WO2020252879A1 - 基于超声波喷雾的手机屏缺陷检测系统 - Google Patents

基于超声波喷雾的手机屏缺陷检测系统 Download PDF

Info

Publication number
WO2020252879A1
WO2020252879A1 PCT/CN2019/101306 CN2019101306W WO2020252879A1 WO 2020252879 A1 WO2020252879 A1 WO 2020252879A1 CN 2019101306 W CN2019101306 W CN 2019101306W WO 2020252879 A1 WO2020252879 A1 WO 2020252879A1
Authority
WO
WIPO (PCT)
Prior art keywords
mobile phone
phone screen
image
section
system based
Prior art date
Application number
PCT/CN2019/101306
Other languages
English (en)
French (fr)
Inventor
张学强
戴军
张建伟
罗银兵
Original Assignee
罗博特科智能科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 罗博特科智能科技股份有限公司 filed Critical 罗博特科智能科技股份有限公司
Publication of WO2020252879A1 publication Critical patent/WO2020252879A1/zh

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
    • B07C5/02—Measures preceding sorting, e.g. arranging articles in a stream orientating
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
    • B07C5/34—Sorting according to other particular properties
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
    • B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
    • B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
    • B07C5/36—Sorting apparatus characterised by the means used for distribution
    • B07C5/361—Processing or control devices therefor, e.g. escort memory
    • B07C5/362—Separating or distributor mechanisms

Definitions

  • the invention relates to the field of mobile phone screen production, in particular to a mobile phone screen defect detection system based on ultrasonic spray.
  • the mobile phone screen is also called the display screen, which is used to display images and colors.
  • the screen size is calculated on the diagonal of the screen, usually in inches (inch), which refers to the length of the diagonal of the screen.
  • Screen material introduction As the color screen of mobile phones becomes more and more common, the material of mobile phone screens becomes more and more important.
  • the waterproof layer or oleophobic layer on the surface of the screen is also extremely critical, and the quality of the processing directly affects the experience of use.
  • the technical problem to be solved by the present invention is to provide a mobile phone screen defect detection system based on ultrasonic spray, which has low preparation difficulty, low cost, high precision, and greatly improves production efficiency.
  • the present invention provides a mobile phone screen defect detection system based on ultrasonic spray, which includes a loading section, a processing detection section, and a discharging section that are sequentially connected, and the processing detection section and the discharging section are both provided
  • a roller conveyor line the roller conveyor of the roller conveyor line is provided with a partition ring, and a spray chamber and a visual inspection room are sequentially arranged in the processing and detection section along the mobile phone screen transmission direction, and the spray chamber is along the mobile phone screen transmission direction
  • An ion wind assembly and a spray assembly are arranged in sequence.
  • a camera and a light source are arranged in the visual inspection room. The light source and the camera cooperate to shoot the light source projection image on the mobile phone screen.
  • a shaft manipulator, the multi-axis manipulator transfers the mobile phone screen to the processing and detection section or transfers the mobile phone screen to the defective section, and the defective section is arranged on one side of the discharge section.
  • its processing method for processing the detection section includes the following steps:
  • Step 1) The surface of the mobile phone screen is cleaned and electrostatically removed through the ion wind assembly
  • Step 2) Spray water mist on the surface of the mobile phone screen through the spray component, and stay on the surface of the mobile phone screen;
  • Step 3 Use a camera and cooperate with a light source for image acquisition
  • Step 4) Perform preprocessing on the image collected by the camera, first perform the segmentation operation on the image, and reserve the region of interest to obtain the detection area;
  • Step 5 Set the threshold for the gray value of the image, and then filter the pixels in the detection area, and filter the domains that need to be processed through the set threshold;
  • the threshold selects pixels whose gray values meet the following conditions from the input image:
  • All feature points of the image that meet the conditions are returned as a region; when there are intervals with multiple gray values, a separate region is returned for each interval;
  • the length of the vector diagram is used as the basis for judgment to obtain the domains that need to be filtered
  • Step 6 Use Gaussian algorithm to smooth the image in the domain to be processed.
  • the smoothing effect increases with the increase of the filter operator, and the following filters are used:
  • the smoothed image is judged and classified, and the current mobile phone screen is screened from the classification result whether the mobile phone screen is qualified; when the mobile phone screen is unqualified, a multi-axis manipulator is used to remove the unqualified product to complete the screening.
  • the width and height are adjusted to satisfy all image segmentation and the size of each rectangle is equal; the segmentation is performed when the size of the area is at least 1.5 times the size of the rectangle given by the parameter.
  • the defect is used as a test sample, after collecting a large amount of information, it is stored in the defect sample library, and the defect detection is performed through different sample deep learning algorithms and the Gaussian algorithm is adjusted.
  • a single mobile phone screen captures three images.
  • the mobile phone screen enters the collection area as a whole, the first image is collected, and the second image is collected when the mobile phone screen is located in the center of the collection area.
  • the third image is acquired.
  • the multi-axis manipulator is provided with a grabbing sucker.
  • a humidity detector and an exhaust fan are arranged in the spray chamber, and when the humidity reaches a standard value, the exhaust fan will draw away excess water mist.
  • the air pressure in the spray chamber is lower than the visual inspection chamber.
  • a light source to project on a mobile phone screen with a mirror or reflective surface and cooperate with the camera to collect the projection vision, it can quickly and accurately obtain the image of the defect, so as to meet the judgment demand, improve the degree of automation, and increase the production capacity.
  • Figure 1 is a schematic diagram of the overall structure of the present invention.
  • FIG. 2 is a schematic diagram of the structure of the detection part of the present invention.
  • FIG. 3 is a partial schematic diagram of the roller table of the present invention.
  • Figure 4 is a detection flow chart of the present invention
  • Figure 5 is a schematic diagram of the present invention when it is divided
  • FIG. 6 is a schematic diagram of the present invention after selecting and screening part of the images in FIG. 5;
  • FIG. 7 is a schematic diagram of FIG. 6 optimized by the present invention.
  • Figure 5 is a schematic diagram of multiple photographs of the present invention.
  • an embodiment of the mobile phone screen defect detection system based on ultrasonic spray of the present invention includes a loading section 1, a processing detection section 2 and a discharging section 3 connected in sequence, a processing detection section and a discharge section.
  • a roller conveyor line 4 is provided in the material section, and a partition ring 5 is provided on the roller conveyor of the roller conveyor line.
  • a spray chamber 6 and a visual inspection chamber 7 are arranged in the processing and detection section along the transmission direction of the mobile phone screen.
  • the spray chamber is along the mobile phone Ion wind assembly 8 and spray assembly 9 are arranged in the transmission direction of the screen in turn.
  • a camera 10 and a light source 11 are arranged in the visual inspection room. The light source and the camera cooperate to shoot the light source projection image on the mobile phone screen.
  • the loading section and the discharging section are both set Multi-axis manipulator 12, the multi-axis manipulator transfers the mobile phone screen to the processing and detection section or transfers the mobile phone screen to the defective section 13, which is set on one side of the discharge section.
  • the mobile phone screen When in use, the mobile phone screen is installed in the turnover tray in advance, and the multi-axis manipulator in the feeding section grabs the mobile phone screen from the tray to the roller conveyor line.
  • the separation ring can be used to place multiple mobile phone screens in an orderly manner.
  • the roller conveyor line is limited in position, which has a high positioning effect when collecting images, which facilitates the subsequent accurate image segmentation and reduces the difficulty of image correction caused by the skew of the mobile phone screen;
  • the bottom of the mobile phone screen can also be installed with raised parts and light sources Avoid being blocked by the separation ring; of course, it can also be a stainless steel mesh belt conveyor line, and the separation ring supports the stainless steel mesh;
  • the mobile phone screen enters the spraying room.
  • the mobile phone screen In the spraying room, the mobile phone screen is blown with ion wind to remove static electricity, and then water mist is sprayed.
  • the water mist is generated by an ultrasonic humidifier, and a thin layer of water mist (water temperature) is continuously sprayed on the mobile phone screen through the ultrasonic nozzle. adjustable).
  • the mobile phone screen enters the visual inspection room to detect the state of the water mist on the mobile phone screen, and records the NG materials.
  • the mobile phone screen enters the discharge section, the good products continue to be transmitted, and the defective products can be captured into the bad section.
  • the air pressure in the visual inspection room is greater than the air pressure in the spray chamber to ensure that the lens will not be affected by water vapor detection.
  • the present invention also provides a judgment processing method based on the above mechanism, which mainly uses projection detection to image a screen with a water film. And processed by the later algorithm. Due to the existence of the waterproof layer, the water droplets will not condense together under the action of the waterproof layer, but will be scattered into small droplets. The place where there is no waterproof layer is the defect, and the water droplets will condense together to form a larger part of the water stain. In the image, the water mist part and the water stain part have a large gray-scale contrast, which can be clearly stuck by the algorithm. At the same time, the shape of defects left by other processing on the mobile phone screen is also diverse and irregular, and this method can also be used to judge poorly.
  • the surface of the mobile phone screen is cleaned and electrostatically removed through the ion wind component, and impurities such as dust particles are taken out;
  • the image collected by the camera is then preprocessed, and the image is segmented first, and the region of interest is retained to obtain the detection area;
  • the region is represented by a rectangle.
  • the quantized size of the rectangle is determined by its defined pixel coordinates and size; in the case of a large rectangular area, if the entire area is processed, Consumes a lot of memory, and the processing time is slow.
  • the usual processing method is to divide the input area into rectangular areas with width times height.
  • the width and height are adjusted to satisfy all image segmentation and the size of each rectangle is equal; the segmentation is performed when the size of the area is at least 1.5 times the size of the rectangle given by the parameter.
  • threshold setting is performed on the gray value of the image, and then the detection area is filtered, and the domains that need to be processed are filtered through the set threshold;
  • the threshold selects pixels whose gray values meet the following conditions from the input image:
  • All feature points of the image that meet the conditions are returned as a region; when there are intervals with multiple gray values, a separate region is returned for each interval;
  • the length of the vector diagram is used as the basis for judgment to obtain the domains that need to be filtered
  • the Gaussian algorithm is used to smooth the image in the domain to be processed.
  • the smoothing effect increases with the increase of the filter operator, and the following filters are used (the sigma value of the Gaussian function is indicated by brackets):
  • the smoothed image is judged and classified, and the current mobile phone screen is screened from the classification result whether the mobile phone screen is qualified; when the mobile phone screen is unqualified, a multi-axis manipulator is used to remove the unqualified product to complete the screening.
  • Gaussian filtering For boundary processing, the gray value of the image has a partial impact on the image boundary. Therefore, contrary to Gaussian image algorithm processing, the relationship between the filter mask size and the corresponding value of the sigma parameter is linear.
  • Gaussian filters are executed on OpenCL devices for all supported image types. Gaussian operators under different optimization levels will have certain optimizations in processing time and processing results, so it is an extremely important processing method in image processing. After Gaussian filtering and smoothing, the effect is obvious.
  • the defect is used as the detection sample, after collecting a large amount of information, it is stored in the defect sample library, and the defect detection is performed through different sample deep learning algorithms and the Gaussian algorithm is adjusted to achieve better detection accuracy.
  • a single mobile phone screen captures three images.
  • the first image a is captured
  • the second image b is captured when the mobile phone screen is in the center of the capture area.
  • the third image c is acquired.
  • the first image is mainly used for image acquisition and judgment of the side at the beginning of the movement direction
  • the third image is mainly used for image acquisition and judgment of the side at the end of the movement direction, suitable for 2.5D and 3D mobile phone screen detection , Improve detection accuracy and scope of application.

Landscapes

  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)

Abstract

基于超声波喷雾的手机屏缺陷检测系统，包括依次连接的上料段（1）、处理检测段（2）和排料段（3），处理检测段（2）和排料段（3）内均设置有辊道传送线（4），辊道传送线（4）的辊道上设置有分隔环（5），处理检测段（2）内沿手机屏传送方向依次设置喷雾室（6）和视觉检测室（7），喷雾室（6）内沿手机屏传送方向依次设置离子风组件（8）和喷雾组件（9），视觉检测室（7）内设置有相机（10）和光源（11），光源（11）和相机（10）配合拍摄手机屏上的光源投影图像，上料段（1）和排料段（3）内均设置有多轴机械手（12）。

Description

基于超声波喷雾的手机屏缺陷检测系统 技术领域
本发明涉及手机屏幕生产领域，具体涉及一种基于超声波喷雾的手机屏缺陷检测系统。
背景技术
手机屏幕也称显示屏，用于显示图像及色彩。荧幕尺寸依荧幕对角线计算，通常以英寸(inch)作单位，指荧幕对角的长度。屏幕材质引随着手机彩屏的逐渐普遍，手机屏幕的材质也越来越显得重要。
并且手机屏幕在使用时，对于屏幕表面的防水层或者疏油层也是极为关键的，加工的好坏直接影响使用体验感。
现有的检测一般采用人工判断较多，也有部分采用光学反射、折射等方式进行判断，总体检测结构设计复杂，制备难度大，并且成本高。
发明内容
本发明要解决的技术问题是提供一种基于超声波喷雾的手机屏缺陷检测系统，制备难度小，成本低，精度高，大大提高生产效率。
为了解决上述技术问题，本发明提供了一种基于超声波喷雾的手机屏缺陷检测系统，包括依次连接的上料段、处理检测段和排料段，所述处理检测段和排料段内均设置有辊道传送线，所述辊道传送线的辊道上设置有分隔环，所述处理检测段内沿手机屏传送方向依次设置有喷雾室和视觉检测室，所述喷雾室内沿手机屏传送方向依次设置有离子风组件和喷雾组件，所述视觉检测室内设 置有相机和光源，所述光源和相机配合拍摄手机屏上的光源投影图像，所述上料段和排料段内均设置有多轴机械手，所述多轴机械手将手机屏转载至处理检测段内或者将手机屏转载至不良段内，所述不良段设置在排料段一侧。
进一步的，其处理检测段的处理方法包括以下步骤：
步骤1)将手机屏表面通过离子风组件进行清洁和除静电；
步骤2)通过喷雾组件在手机屏表面喷水雾，并在手机屏表面滞留；
步骤3)采用相机并配合光源进行图像采集；
步骤4)对相机采集的图像进行预处理，先对图像进行分割操作，对感兴趣的区域进行保留，得到检测区；
步骤5)对图像的灰度值进行阈值设定，然后对检测区进行像素的筛选，通过设定的阈值，筛选需要处理的域；
其中阈值从输入图像中选择灰度值满足以下条件的像素:
MinGray≤G≤MaxGray；
满足条件的图像的所有特征点都作为一个区域返回；当具有多个灰度值的间隔，则为每个间隔返回一个单独的区域；
当区域图像为矢量图时，以矢量图的长度作为判断依据，得到筛选需要处理的域；
步骤6)采用高斯算法对需要处理的域内的图像进行平滑处理，平滑效果随着滤波算子的增加而增加，并使用以下过滤器：
Figure PCTCN2019101306-appb-000001
对平滑处理后的图像进行判断，并进行分类，从分类结果筛选当前手机屏是否合格；当手机屏不合格时，采用多轴机械手将不合格产品取走，完成筛选。
进一步的，在分割图像时，将图像剪切一个或多个区域，区域由矩形表示；
矩形分割时，通过调整宽度和高度满足全部图像分割并且每个矩形大小相等；当区域的大小至少是参数给定的矩形大小的1.5倍时才进行分割。
进一步的，将缺陷作为检测样品，在采集大量信息后，存入缺陷样品库中，并通过不同的样品深度学习算法进行缺陷检测并调整高斯算法。
进一步的，在采集图像时，单个手机屏采集三次图像，当手机屏整体进入采集区时采集第一次图像，当手机屏位于采集区中心时采集第二次图像，当手机屏与采集区边界接触并移出采集区时，采集第三次图像。
进一步的，所述多轴机械手上设置有抓取吸盘。
进一步的，所述喷雾室内设置有湿度检测仪和抽风机，当湿度达到标准值时，抽风机将多余水雾抽走。
进一步的，所述喷雾室内气压小于视觉检测室。
本发明的有益效果：
采用光源在具有镜面或反光面手机屏上进行投射并与相机配合采集投影视觉，能够快速且精准的得到缺陷处的图像，从而满足判断需求，提高自动化程度，提高生产能力。
附图说明
图1是本发明的整体结构示意图；
图2是本发明的检测部分结构示意图；
图3是本发明辊道部分示意图；
图4是本发明检测流程图；
图5是本发明分割时的示意图；
图6是本发明对图5中部分图像选取并筛选后的示意图；
图7是本发明对图6优化后的示意图；
图5是本发明多次拍照的示意图。
具体实施方式
下面结合附图和具体实施例对本发明作进一步说明，以使本领域的技术人员可以更好地理解本发明并能予以实施，但所举实施例不作为对本发明的限定。
参照图1至图3所示，本发明的基于超声波喷雾的手机屏缺陷检测系统的一实施例，包括依次连接的上料段1、处理检测段2和排料段3，处理检测段和排料段内均设置有辊道传送线4，辊道传送线的辊道上设置有分隔环5，处理检测段内沿手机屏传送方向依次设置有喷雾室6和视觉检测室7，喷雾室内沿手机屏传送方向依次设置有离子风组件8和喷雾组件9，视觉检测室内设置有相机10和光源11，光源和相机配合拍摄手机屏上的光源投影图像，上料段和排料段内均设置有多轴机械手12，多轴机械手将手机屏转载至处理检测段内或者将手机屏转载至不良段13内，不良段设置在排料段一侧。
使用时，手机屏事先装在周转盘中，上料段的多轴机械手将手机屏从料盘中抓取到辊道传送线上，采用分隔环能够将多个手机屏有序分开摆放在辊道传送线上，得到限位，在采集图像时具有较高的定位效果，便于后续精准的分割图像，降低手机屏歪斜导致的图像矫正难度；在手机屏底部还可以安装垫高部件，光源的照射避免被分隔环阻挡；当然也可以为不锈钢网带输送线，分隔环将不锈钢网支撑；
手机屏进入喷雾室，喷雾室内先将手机屏吹离子风除静电，而后喷水雾， 水雾由超声波加湿器产生，经超声波喷头持续在手机屏幕上喷上薄薄的一层水雾(水温可调节)。随后手机屏幕进入视觉检测室内检测水雾在手机屏幕上的状态，并记录下NG物料，当手机屏进入排料段后，良品继续传送，不良品被抓取至不良段内即可。其中，视觉检测室内气压大于喷雾室内气压，确保镜头不会有水汽影响检测，每个工作室内都有湿度检测装置，喷雾室上方有排雾管，排雾管与抽风机连接，当湿度达到标准值时，抽风机会持续将多余水雾抽走，使喷雾室内湿度恒定在一小区间内；多轴机械手上设置有抓取吸盘14，不伤手机屏。
参照图4所示，本发明还提供一种基于上述机构的判断处理方法，其主要通过投影式检测，对带有水膜的屏幕进行成像。并由后期算法进行处理。由于防水层的存在，水滴在防水层的作用下不会凝结在一起，而是分散成一颗颗的小水滴存在。而没有防水层存在的地方即为缺陷处，水滴将会凝结在一起，形成较大部分的水渍。在图像中，水雾部分与水渍部分具有较大的灰度对比，可以很明显的被算法卡出。同时手机屏上的其他加工留下的缺陷形状也具有多样性和不规则性，也能够采用本方法进行判断不良。
具体的，包括以下步骤：
将手机屏表面通过离子风组件进行清洁和除静电，取出灰尘颗粒等杂质；
然后通过喷雾组件在手机屏表面喷水雾，形成颗粒状附着在手机屏幕上；
接着采用相机并配合光源进行图像采集；
残渣头5所示，随后对相机采集的图像进行预处理，先对图像进行分割操作，对感兴趣的区域进行保留，得到检测区；
在分割图像时，将图像剪切一个或多个区域，区域由矩形表示，矩形的量化大小由其定义的像素坐标和大小来决定；在矩形区域很大的情况下，如果对 整个区域处理会消耗极大的内存，处理时间也会较慢，通常处理方式是将输入区域划分为具有宽度乘以高度的矩形区域。
矩形分割时，通过调整宽度和高度满足全部图像分割并且每个矩形大小相等；当区域的大小至少是参数给定的矩形大小的1.5倍时才进行分割。
参照图6所示，对图像的灰度值进行阈值设定，然后对检测区进行像素的筛选，通过设定的阈值，筛选需要处理的域；
其中阈值从输入图像中选择灰度值满足以下条件的像素:
MinGray≤G≤MaxGray；
满足条件的图像的所有特征点都作为一个区域返回；当具有多个灰度值的间隔，则为每个间隔返回一个单独的区域；
当区域图像为矢量图时，以矢量图的长度作为判断依据，得到筛选需要处理的域；
参照图7所示，采用高斯算法对需要处理的域内的图像进行平滑处理，平滑效果随着滤波算子的增加而增加，并使用以下过滤器(高斯函数的西格玛值用括号表示)：
Figure PCTCN2019101306-appb-000002
对平滑处理后的图像进行判断，并进行分类，从分类结果筛选当前手机屏是否合格；当手机屏不合格时，采用多轴机械手将不合格产品取走，完成筛选。
对于边界的处理，图像的灰度值在图像边界处有部分的影响，因此与高斯图像算法处理相反，过滤器掩码大小与西格玛参数的相应值之间的关系是线性 的。高斯滤波器在OpenCL设备上针对所有支持的图像类型执行。不同的优化程度下的高斯算子，会在处理时间及处理结果上有一定的优化，因此在图像处理上极其重要的处理方式。在高斯滤波平滑处理后，效果显而易见。
为了提高检测优化效果，将缺陷作为检测样品，在采集大量信息后，存入缺陷样品库中，并通过不同的样品深度学习算法进行缺陷检测并调整高斯算法，以达到更好地检测精度。
参照图8所示，在采集图像时，单个手机屏采集三次图像，当手机屏整体进入采集区时采集第一次图像a，当手机屏位于采集区中心时采集第二次图像b，当手机屏与采集区边界接触并移出采集区时，采集第三次图像c。其中第一次图像主要用于对移动方向首端的侧边进行图像采集并判断，第三次图像主要用于对移动方向尾端的侧边进行图像采集并判断，适合2.5D以及3D手机屏的检测，提高检测精度以及适用范围。
以上实施例仅是为充分说明本发明而所举的较佳的实施例，本发明的保护范围不限于此。本技术领域的技术人员在本发明基础上所作的等同替代或变换，均在本发明的保护范围之内。本发明的保护范围以权利要求书为准。

Claims (8)

  1. 一种基于超声波喷雾的手机屏缺陷检测系统，其特征在于，包括依次连接的上料段、处理检测段和排料段，所述处理检测段和排料段内均设置有辊道传送线，所述辊道传送线的辊道上设置有分隔环，所述处理检测段内沿手机屏传送方向依次设置有喷雾室和视觉检测室，所述喷雾室内沿手机屏传送方向依次设置有离子风组件和喷雾组件，所述视觉检测室内设置有相机和光源，所述光源和相机配合拍摄手机屏上的光源投影图像，所述上料段和排料段内均设置有多轴机械手，所述多轴机械手将手机屏转载至处理检测段内或者将手机屏转载至不良段内，所述不良段设置在排料段一侧。
  2. 如权利要求1所述的基于超声波喷雾的手机屏缺陷检测系统，其特征在于，其处理检测段的处理方法包括以下步骤：
    步骤1)将手机屏表面通过离子风组件进行清洁和除静电；
    步骤2)通过喷雾组件在手机屏表面喷水雾，并在手机屏表面滞留；
    步骤3)采用相机并配合光源进行图像采集；
    步骤4)对相机采集的图像进行预处理，先对图像进行分割操作，对感兴趣的区域进行保留，得到检测区；
    步骤5)对图像的灰度值进行阈值设定，然后对检测区进行像素的筛选，通过设定的阈值，筛选需要处理的域；
    其中阈值从输入图像中选择灰度值满足以下条件的像素:
    MinGray≤G≤MaxGray；
    满足条件的图像的所有特征点都作为一个区域返回；当具有多个灰度值的间隔，则为每个间隔返回一个单独的区域；
    当区域图像为矢量图时，以矢量图的长度作为判断依据，得到筛选需要处理的域；
    步骤6)采用高斯算法对需要处理的域内的图像进行平滑处理，平滑效果随着滤波算子的增加而增加，并使用以下过滤器：
    Figure PCTCN2019101306-appb-100001
    对平滑处理后的图像进行判断，并进行分类，从分类结果筛选当前手机屏是否合格；当手机屏不合格时，采用多轴机械手将不合格产品取走，完成筛选。
  3. 如权利要求2所述的基于超声波喷雾的手机屏缺陷检测系统，其特征在于，在分割图像时，将图像剪切一个或多个区域，区域由矩形表示；
    矩形分割时，通过调整宽度和高度满足全部图像分割并且每个矩形大小相等；当区域的大小至少是参数给定的矩形大小的1.5倍时才进行分割。
  4. 如权利要求2所述的基于超声波喷雾的手机屏缺陷检测系统，其特征在于，将缺陷作为检测样品，在采集大量信息后，存入缺陷样品库中，并通过不同的样品深度学习算法进行缺陷检测并调整高斯算法。
  5. 如权利要求2所述的基于超声波喷雾的手机屏缺陷检测系统，其特征在于，在采集图像时，单个手机屏采集三次图像，当手机屏整体进入采集区时采集第一次图像，当手机屏位于采集区中心时采集第二次图像，当手机屏与采集区边界接触并移出采集区时，采集第三次图像。
  6. 如权利要求1所述的基于超声波喷雾的手机屏缺陷检测系统，其特征在于，所述多轴机械手上设置有抓取吸盘。
  7. 如权利要求1所述的基于超声波喷雾的手机屏缺陷检测系统，其特征在 于，所述喷雾室内设置有湿度检测仪和抽风机，当湿度达到标准值时，抽风机将多余水雾抽走。
  8. 如权利要求1所述的基于超声波喷雾的手机屏缺陷检测系统，其特征在于，所述喷雾室内气压小于视觉检测室。
PCT/CN2019/101306 2019-06-19 2019-08-19 基于超声波喷雾的手机屏缺陷检测系统 WO2020252879A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910533341.8 2019-06-19
CN201910533341.8A CN110238080A (zh) 2019-06-19 2019-06-19 基于超声波喷雾的手机屏缺陷检测系统

Publications (1)

Publication Number Publication Date
WO2020252879A1 true WO2020252879A1 (zh) 2020-12-24

Family

ID=67888280

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/101306 WO2020252879A1 (zh) 2019-06-19 2019-08-19 基于超声波喷雾的手机屏缺陷检测系统

Country Status (2)

Country Link
CN (1) CN110238080A (zh)
WO (1) WO2020252879A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112945977B (zh) * 2021-02-09 2022-10-28 苏州天准科技股份有限公司 用于屏幕表面缺陷的检测设备

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110102643A1 (en) * 2004-02-04 2011-05-05 Tessera Technologies Ireland Limited Partial Face Detector Red-Eye Filter Method and Apparatus
CN106024661A (zh) * 2015-03-31 2016-10-12 朗姆研究公司 用于检测晶片上的缺陷的方法和装置
CN108918543A (zh) * 2018-05-14 2018-11-30 上海工程技术大学 一种镜子表面划伤的动态检测装置及方法

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1847831A (zh) * 2005-04-12 2006-10-18 比亚迪股份有限公司 一种检查平面质量缺陷的方法
CN108169236A (zh) * 2016-12-07 2018-06-15 广州映博智能科技有限公司 一种基于视觉的金属表面缺陷检测方法
CN207746189U (zh) * 2017-11-17 2018-08-21 东莞市晶博光电有限公司 一种具有电镀检测功能的手机玻璃清洗机
CN208513297U (zh) * 2018-05-17 2019-02-19 广东拓斯达科技股份有限公司 玻璃清洗分拣一体化系统
CN109030501A (zh) * 2018-07-31 2018-12-18 苏州天准科技股份有限公司 一种手机玻璃af膜表面瑕疵检测装置及检测方法
CN109490316B (zh) * 2018-11-30 2021-08-03 熵智科技(深圳)有限公司 一种基于机器视觉的表面缺陷检测算法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110102643A1 (en) * 2004-02-04 2011-05-05 Tessera Technologies Ireland Limited Partial Face Detector Red-Eye Filter Method and Apparatus
CN106024661A (zh) * 2015-03-31 2016-10-12 朗姆研究公司 用于检测晶片上的缺陷的方法和装置
CN108918543A (zh) * 2018-05-14 2018-11-30 上海工程技术大学 一种镜子表面划伤的动态检测装置及方法

Also Published As

Publication number Publication date
CN110238080A (zh) 2019-09-17

Similar Documents

Publication Publication Date Title
CN108230324B (zh) 磁瓦表面微缺陷视觉检测方法
WO2021248554A1 (zh) 高速高精度锂离子电池极片的毛刺检测方法及检测系统
JP2024069188A (ja) 空気サンプルの自動分析をするためのシステムおよび方法
KR100924985B1 (ko) 결함 검출 장치, 결함 검출 방법, 정보 처리 장치, 정보 처리 방법 및 그 프로그램
CN107064160B (zh) 基于显著性检测的纺织品表面瑕疵检测方法及系统
JP4102842B1 (ja) 欠陥検出装置、欠陥検出方法、情報処理装置、情報処理方法及びそのプログラム
JP2001527645A (ja) ムラ欠陥検出方法および検出装置
WO2007074770A1 (ja) 画像解析によって欠陥検査を行う欠陥検査装置
KR101297395B1 (ko) 이차 전지용 극판 비젼 검사 방법
JP2008139201A (ja) 欠陥検出装置、欠陥検出方法、情報処理装置、情報処理方法及びそのプログラム
CN105866136A (zh) 摄像头模组滤光片涂胶检测装置和方法
CN205229061U (zh) 一种基于线扫描相机的lcd导光板缺陷检测系统
CN112816490A (zh) 一种电子产品的自动检测设备及自动检测方法
WO2012006221A1 (en) Scratch detection method and apparatus
CN111260617A (zh) 一种基于深度学习的太阳能电池板缺陷检测方法
CN107767385A (zh) 一种基于机器视觉的金刚砂线颗粒计数方法和装置
CN111476712B (zh) 一种烧结机的台车箅条图像的摄取及检测方法、系统
WO2020252879A1 (zh) 基于超声波喷雾的手机屏缺陷检测系统
CN111307812A (zh) 基于机器视觉的焊点外观检测方法
JP2006242759A (ja) 周期性パターンのムラ検査方法
CN106093052A (zh) 一种断纱检测方法
CN112730420A (zh) 外观瑕疵检测工站
CN116997927A (zh) 曲面基板气泡检测方法及检测系统
CN216560298U (zh) 一种电子产品的自动检测设备
CN112837285A (zh) 一种板面图像的边缘检测方法及装置

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: 19933306

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: 19933306

Country of ref document: EP

Kind code of ref document: A1