CN108593653A - A kind of screw classification of view-based access control model and damage detection device and method - Google Patents

Abstract

The invention discloses a vision-based screw classification and damage detection device, which includes a detection station, a top camera, a side camera, a light source, and an industrial computer. Screws are set on the detection station, and the top camera is used to photograph the head of the screw. The side image, the side camera is used to detect the screw image of the screw, the light source is used to provide the required light for the shooting of the top camera and the side camera, and the top camera and the side camera transmit the image data taken to the industrial computer respectively The image recognition processing is carried out in the computer, and the industrial computer classifies and detects the damage of the screws through the image recognition processing. The advantages of the present invention are: the screws can be effectively dispersed through the loading and unloading subsystem, and the screws can be clamped by robots and industrial cameras, laying the foundation for subsequent classification and detection; the notch can be obtained by the boundary between the screw head and the screw rod Area, thread height and other information to judge the wear condition.

Description

A vision-based screw classification and damage detection device and method

technical field

The invention relates to the field of automation technology, in particular to a device and method for detecting the damage degree and screw parameters of screw heads and threaded parts, and realizing the classification of screws and the detection of damaged screws.

Background technique

Screws are the most commonly used small parts in industry. Usually, screws are widely used in industrial sites, and there are many types, models and materials. It is of great significance to recycle and reuse some screws that are used in large quantities and whose materials are relatively precious. Therefore, it is necessary to sort the used screws. The existing screw classification system is to classify standard screws, which cannot classify worn screws, let alone detect the degree of wear of screws. This patent uses a non-contact visual scheme to sort screws, which can effectively detect the degree of screw wear and classify worn screws.

Contents of the invention

The object of the present invention is to overcome the deficiencies of the prior art, provide a vision-based screw classification and damage detection device and method, and realize screw damage detection.

In order to achieve the above object, the technical solution adopted by the present invention is: a vision-based screw classification and damage detection device, including a detection station, a top camera, a side camera, a light source, and an industrial computer. Screws are set on the detection station. The top camera is used to take the head image of the screw, the side camera is used to detect the screw image of the screw, the light source is used to provide the required light for the top camera and the side camera, and the top camera and the side camera will The captured image data are respectively transmitted to the industrial computer for image recognition processing, and the industrial computer classifies the screws and detects damage through the image recognition processing.

The industrial computer obtains the wear data of the screws by processing the image to detect the damage and classifies the screws by the shape and size of the obtained screws.

The industrial computer is connected to the robot system, and the robot system stores the screws according to the classification according to the control signal sent by the industrial computer.

The robot system includes a robot controller, a robot body and a gripper system. The robot controller controls the movement of the robot body and the gripper according to the received signal, and places the clamped screw at a corresponding recovery position.

The device also includes a loading and unloading system. The loading and unloading system includes a material platform and screws placed on the material platform. The robot system clamps the screws on the material platform and places them on the detection station for classifying damage. detection.

The material table is a vibrating table, an inclined material box connected with the vibrating table, the vibration of the vibrating table is controlled by a vibrating table controller, the vibrating table controller is connected with an industrial computer, and an industrial computer is set above the vibrating table. The camera is used to collect the screw image data on the vibrating table and send the image to the industrial computer; the screws in the material box roll down on the vibrating table along the inclined material box, and the vibrating table works to disperse the screws, The image data on the vibrating table captured by the industrial camera is sent to the industrial computer, and the industrial computer recognizes the position of the screw according to the image and controls the robot system to grab the screw and place it on the detection station.

The detection method of the screw classification and damage detection device based on vision comprises the following steps:

(1) Put the screw in the material box, the screw part of the material box enters the vibrating table, the industrial computer controls the vibrating table controller to work, the vibrating table controller controls the vibration of the vibrating table, and disperses the screws on the vibrating table;

(2) The industrial camera above the vibrating table captures the image data on the vibrating table, and sends the image to the industrial computer, which processes the image, identifies the screw in the image and obtains the screw position information;

(3) The industrial computer sends a control signal to the robot system, and the robot grabs the screws on the vibration table according to the coordinate information, and places the screws on the detection station;

(4) The top camera and side camera on the detection station take pictures of the head image and screw rod image of the screw respectively and send them to the industrial computer;

(5) The industrial computer obtains the head shape and profile of the screw and the thread length and diameter of the screw according to the head image of the screw and the image of the screw rod; and classifies and judges damage according to the data.

(6) The industrial computer controls the robot to pick up the screws and store them in the classified storage corresponding to the screws.

In step (2), the industrial computer first performs Gaussian filtering on the collected image, uses the Canny operator to obtain the boundary of the image, and obtains a relatively stable boundary feature, and performs a morphological opening operation on the acquired boundary according to the eigenvalue defined by the screw. The boundary of the screw is screened out, and the direction of the screw and the predefined center coordinate value are calculated based on the boundary characteristics of the screw.

The image processing of the screw head by the industrial computer includes: identifying the outer circle boundary of the screw head and the boundary of the screw notch in the image, calculating the area of the notch and the area of the screw head according to the boundary, and obtaining the preset value according to the area of the screw head For the corresponding standard notch area, the wear condition of the screw head is judged by the difference between the standard notch area and the obtained notch area.

The processing of the screw image by the industrial computer includes: performing Gaussian filtering on the original image and image sharpening with the Laplacian operator method, and then using an adaptive binarization algorithm to find the connected domain of the screw part and obtain the boundary information of the screw part , the thread information of the screw to be detected is calculated based on the connected domain of the screw, and the thread wear is judged according to the thread height.

The advantages of the present invention are: the screws can be effectively dispersed through the loading and unloading subsystem, and the screws can be clamped by robots and industrial cameras, laying the foundation for subsequent classification and detection; the notch can be obtained by the boundary between the screw head and the screw rod The area, thread height and other information can be used to judge the wear condition, the wear degree of the screw can be detected, and the screw can be classified according to the shape and size of the obtained screw.

Description of drawings

The content expressed in each accompanying drawing of the description of the present invention and the marks in the figure are briefly described below:

Fig. 1 is a schematic diagram of a vibrating table of the present invention;

Fig. 2 is a principle diagram of obtaining the image of the screw at the detection station by the industrial computer of the present invention.

Detailed ways

The specific implementation manner of the present invention will be described in further detail below by describing the best embodiment with reference to the accompanying drawings.

As shown in Figure 1, the loading and unloading system includes a material table and screws placed on the material table. The robot system picks up the screws on the material table and places them on the detection station for classification and damage detection.

The material table is a vibrating table, and the obliquely set material box is connected with the vibrating table. The vibrating table is controlled by the vibrating table controller. Screw image data and send the image to the industrial computer; the screw in the material box rolls down on the vibrating table along the inclined material box, the vibrating table works to disperse the screws, and the industrial camera captures the image data on the vibrating table and sends it to the industrial control In the machine, the industrial computer recognizes the position of the screw according to the image and controls the robot system to grab the screw and place it on the detection station.

The robot system includes a robot controller, a robot body, and a fixture system. Specifically, a fixture system is set on the mechanical arm of the robot body for clamping screws. The movement of the robotic arm of the robot is run by the robot controller according to the received control instructions from the industrial computer. . The robot controller controls the movement of the robot body and the fixture according to the received signal, and places the clamped screw at the corresponding position. Here, the robot moves from the position of the vibrating table to place the screw at the position of the detection station, which is used to image the screw. Collect and process for detection.

The industrial computer controls the screw that the robot grips through the mechanical arm, and then moves the screw to the preset fixed position, that is, the position where the detection station is located. It is held by the fixture or placed on the fixed tool on the detection station. The top camera uses For shooting the head image of the screw, the side camera is used to detect the screw image of the screw, the light source is used to provide the required light for the shooting of the top camera and the side camera, and the image data captured by the top camera and the side camera They are respectively transmitted to the industrial computer for image recognition processing, and the industrial computer classifies the screws and detects damage through the image recognition processing. The industrial computer obtains the wear data of the screw by processing the image for damage detection and classifies the screw by the shape and size of the obtained screw. The wear data of the screw is determined according to the ratio of the notch area of the screw head to the standard area, and the thread wear is determined according to the length of the thread wear.

Vision-based detection method for screw classification and breakage detection device,

(1) Put the screw in the material box, the screw part of the material box enters the vibrating table, the industrial computer controls the vibrating table controller to work, the vibrating table controller controls the vibration of the vibrating table, and disperses the screws on the vibrating table;

(2) The industrial camera above the vibrating table captures the image data on the vibrating table, and sends the image to the industrial computer, which processes the image, identifies the screws in the image and obtains the screw position information; specifically, the industrial computer Gaussian filtering is first performed on the collected image, and the Canny operator is used to obtain the boundary of the image to obtain a relatively stable boundary feature. According to the eigenvalue defined by the screw, the morphological opening operation is performed on the acquired boundary to filter out the boundary of the screw. Based on the screw The boundary feature of the screw, calculate the direction of the screw and the predefined center coordinate value.

(3) The industrial computer sends a control signal to the robot system, and the robot grabs the screws on the vibrating table according to the coordinate information, and places the screws on the detection station; the detection station is a fixed position, and its position coordinates can be calibrated in advance, stored in the robot controller.

(4) The top camera and side camera on the detection station take pictures of the head image and screw rod image of the screw respectively and send them to the industrial computer;

(5) The industrial computer obtains the head shape and profile of the screw and the thread length and diameter of the screw according to the head image of the screw and the image of the screw rod; and classifies and judges damage according to the data. The image processing of the screw head by the industrial computer includes: identifying the outer circle boundary of the screw head and the boundary of the screw notch in the image, calculating the area of the notch and the area of the screw head according to the boundary, and obtaining the preset value according to the area of the screw head For the corresponding standard notch area, the wear condition of the screw head is judged by the difference between the standard notch area and the obtained notch area. The processing of the screw image by the industrial computer includes: performing Gaussian filtering on the original image and image sharpening with the Laplacian operator method, and then using an adaptive binarization algorithm to find the connected domain of the screw part and obtain the boundary information of the screw part , the thread information of the screw to be detected is calculated based on the connected domain of the screw, and the thread wear is judged according to the thread height.

(6) The industrial computer controls the robot to pick up the screws and store them in the classified storage corresponding to the screws. Judging whether the screw wear is qualified according to the wear condition, unqualified ones are not allowed to be recycled, and are directly picked up by the robot and placed in the unqualified recycling device. The qualified screws obtain the size, shape, and thread shape of the screw according to the boundary information of the identified screw. Through the preset classification, the standard, according to the size, shape and thread shape of the detected screws, the qualified screws are clamped and placed in the corresponding classification device for storage.

bolt head

If the wear area of the bolt head cross groove exceeds 20%, it is unqualified, and if it is less than 20%, it is a part that can be polished. Through the pre-set standard screw notch area and the detected screw notch area, judge whether it is qualified according to the ratio of the two. For example, if the area of the standard screw cross groove is set as S, the area of the screw cross groove after use It is S1, when S1/S>1.2, the screw is unqualified; when S1/S<1.2, the screw is qualified. The detected screw area is obtained from the screw boundary data obtained from the image of the top camera. Since the notch area of the screw is related to the size of the screw head and the shape of the notch, the preset standard notch area is a plurality of standard notch areas corresponding to the size of the screw head and the shape of the notch. Boundary is easy to get screw size and notch shape, so standard notch area can be obtained.

Screw part

When the thread wear part length exceeds 1/2 of the total thread length of the bolt or screw, it is unqualified. The judgment rule for thread wear is that when the remaining height of a certain point of the thread is lower than 1/2 of the standard thread height, set the thread detection point as the center, and the thread within 1/4 of the thread lead is unqualified.

The camera in the loading and unloading subsystem is mainly used to locate the scattered bolts and screws, that is, to obtain the position coordinates of the screws on the vibrating plate through image processing, and transmit the coordinates to the robot system through the TCP/IP protocol, and the robot realizes Grabbing tasks for screws and bolts.

Since there is a certain gap in the surface gray level of the removed screw, and the accuracy of the screw position coordinates is high for robot grasping, the image processing algorithm needs to accurately identify the screw and calculate the screw position, and the gray level of the screw surface, etc. Better robustness. Gaussian filtering is performed on the collected image first, and the boundary of the image is obtained by using the Canny operator to obtain a relatively stable boundary feature. In order to stabilize the image processing and reduce the amount of calculation, according to the eigenvalues defined by the screw, the morphological opening operation is performed on the acquired boundary, and the boundary of the screw is screened out. Based on the boundary features of the screw, the direction of the screw and the predefined center coordinate value are calculated.

A camera is installed directly above the bolt to be detected, which becomes the top camera; a camera is distributed around the bolt to be detected to detect the thread condition of the bolt. Its main functions are:

(1) The top camera is used to detect the condition of the bolt head and detect the damage of the cross groove or slot;

(2) The side camera detects the condition of the screw, including full thread and half thread, and thread damage;

The judgment index for whether the screw head wear detection is qualified is the notch area, so the notch area needs to be accurately calculated in the image. The slot area calculation is based on the extraction of slot boundaries. Perform median filtering on the collected original image, and then sharpen the image, such as using gradient algorithm or sobel operator, to strengthen the boundary of the object in the image. The boundary in the image is obtained based on the Canny operator. Since the screw detection is placed on a fixed detection station, it is only necessary to find the boundary of the screw notch and the boundary of the outer circle of the screw top in the fixed area in the image. The screw wear can be judged by calculating the area of the two boundary envelopes found in the image.

The thread detection part of the screw and bolt adopts the backlight lighting method, and the boundary of the screw part of the screw and bolt can be obtained through binarization. The image processing flow is to perform Gaussian filtering and Laplacian operator method on the original image to sharpen the image, and then use the adaptive binarization algorithm to find the connected domain of the screw part. Based on the connected domain of the screw, the major diameter, minor diameter, thread length, number of threads, and height of each thread of the screw to be detected are calculated. After the industrial computer obtains the relevant information, judge whether the wear of the screw head meets the requirements according to the notch area, and judge whether the thread wear length of the screw meets the requirements at the same time. The head wear is judged by the ratio of the worn area to the standard area, and the screw wear is judged by wear Determine the ratio of the thread length to the total thread length, divide the screws into qualified and unqualified, put the unqualified screws in the corresponding collection device, and then put the qualified screws according to the size of the screw head and the shape of the notch According to the obtained screw head size, notch shape, and screw length information, the screws are respectively placed in the corresponding recovery device, so as to complete the wear identification of the screws and the qualified screws according to the shape and size of the notch And so on for sorting and recycling.

Apparently, the specific implementation of the present invention is not limited by the above methods, as long as various insubstantial improvements are made by adopting the method concept and technical solutions of the present invention, they all fall within the protection scope of the present invention.

Claims (10)

1. A vision-based screw classification and breakage detection device, characterized in that: it includes a detection station, a top camera, a side camera, a light source, and an industrial computer. Screws are set on the detection station, and the top camera is used to photograph screws head image, the side camera is used to detect the screw image of the screw, and the light source is used to provide the required light for the shooting of the top camera and the side camera, and the top camera and the side camera transmit the captured image data to the The image recognition processing is carried out in the industrial computer, and the industrial computer classifies and detects the damage of the screws through the image recognition processing. 2. A vision-based screw classification and damage detection device as claimed in claim 1, characterized in that: the industrial computer obtains the wear data of the screw by processing the image for damage detection and obtains the shape of the screw , size to classify the screws. 3. A vision-based screw classification and damage detection device as claimed in claim 1 or 2, characterized in that: the industrial computer is connected to a robot system, and the robot system sends the screw according to the control signal sent by the industrial computer. Store by category. 4. A vision-based screw classification and breakage detection device as claimed in claim 3, characterized in that: the robot system includes a robot controller, a robot body and a clamp system, and the robot controller according to the received The signal controls the movement of the robot body and the fixture, and places the clamped screw in the corresponding recovery position. 5. A vision-based screw classification and damage detection device as claimed in claim 3, characterized in that: the device also includes a loading and unloading system, and the loading and unloading system includes a material table and screws placed on the material table , the robot system clamps the screws on the material table and places them on the detection station for classification and damage detection. 6. A kind of vision-based screw classification and damage detection device as claimed in claim 5, characterized in that: the material table is a vibrating table, the obliquely arranged material box connected with the vibrating table, the vibrating table Vibration is controlled by a vibrating table controller, the vibrating table controller is connected with the industrial computer, and an industrial camera is arranged above the vibrating table for collecting screw image data on the vibrating table and sending the image into the industrial computer; in the material box The screws roll down on the vibrating table along the inclined material box, the vibrating table works to disperse the screws, the industrial camera captures the image data on the vibrating table and sends it to the industrial computer, and the industrial computer recognizes the screw according to the image position and control the robot system to grab the screw and place it on the detection station. 7. A detection method based on a vision-based screw classification and damage detection device according to any one of claims 1-6, characterized in that: (1) Put the screw in the material box, the screw part of the material box enters the vibrating table, the industrial computer controls the vibrating table controller to work, the vibrating table controller controls the vibration of the vibrating table, and disperses the screws on the vibrating table; (2) The industrial camera above the vibrating table captures the image data on the vibrating table, and sends the image to the industrial computer, which processes the image, identifies the screw in the image and obtains the screw position information; (3) The industrial computer sends a control signal to the robot system, and the robot grabs the screws on the vibration table according to the coordinate information, and places the screws on the detection station; (4) The top camera and side camera on the detection station take pictures of the head image and screw rod image of the screw respectively and send them to the industrial computer; (5) The industrial computer obtains the head shape and profile of the screw and the thread length and diameter of the screw according to the head image of the screw and the image of the screw rod; and classifies and judges damage according to the data. (6) The industrial computer controls the robot to pick up the screws and store them in the classified storage corresponding to the screws. 8. the detection method of screw classification and damage detection device based on vision as claimed in claim 7, it is characterized in that: in step (2), industrial computer first carries out Gaussian filter to the image collected, adopts Canny operator to obtain image According to the eigenvalues defined by the screw, the morphological opening operation is performed on the obtained boundary, and the boundary of the screw is screened out. Based on the boundary characteristics of the screw, the direction of the screw and the predefined center coordinate value are calculated. . 9. The vision-based detection method for screw classification and damage detection device according to claim 7, characterized in that: processing the image of the screw head by the industrial computer includes: identifying the outer circle boundary of the screw head and the screw notch in the image Boundary, calculate the area of the notch and the area of the screw head according to the boundary, obtain the preset standard notch area corresponding to it according to the area of the screw head, and judge the screw by the difference between the standard notch area and the obtained notch area Head wear. 10. The detection method of screw classification and damage detection device based on vision as claimed in claim 7, characterized in that: the processing of the screw image by the industrial computer includes: performing Gaussian filtering on the original image and image processing by the Laplacian operator method After sharpening, an adaptive binarization algorithm is used to find the connected domain of the screw part, obtain the boundary information of the screw part, calculate the thread information of the screw to be detected based on the connected domain of the screw, and judge the thread wear condition according to the thread height.