Machine vision
Cylinder product surface quality inspection case
Defects such as scratches, bristles, and holes on the surface of the product not only affect the aesthetics and user experience, but also have an adverse effect on the enterprise. Machine vision technology has greatly improved the accuracy and efficiency of defect detection and promoted manufacturers The high-quality production of products has promoted the intelligent development of manufacturing.
This article mainly introduces the VisionBank intelligent vision system based on the VisionBank intelligent vision system for the "one drag five" defect detection scheme for cylindrical products, which can effectively improve the detection efficiency.
Overall solution analysis
1. Solution presentation
This solution uses five cameras to detect five sides of the product, and adopts the "one-to-five" scheme, which is compatible with the following models of products: matte surface and glossy surface.
1. Turn on coaxial light and turn off bar light when detecting matte surface;
2. Turn on the strip light and turn off the coaxial light when detecting the light surface.
Industrial Camera Station
Note: The main reason for the design of the five stations in this solution is that only the use of a large-sized coaxial light source can ensure that defects in the "arc chamfer" part of the product can be detected. If two vision systems are installed in one station, the light source cannot be installed.
2.Single-station vision system solution display
Installation distance description: The light sources configured for the five cameras are the same, and the installation positions of the camera, lens, and light source are also the same.
3.Multi-model product testing
This visual inspection line is compatible with various types of product inspection requirements, and its usage is as follows:
1. When the system is debugged, various types of testing "project files" are preset under the current conditions. Example: If you need to test C products, double-click the C project file to open the testing program that is compatible with C products.
Product defect detection
2. The camera, lens, light source and software configuration parameters cannot be changed whether the device is running, shutting down or switching product models. If the above parameters are changed due to other factors or new product models need to be added or new testing items need to be added, the relevant "qualified products" and "non-qualified products" pictures should be stored and sent to our company according to the instructions. After you have configured a new project file, replace the old project file.
3. The system is compatible with the product model principle: for unverified product models, as long as the defects visible to the human eye at any angle can be detected. The system cannot detect defects that must be seen from a specific angle.
4. Project defect detection content
Defect detection
Function description of each station camera
Camera No. 1 function description: Camera No. 1 is responsible for shooting the top surface of the product and detecting abnormalities on the top surface
Camera # 1
Detects two types of defects: one is the abnormality of the character pattern; the other is the surface defect (the scheme designed now can detect similar to the fingerprint defect)
The "pattern comparison" function is used for poor pattern printing. The consistency of the tested qualified products is very good, and all unqualified defects are detected.
Camera 2
Function description of camera # 2: There will be a pattern on the side of camera # 2. Camera # 2 is responsible for detecting characters and product surface defects.
Camera 2 1
WeChat screenshot_20191118102113
Function description of cameras 3, 4, and 5: Cameras 3, 4, and 5 are responsible for defect detection without character areas, as long as it is any abnormal "white point", "black point", "white line" different from the standard , "Black lines" or "black and white patches", etc.
Defect detection
Advantage function introduction
01. Sub-pixel precision processing
VisionBank uses 1/4 sub-pixel accuracy. For this project, one pixel represents 0.05mm. Based on this algorithm, its actual accuracy can reach 0.012mm. Make the system more accurate detection of subtle defects.
Industrial camera inspection
02. Project file automatic switching
In the future, if a device wants to switch between different models of products for testing, it can realize automatic switching, and a device can create new types of product testing programs.
Engineering tasks.webp
03. Multi-level authority management
Multi-level rights management
04.Support auto exposure strategy
05. Smart Assistant
Double-click on the image feature, and the software will intelligently analyze whether the feature at the location is "circle", "edge" or "arc".
06.Remote data storage
The user can connect the processor to the local area network and upload the image during the detection process to the designated computer via FTP.
07. Support user-defined data storage
Custom data storage
Users can define which data is stored and in what format.
08. Other gadgets
1. Comes with a soft keyboard
2.External communication test assistant
3. Image Effect Evaluation Tool
4. System operation log
5. Detection data processing histogram
6. Display interface customization settings, etc.
Other gadgets