云的表面平整度的检测方法、系统及介质 | |
| CN110853018B (zh) | 一种基于计算机视觉的振动台疲劳裂纹在线检测系统及检测方法 | |
| CN104700395A (zh) | 一种构造物外观裂缝检测方法及系统 | |
| CN115100206B (zh) | 用于具有周期图案纺织物的印花缺陷识别方法 | |
| CN110189375A (zh) | 一种基于单目视觉测量的图像目标识别方法 | |
| CN112489042A (zh) | 基于超分辨重建的金属品印刷缺陷与表面损伤的检测方法 | |
| CN113781537A (zh) | 一种轨道弹条扣件缺陷识别方法、装置和计算机设备 | |
| CN106403818A (zh) | 多规格大型方矩形管在线尺寸参数检测用的系统及方法 | |
| Fu et al. | Research on image-based detection and recognition technologies for cracks on rail surface | |
| CN111583241B (zh) | 超大面积蜂窝产品规整度的移动式检测方法及装置 | |
| WO2021227286A1 (zh) | 超大面积蜂窝产品规整度的移动式检测方法及装置 | |
| CN111583242B (zh) | 蜂窝规整度检测的形态学顶点提取方法及检测装置 |
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: 20935777 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 20935777 Country of ref document: EP Kind code of ref document: A1 |