WO2020119246A1 - 一种基于反射光的高光物体表面缺陷检测系统及方法 - Google Patents
一种基于反射光的高光物体表面缺陷检测系统及方法 Download PDFInfo
- Publication number
- WO2020119246A1 WO2020119246A1 PCT/CN2019/111534 CN2019111534W WO2020119246A1 WO 2020119246 A1 WO2020119246 A1 WO 2020119246A1 CN 2019111534 W CN2019111534 W CN 2019111534W WO 2020119246 A1 WO2020119246 A1 WO 2020119246A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- reflected light
- image
- workpiece
- straight line
- threshold
- 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
- G06T7/0008—Industrial image inspection checking presence/absence
-
- 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
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
Definitions
- the invention relates to the field of digital image processing, in particular to a system and method for detecting surface defects of highlight objects based on reflected light.
- Machine vision technology is a process that uses an image acquisition system to convert an image into a digital signal, and then uses a host computer to analyze the digital relationship between the digital images, and finally extracts and analyzes image features to achieve human goals.
- the required object characteristics are covered to a certain extent.
- the more general method is to use polarized lenses and other filters to add high-efficiency light sources to eliminate the highlight area for detection, but eliminating the highlight area is easy to cause the weakening of defect characteristics. It is difficult to detect some subtle defects, which has become a constraint for some industries. An important bottleneck for automated inspection of machine vision equipment.
- the present invention provides a system and method for detecting surface defects of highlight objects based on reflected light.
- a high-gloss object surface defect detection system based on reflected light characterized in that it includes a controller, a reflected light receiving board, a linear light source and a camera, and transports the workpiece to be tested to the detection position, so that the light emitted from the linear light source directly hits the workpiece to be tested At this time, the reflected light receiving plate receives the reflected light from the surface of the workpiece to be measured.
- the reflected light receiving plate adjusts the workpiece to be measured within the collection field of view of the camera so that the light emitted by the linear light source can traverse the entire area to be detected.
- the camera collection includes The image sequence reflecting the light is input to the controller.
- a transmission device which includes a belt, a transmission gear, and a DC motor, and the DC motor is respectively connected to the transmission gear and the controller.
- a clamping device which includes a clamp that clamps and rotates the workpiece to be measured and a rotary stepper motor that controls the rotation of the clamp.
- a light blocking plate which is arranged directly in front of the reflected light receiving plate.
- the reflected light receiving plate is opaque.
- the light blocking plate has a hole.
- a method for detecting a surface defect of a highlight object based on reflected light includes the following:
- S1 reads the image containing the reflected light, and compares the weighted gray value with the defined threshold to find an effective image
- S2 filters the gray image, selects the upper threshold and the lower threshold according to the pixel characteristics of the picture, removes the gray areas below the lower threshold and above the upper threshold, and retains the gray of the light scattering caused by the edges and defects of the workpiece to be measured Degree area
- S3 performs binarization processing, and performs local mean filtering on the upper and lower edges of the binarized image
- S4 calculates the sum of the number of pixels in each column of non-zero values for the filtered image, selects a threshold, and determines the location of surface defects;
- S5 calculates the degree of distortion of the edge of the straight line in the image after the average filtering, finds out the deformation area of the straight line, and determines whether there is a deformation area on the surface of the workpiece to be measured;
- S6 calculates all defect locations and marks them, and then displays the marked images.
- the mean filtering in S3 is,
- Edge(i) corresponds to the i-th column of the upper and lower edges
- n is a local filtering degree coefficient, and the larger the value, the stronger the filtering ability.
- the S5 calculates the degree of distortion of the edge of the straight line in the image after the mean filtering, finds out the deformation area of the straight line, and determines whether there is a deformation area on the surface of the workpiece to be measured;
- Pixels with a projection value of 0 in S5.1 are fitted with a straight line with a slope of line_k, where line_k is calculated as follows:
- left and right are the gray value of the first column that appears on the left and the right and the specific number of columns that are not zero;
- S5.2 Perform average filtering on the upper and lower edges, and then find each pole by subtraction between adjacent pixels and setting of judgment conditions, and record it;
- each line of the binarized image is weighted and summed, and the point with the largest value is found as the ordinate value of the reference straight line g(x).
- the straight line equation is as follows:
- S5.5 judge whether the pole belongs to the deformation point and the degree of deformation according to the threshold.
- the deformation point also includes a point with a projection value of 0 along the vertical direction of the image.
- the present invention has low requirements on the environment, light source, image acquisition system, etc., the hardware cost is very low, and the detection effect is good, and it has a large advantage in promotion in the industrial production environment;
- the present invention uses reflected light to detect the surface of a high-gloss object, and can enlarge the surface defect of the object by increasing the distance from the reflection point to the reflection receiving plane, and the detection result is more accurate;
- the present invention is also extensible. By increasing the frame rate and pixels of the camera and enhancing the stability of the linear light source, the method can provide more accurate and effective information for the detection and further improve the accuracy of the detection;
- the present invention can have multiple implementation forms, and can be implemented according to different hardware structures of the actual application environment and different applications of the workpiece to be tested;
- the present invention can convert regular curved surfaces into quasi-planes for processing, and can detect complex curved surfaces by adding a mask.
- the present invention has a wide application range, and only needs to adjust the upper and lower limits of the surface defect threshold, the upper and lower limits of the linear grayscale threshold and the upper and lower limits of the surface deformation defect threshold, etc., so that different objects can be detected, the adjustment of the parameters is simple, and it is suitable for embedding as a module Among different detection systems;
- Example 1 is a schematic structural view of Example 1 of the present invention.
- Embodiment 2 is a schematic structural diagram of Embodiment 2 of the present invention.
- FIG. 3 is a working flowchart of the present invention.
- a system for detecting surface defects of highlight objects based on reflected light includes a controller, a reflected light receiving board, a linear light source 2, a transmission device, a clamping device 4 and a camera 1.
- the camera is located on the reflection Below the light receiving plate, the reflected light receiving plate is within the field of view of the camera.
- the positions of the linear light source, the camera, and the reflected light receiving plate 6 in the present invention are not fixed, and can be set specifically according to the specific situation, as long as the output light 3 of the linear light source is directly emitted to the waiting position when the workpiece to be tested is transported to the detection position Measuring the surface of the workpiece, because the workpiece to be measured is a high-gloss object, the reflected light receiving plate receives the reflected light 5 from the surface of the workpiece to be measured, adjust the workpiece to be measured, so that the linear light can traverse the entire area to be inspected, and then use the high frame rate
- the camera acquires sequence images and inputs them to the controller for analysis to obtain surface defects.
- the clamping device includes a clamp, a non-slip rubber, a rotating stepper motor, and a pressure sensor.
- the workpiece to be tested enters the defect detection device and is placed in the clamp.
- a pressure sensor is provided between the clamp and the workpiece to be tested. When the pressure sensor detects clamping When the pressure reaches the specified pressure, confirm that the fixture clamps the workpiece, and the fixture drive motor stops working to achieve the clamping process of the workpiece.
- the transmission equipment includes a belt, a transmission gear and a DC motor, the DC motor is respectively connected to the transmission gear and the controller, the controller sends a signal to the transmission equipment, the controller drives the rotating motor to let the fixture carry the test The workpiece rotates at a specified angle, and then the transmission equipment is used to move the workpiece to be measured, so that the linear light can traverse the entire surface of the object to be measured.
- the reflected light receiving plate is opaque. In this embodiment, it is composed of paper material.
- the reflected light receiving plate receives better effect under dark conditions. Therefore, a light blocking plate is provided in front of the reflected light receiving plate.
- the linear light source is a red light source.
- the reflection line can be changed into a line that is several times the surface defect of the workpiece to be measured by extending the reflection distance.
- the controller processes the image to obtain surface defects, which are displayed after being marked.
- the R channel of the image is separately extracted for analysis, and the first effective image is found by weighting the gray value and further finding all effective images greater than the specified threshold.
- S2 filters the effective image, selects the upper threshold and the lower threshold according to the pixel characteristics of the picture, removes the gray areas below the lower threshold and above the upper threshold, and only retains the middle and low gray areas, that is, the workpiece to be measured is retained Low-gray area of light scattering caused by edges and defects, which is convenient for subsequent defect analysis;
- S3 selects an appropriate threshold, performs binarization processing on the reflected light, and then performs local average filtering on the upper and lower edges of the binarized image to prevent the influence of noise on the detection result.
- the specific steps are:
- Edge(i) corresponds to the i-th column of the upper and lower edges
- n is the local filtering degree coefficient. The larger the value, the stronger the filtering ability, which can be selected by the user according to the actual situation;
- S4 Calculate the sum of the number of pixels in each column of the image that is not zero, and select a threshold based on this value to determine the location of surface defects;
- S5 calculates the degree of distortion of the edge of the straight line according to the formula, selects an appropriate threshold, finds the deformation area of the straight line, and thereby determines whether there is a deformation area on the surface of the workpiece to be measured;
- left and right are the gray value of the first column that appears on the left and the right and the specific number of columns that are not zero.
- S5.2 Perform average filtering on the upper and lower edges, and then find each pole by subtraction between adjacent pixels and setting of judgment conditions, and record it;
- each line of the binarized image is weighted and summed, and the point with the largest value is found as the ordinate value of the reference straight line g(x).
- the straight line equation is as follows:
- i is the value of the abscissa of each pole
- the meaning of using an exponential function to describe this deviation value is that when distance(i) is small, the impact on the degree of deformation is small, and the main difference is through the shrinkage value of shrink( i) Judgment. If the value is too large, it proves that the point deviates too far from the standard straight line, and the probability of the deformation point increases. At this time, the judgment is mainly made by the deviation value distance (i).
- a point with a projection value of 0 in the vertical direction is also regarded as a deformation point.
- S6 calculates all defect locations and marks them, and then sends the marked images to the controller for display.
- a system for detecting surface defects of specular objects based on reflected light includes a reflected light receiving plate, a linear light source and a camera.
- the linear light source is emitted at a small angle and directly hits the surface of the workpiece 9 to be measured.
- a hole is opened at a suitable position of the light blocking plate, and the reflected light is reflected through the hole 8 onto the opaque reflected light receiving plate at the rear, and a picture sequence with complete reflected light is collected by the camera and input to the controller for detection.
- This embodiment uses low-angle exit light for reflection. Therefore, as long as the surface curvature of the workpiece to be measured is continuous, the reflected light will be displayed in the form of an approximately straight line, which achieves the effect of turning the curve straight, which is more conducive to the subsequent detection.
- This embodiment usually acts on both ends of the industrial assembly line, and uses the transmission structure of the assembly line to realize automatic detection of the workpiece to be tested.
Abstract
本发明公开了一种基于反射光的高光物体表面缺陷检测系统及方法，包括控制器、反射光接收板、线性光源及摄像头，将待测工件运送至检测位，使线性光源的出射光直射待测工件的表面，此时反射光接收板接收待测工件表面的反射光线，反射光接收板在摄像头的采集视野内，调整待测工件，使线性光能够遍历整个待检测区域，摄像头采集包含反射光线的图像序列输入控制器。本发明利用反射光线对高光物体表面的检测，能够通过增加反射点到反射接收平面的距离实现对物体表面缺陷的放大，检测结果更为精确。
Description
本发明涉及数字图像处理领域，具体涉及一种基于反射光的高光物体表面缺陷检测系统及方法。
机器视觉技术是利用图像采集系统将图像转换成数字信号，再利用上位机通过分析数字图像之间的数字关系，最终通过提取、分析图像特征，实现人类目的的过程。但是高光物体由于其表面对光线有极强的反射作用，一定程度上遮盖了所需的物体特征。现在比较通用的方法是通过偏振镜片等滤镜加之以有效的光源消除高光区域进行检测，但是消除高光区域的同时易造成缺陷特征的弱化，对于一些细微的缺陷难以检测，这成为制约一些行业引入机器视觉设备进行自动化检测的重要瓶颈。
发明内容
为了克服现有技术存在的缺点与不足，本发明提供一种基于反射光的高光物体表面缺陷检测系统及方法。
本发明采用如下技术方案：
一种基于反射光的高光物体表面缺陷检测系统，其特征在于，包括控制器、反射光接收板、线性光源及摄像头，将待测工件运送至检测位，使线性光源的出射光直射待测工件的表面，此时反射光接收板接收待测工件表面的反射光线，反射光接收板在摄像头的采集视野内，调整待测工件，使线性光源的出射光能够遍历整个待检测区域，摄像头采集包含反射光线的图像序列输入控制器。
进一步的，还包括传动设备,所述传动设备包括皮带、传动齿轮及直流电机，所述直流电机分别与传动齿轮及控制器连接。
进一步的，还包括夹取设备，所述夹取设备包括夹取待测工件并进行旋转的夹具及控制夹具旋转的旋转步进电机。
进一步的，还包括挡光板，所述挡光板设置在反射光接收板的正前方。
进一步的，所述反射光接收板是不透明的。
进一步的，所述挡光板开有一个孔。
一种基于反射光的高光物体表面缺陷检测系统的方法，包括如下：
S1读取包含反射光线的图像，通过灰度值加权和与限定阈值相比较，找出有效图像；
S2对灰度图像进行滤波处理，根据图片像素特征选取上界阈值及下界阈值，剔除低于下界阈值及高于上界阈值的灰度区域，保留待测工件边缘及缺陷造成的光线散射的灰度区域；
S3进行二值化处理，对二值化图像进行上下边缘的局部均值滤波；
S4对滤波后的图像计算每一列不为零的像素点的个数和，选取阈值，判断表面缺陷存在的位置；
S5计算均值滤波后图像中直线边缘的扭曲程度，找出直线形变区域，判断出待测工件表面是否存在形变区域；
S6计算出所有缺陷位置，并标记出来，再将标记后的图像进行显示。
所述S3中均值滤波为，
S3.1找到图像的上下边缘，即每一列第一个以及最后一个不为零的像素点，并将其记录下来；
S3.2对图像的上下边缘进行局部均值滤波，其具体方法如下式所示：
其中，Edge(i)对应上下边缘的第i列，而n是局部滤波程度系数，该值越大滤波能力越强。
所述S5计算均值滤波后图像中直线边缘的扭曲程度，找出直线形变区域，判断出待测工件表面是否存在形变区域；
S5.1中投影值为0的像素点，用斜率为line_k的直线拟合，其中line_k的计算方法如下：
其中left和right为左边和右边出现的第一个列灰度值和不为零的具体列数；
用于拟合的直线方程f(x)如下：
f(x)＝line_k·x+edge(left)；
S5.2对上下两边缘进行均值滤波，再通过相邻像素点之间的减法以及判断条件的设置找出每个极点，并记录下来；
S5.3根据霍夫直线的基本原理，将二值化图像的每一行进行加权求和，寻找该值最大的点作为参考直线g(x)的纵坐标值，其中直线方程如下：
其中
是一组列向量，表示二值图像的每一行的加权和；
S5.4对每个极点进行形变程度的判断；
S5.5根据阈值判断极点是否属于形变点及形变的程度。
所述形变点还包括沿图像竖直方向的投影值为0的点。
本发明的有益效果：
(1)本发明对环境、光源、图像采集系统等的要求不高，硬件成本很低，且检测效果较好，在工业生产环境中进行推广具有较大优势；
(2)本发明利用反射光线对高光物体表面的检测，能够通过增加反射点到反射接收平面的距离实现对物体表面缺陷的放大，检测结果更为精确；
(3)本发明采用的算法均为简单的加权求和取值以及简单的逻辑运算，计算量低、运算速度块；
(4)本发明还具有可扩展性，通过提高摄像头的帧率及像素、增强线性光源稳定性等方法可以为检测提供更加精确有效的信息，进一步提高检测的准确性；
(5)本发明可具有多种实施形式，可根据实际应用环境与待测工件的不同应用不同的硬件结构进行实施；
(6)本发明能够将规则曲面转化为类平面进行处理，并且对复杂曲面能够通过增加mask(掩模)的方法进行检测。
(7)本发明应用范围广，只需要调节表面缺陷阈值上下限，直线灰度阈值上下限及表面形变缺陷阈值上下限等，便可实现不同物体的检测，参数的调节简单，适合作为模块嵌入不同的检测系统之中；
图1是本发明实施例1的结构示意图；
图2是本发明实施例2的结构示意图；
图3是本发明的工作流程图。
下面结合实施例及附图，对本发明作进一步地详细说明，但本发明的实施方式不限于此。
实施例
如图1所示，一种基于反射光的高光物体表面缺陷检测系统，包括控制器、反射光接收板、线性光源2、传动设备、夹取设备4及摄像头1，该实施例中摄像头位于反射光接收板的下方，反射光接收板在摄像头的视野范围内。本发明中的线性光源、摄像头、反射光接收板6的位置均不固定，可通过具体情况具体设置，只要满足当待测工件运送至检测位的时候，使线性光源的出射光3直射到待测工件的表面，由于待测工件为高光物体，此时反射光接收板接收待测工件表面的反射光线5，调整待测工件，使线性光能够遍历整个待检测区域，然后采用高帧率的摄像头获取序列图像，输入控制器进行分析得到表面缺陷。
所述夹取设备包括夹具、防滑胶、旋转步进电机以及压力传感器，待测工件进入缺陷检测设备后放入夹具中，夹具与待测工件之间设置压力传感器，当压力传感器检测到夹紧压力达到指定压力，确认夹具夹紧工件，则夹具传动电机停止工作，实现对工件的夹紧过程。
所述传动设备,所述传动设备包括皮带、传动齿轮及直流电机，所述直流电机分别与传动齿轮及控制器连接，控制器向传动设备发送信号，控制器驱动旋转电机让夹具带着待测工件旋转指定角度，再利用传动设备移动待测工件，使线性光线能够遍历整个待测物体表面。
所述反射光接收板是不透明的，本实施例中由纸质材料构成，反射光接收板在黑暗条件下接收效果比较好，因此在反射光接收板的正前方设置挡光板。
本实施例中线性光源为红色光源。
本发明可以通过延长反射距离，将反射线变为几倍于待测工件物体表面缺陷的线。
如图3所示，控制器接收到包含反射光线的序列图像后，对图像进行处理得到表面缺陷，进行标记后显示。
具体步骤为：
S1由于本实施例采用的红色的线性光源，因此单独提取出图像的R通道进行分析，通过灰度值加权和寻找出第一张有效图像，进一步找出大于指定阈值的所有有效图像。
S2对有效图像进行滤波处理，根据图片像素特征选取上界阈值及下界阈值，剔除低于下界阈值及高于上界阈值的灰度区域，只保留中低灰度区域部分，即保留待测工件边缘及缺陷造成的光线散射的低灰度区域，便于后续的缺陷分析；
S3选取适当阈值，对反射光线进行二值化处理，再对二值化图像的上下边 缘进行局部均值滤波，以防止噪声对检测结果的影响，具体步骤为：
S3.1找到图像的上下边缘，即每一列第一个以及最后一个不为零的像素点，并将其记录下来；
S3.2对图像的上下边缘进行局部均值滤波，其具体方法如下式所示：
其中，Edge(i)对应上下边缘的第i列，而n是局部滤波程度系数，该值越大滤波能力越强，可由使用者根据实际情况进行选用；
S4计算图像每一列不为零的像素点的个数和，并根据该值选取阈值，判断表面缺陷存在的位置；
S5根据公式计算出直线边缘的扭曲程度，选取合适阈值，找出直线形变区域，从而判断出待测工件表面是否存在形变区域；
S5.1对步骤S4中投影值为0的像素点，用斜率为line_k的直线拟合，其中line_k的计算方法如下：
其中left和right为左边和右边出现的第一个列灰度值和不为零的具体列数。
用于拟合的直线方程f(x)如下：
f(x)＝line_k·x+edge(left)
S5.2对上下两边缘进行均值滤波，再通过相邻像素点之间的减法以及判断条件的设置找出每个极点，并记录下来；
S5.3根据霍夫直线的基本原理，将二值化图像的每一行进行加权求和，寻找该值最大的点作为参考直线g(x)的纵坐标值，其中直线方程如下：
其中
是一组列向量，表示二值图像的每一行的加权和；
S5.4对每个极点进行形变程度的判断，可以由以下几个式子表述：
1)极点到参考直线的距离distance的计算方法如下：
distance(i)＝|Edge(i)-g(x)|
2)直线的形变程度shrink的计算方法如下：
3)结合上述两式，形变程度的最终判断方法可由下式表示：
distortion(i)＝1.2
|distance(i)|×0.001+shrink(i)
其中i为每个极点的横坐标的值，使用指数函数的形式去描述这个偏离值的意义在于当distance(i)较小的时候对形变程度的影响较小，此时主要通过落差值shrink(i)进行判断。该值如若过大，则证明该点偏离标准直线过远，是形变点的概率增大，此时主要通过偏离值distance(i)进行判断。
S5.5根据实际情况选取阈值，判断该极点是否属于形变点以及形变的严重程度；
S5.6将沿竖直方向投影值为0的点也算作是形变点。
S6计算出所有缺陷位置，并标记出来，再将标记后的图像传送至控制器进行显示。
实施例2
如图2所示，一种基于反射光的高光物体表面缺陷检测系统，其包括反射光接收板、线性光源和摄像头，该实施例中线性光源以小角度射出，直射至待测工件9表面，在挡光板合适的位置开有一个孔，反射光线通过孔8反射至后方不透明的反射光接收板之上，通过摄像头采集具有完整反射光线的图片序列并输入控制器进行检测。
本实施例采用低角度出射光进行反射，因此只要待测工件表面弧度连续，反射光线将以近似直线的形式显示，实现了化曲为直的效果，更加有利于后续检测的进行。
其它特征与实施例1相同。
本实施例通常作用于工业流水线两端，利用流水线的传动结构实现对待测工件的自动检测。
上述实施例为本发明较佳的实施方式，但本发明的实施方式并不受所述实施例的限制，其他的任何未背离本发明的精神实质与原理下所作的改变、修饰、替代、组合、简化，均应为等效的置换方式，都包含在本发明的保护范围之内。
Claims (10)
- 一种基于反射光的高光物体表面缺陷检测系统，其特征在于，包括控制器、反射光接收板、线性光源及摄像头，将待测工件运送至检测位，使线性光源的出射光直射待测工件的表面，此时反射光接收板接收待测工件表面的反射光线，反射光接收板在摄像头的采集视野内，调整待测工件，使线性光源的出射光能够遍历整个待检测区域，摄像头采集包含反射光线的图像序列输入控制器。
- 根据权利要求1所述的一种基于反射光的高光物体表面缺陷检测系统，其特征在于，还包括传动设备,所述传动设备包括皮带、传动齿轮及直流电机，所述直流电机分别与传动齿轮及控制器连接。
- 根据权利要求2所述的一种基于反射光的高光物体表面缺陷检测系统，其特征在于，还包括夹取设备，所述夹取设备包括夹取待测工件并进行旋转的夹具及控制夹具旋转的旋转步进电机。
- 根据权利要求1所述的一种基于反射光的高光物体表面缺陷检测系统，其特征在于，还包括挡光板，所述挡光板设置在反射光接收板的正前方。
- 根据权利要求1所述的一种基于反射光的高光物体表面缺陷检测系统，其特征在于，所述反射光接收板是不透明的。
- 根据权利要求4所述的一种基于反射光的高光物体表面缺陷检测系统，其特征在于，所述挡光板开有一个孔。
- 一种如权利要求1所述的一种基于反射光的高光物体表面缺陷检测系统的方法，其特征在于，包括如下：S1读取包含反射光线的图像，通过灰度值加权和与限定阈值相比较，找出有效图像；S2对灰度图像进行滤波处理，根据图片像素特征选取上界阈值及下界阈值，剔除低于下界阈值及高于上界阈值的灰度区域，保留待测工件边缘及缺陷造成的光线散射的灰度区域；S3进行二值化处理，对二值化图像进行上下边缘的局部均值滤波；S4对滤波后的图像计算每一列不为零的像素点的个数和，选取阈值，判断表面缺陷存在的位置；S5计算均值滤波后图像中直线边缘的扭曲程度，找出直线形变区域，判断出待测工件表面是否存在形变区域；S6计算出所有缺陷位置，并标记出来，再将标记后的图像进行显示。
- 根据权利要求7所述的方法，其特征在于，所述S3中均值滤波算法为，S3.1找到图像的上下边缘，即每一列第一个以及最后一个不为零的像素点，并将其记录下来；S3.2对图像的上下边缘进行局部均值滤波，其具体方法如下式所示：
其中，Edge(i)对应上下边缘的第i列，而n是局部滤波程度系数，该值越大滤波能力越强。 - 根据权利要求7所述的方法，其特征在于，所述S5计算均值滤波后图像中直线边缘的扭曲程度，找出直线形变区域，判断出待测工件表面是否存在形变区域；S5.1中投影值为0的像素点，用斜率为line_k的直线拟合，其中line_k的计算方法如下：
其中left和right为左边和右边出现的第一个列灰度值和不为零的具体列数；用于拟合的直线方程f(x)如下：f(x)＝line_k·x+edge(left)；S5.2对上下两边缘进行均值滤波，再通过相邻像素点之间的减法以及判断条件的设置找出每个极点，并记录下来；S5.3根据霍夫直线的基本原理，将二值化图像的每一行进行加权求和，寻找该值最大的点作为参考直线g(x)的纵坐标值，其中直线方程如下：
其中是一组列向量，表示二值图像的每一行的加权和；
S5.4对每个极点进行形变程度的判断；S5.5根据阈值判断极点是否属于形变点及形变的程度。 - 根据权利要求9所述的方法，其特征在于，形变点还包括沿图像竖直方向的投影值为0的点。
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811521895.8A CN109584239B (zh) | 2018-12-13 | 2018-12-13 | 一种基于反射光的高光物体表面缺陷检测系统及方法 |
| CN201811521895.8 | 2018-12-13 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2020119246A1 true WO2020119246A1 (zh) | 2020-06-18 |
Family
ID=65927968
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2019/111534 WO2020119246A1 (zh) | 2018-12-13 | 2019-10-16 | 一种基于反射光的高光物体表面缺陷检测系统及方法 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109584239B (zh) |
| WO (1) | WO2020119246A1 (zh) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109584239B (zh) * | 2018-12-13 | 2024-02-06 | 华南理工大学 | 一种基于反射光的高光物体表面缺陷检测系统及方法 |
| CN110941916A (zh) * | 2019-12-16 | 2020-03-31 | 重庆长安汽车股份有限公司 | 基于铸件仿真软件预测铝合金铸件显微缩孔的方法 |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101393143A (zh) * | 2008-10-31 | 2009-03-25 | 东莞康视达自动化科技有限公司 | 基于显微视觉的光学测量和检测方法及其装置 |
| KR20110073158A (ko) * | 2009-12-22 | 2011-06-29 | 연세대학교 산학협력단 | 국소 표면 플라즈몬 공명 기반의 초고해상도 전반사 형광 현미경 및 전반사 형광 현미경용 검출 모듈 |
| CN107618533A (zh) * | 2017-09-29 | 2018-01-23 | 兰州交通大学 | 一种钢轨表面离散缺陷的机器视觉检测装置及方法 |
| CN109584239A (zh) * | 2018-12-13 | 2019-04-05 | 华南理工大学 | 一种基于反射光的高光物体表面缺陷检测系统及方法 |
| CN209281475U (zh) * | 2018-12-13 | 2019-08-20 | 华南理工大学 | 一种基于反射光的高光物体表面缺陷检测系统 |
Family Cites Families (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5881002B2 (ja) * | 2011-10-11 | 2016-03-09 | 大和製罐株式会社 | 表面欠損検査装置およびその方法 |
| US9839946B2 (en) * | 2015-02-04 | 2017-12-12 | The Boeing Company | System and method for high speed FOD detection |
| CN105701434A (zh) * | 2015-12-30 | 2016-06-22 | 广州卓德信息科技有限公司 | 二维码扭曲图像的图像校正方法 |
| JP6625475B2 (ja) * | 2016-04-22 | 2019-12-25 | 株式会社 東京ウエルズ | 欠陥検査方法および欠陥検査システム |
| CN106442556A (zh) * | 2016-11-16 | 2017-02-22 | 哈尔滨理工大学 | 一种板状带孔工件表面缺陷检测装置和方法 |
| CN106600593B (zh) * | 2016-12-19 | 2019-08-09 | 福州大学 | 一种中铝瓷球表面缺陷检测方法 |
| CN206832698U (zh) * | 2017-05-17 | 2018-01-02 | 华南理工大学 | 一种用于表面光滑高反光物体的缺陷检测装置 |
| CN108072664B (zh) * | 2017-12-29 | 2020-12-08 | 苏州德创测控科技有限公司 | 一种外观检测系统及方法 |
| CN108445007B (zh) * | 2018-01-09 | 2020-11-17 | 深圳市华汉伟业科技有限公司 | 一种基于图像融合的检测方法及其检测装置 |
| CN108629775B (zh) * | 2018-05-14 | 2021-08-03 | 华中科技大学 | 一种热态高速线材表面图像处理方法 |
| CN108375585A (zh) * | 2018-05-18 | 2018-08-07 | 华南理工大学 | 一种多角度视觉缺陷检测平台及检测方法 |
-
2018
- 2018-12-13 CN CN201811521895.8A patent/CN109584239B/zh active Active
-
2019
- 2019-10-16 WO PCT/CN2019/111534 patent/WO2020119246A1/zh active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101393143A (zh) * | 2008-10-31 | 2009-03-25 | 东莞康视达自动化科技有限公司 | 基于显微视觉的光学测量和检测方法及其装置 |
| KR20110073158A (ko) * | 2009-12-22 | 2011-06-29 | 연세대학교 산학협력단 | 국소 표면 플라즈몬 공명 기반의 초고해상도 전반사 형광 현미경 및 전반사 형광 현미경용 검출 모듈 |
| CN107618533A (zh) * | 2017-09-29 | 2018-01-23 | 兰州交通大学 | 一种钢轨表面离散缺陷的机器视觉检测装置及方法 |
| CN109584239A (zh) * | 2018-12-13 | 2019-04-05 | 华南理工大学 | 一种基于反射光的高光物体表面缺陷检测系统及方法 |
| CN209281475U (zh) * | 2018-12-13 | 2019-08-20 | 华南理工大学 | 一种基于反射光的高光物体表面缺陷检测系统 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109584239A (zh) | 2019-04-05 |
| CN109584239B (zh) | 2024-02-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107203990B (zh) | 一种基于模板匹配与图像质量评估的标贴破损检测方法 | |
| CN105067638B (zh) | 基于机器视觉的轮胎胎膜表面字符缺陷检测方法 | |
| GB2569751A (en) | Static infrared thermal image processing-based underground pipe leakage detection method | |
| CN109900711A (zh) | 基于机器视觉的工件缺陷检测方法 | |
| CN115375686B (zh) | 一种基于图像处理的玻璃边缘瑕疵检测方法 | |
| WO2014139231A1 (zh) | 光源光强均一性测调系统及测调方法 | |
| WO2020119246A1 (zh) | 一种基于反射光的高光物体表面缺陷检测系统及方法 | |
| CN110189375A (zh) | 一种基于单目视觉测量的图像目标识别方法 | |
| CN105241400A (zh) | 一种玻璃平整度检测的方法及其装置 | |
| Fu et al. | Medicine glass bottle defect detection based on machine vision | |
| CN111398287A (zh) | 一种电池极片划痕检测系统及检测方法 | |
| CN115100200A (zh) | 基于光学手段的光纤缺陷检测方法及系统 | |
| CN113340909A (zh) | 一种基于机器视觉的胶线缺陷检测方法 | |
| CN112881424A (zh) | Ai+荧光渗透小型管件表面缺陷检测及质量分级方法与系统 | |
| CN114636706A (zh) | 一种太阳能电池片在镀膜后的图像检测的综合方法及装置 | |
| CN112014407A (zh) | 一种集成电路晶圆表面缺陷检测的方法 | |
| CN111693535A (zh) | 基于机器视觉分析的触摸屏缺陷检测设备及方法 | |
| TWI714923B (zh) | 自動光學檢測方法及應用其之自動光學檢測系統 | |
| CN105391998B (zh) | 微光夜视仪分辨率自动检测方法和装置 | |
| CN114359155A (zh) | 一种覆膜方法及系统 | |
| CN111738963B (zh) | 一种基于深度自学习的烟包包装外观检测方法 | |
| CN112484680A (zh) | 一种基于圆检测的蓝宝石晶片定位跟踪方法 | |
| CN112837285B (zh) | 一种板面图像的边缘检测方法及装置 | |
| CN209606340U (zh) | 透明药瓶底部积料缺陷检测系统 | |
| Zhang et al. | Defect detection of mobile phone screen based on improved difference image method |
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: 19895260 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 28.09.2021) |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 19895260 Country of ref document: EP Kind code of ref document: A1 |