Industrial component inspection

Automated visual inspection systems are designed to perform quality control, inspection, and data collection. One or more imaging device captures photographs at various stages of the manufacturing process. These photos are analyzed to obtain important information, for example: rate of manufacture, defects in products, defects in packaging, robot guidance etc.

 

An example is the process of inspecting printed circuit boards. For a PCB to perform as designed, all components must be assembled correctly. This may include components that vary significantly in size: from fractions of a millimeter to a few centimeters. Inspecting these components is done with a range of methods from manual inspections to automated tests. There are a range of testing methods, including manual inspection, automated visual inspection, automated X-ray inspection, and in-circuit testing. A few articles on these methods are found here: [1, 2, 3]

 

In a typical manufacturing unit that generates say 40-50K PCBs a day, inspecting all of them manually is not feasible. An Automated Optical Inspection (AOI) system uses several light sources and one or more still or video cameras to capture the PCBs. Multiple light sources are needed because components can be present on the board with different sizes, and they can create shadows. Multiple lights from various angles will eliminate such shadows. Multiple cameras are needed because a single camera may not be able to capture the board with sufficient detail. An Artificial Intelligence or Computer Vision (AI / CV) algorithm may then detect errors on the board just like a human would. A demo video of such a system is given below: 

 

 

 

Image processing and computer vision algorithms can be used in inspecting products in other industry verticals, for example in the fast moving consumer goods industry. This vertical deals with prepackaged products like beverages, snacks, soaps, shampoos etc. Here, the outer packaging, color, consistency are important parameters for quality control. A video that describes this problem and a high level view of the solution is below:

 




Comments

Post a Comment

Popular posts from this blog

Discrete Wavelet Transform

Code and sample files for this demo can be downloaded from here . Code uses only libraries from the ACV-VirtualMachine.No extra libraries are needed. You can unzip the code to any folder on your system. All paths used in the code are relative. We are using the folder "~/acv/code/dwt_demo/" Compiling and running the code is fairly straight forward. You only need to run the following two commands: acv@acv-vm : ~/acv/code/dwt_demo/bin $ cmake ../src/wavelet_demo_others/ -- The C compiler identification is GNU 7.5.0 -- The CXX compiler identification is GNU 7.5.0 -- Check for working C compiler: /usr/bin/cc -- Check for working C compiler: /usr/bin/cc -- works -- Detecting C compiler ABI info -- Detecting C compiler ABI info - done -- Detecting C compile features -- Detecting C compile features - done -- Check for working CXX compiler: /usr/bin/c++ -- Check for working CXX compiler: /usr/bin/c++ -- works -- Detecting CXX compiler ABI info -- Detecting CXX compiler ABI info - ...
Read more

Geometric image transforms

Image
 The sample code and data for this tutorial can be downloaded from here . The result files are here . The accompanying slide deck for this tutorial is here . After downloading the code, run the following commands:  acv@acv-vm : ~/vision/practice/bin $ cmake ..; make; ./homework transformDemo -- The C compiler identification is GNU 7.5.0 -- The CXX compiler identification is GNU 7.5.0 -- Check for working C compiler: /usr/bin/cc -- Check for working C compiler: /usr/bin/cc -- works -- Detecting C compiler ABI info -- Detecting C compiler ABI info - done -- Detecting C compile features -- Detecting C compile features - done -- Check for working CXX compiler: /usr/bin/c++ -- Check for working CXX compiler: /usr/bin/c++ -- works -- Detecting CXX compiler ABI info -- Detecting CXX compiler ABI info - done -- Detecting CXX compile features -- Detecting CXX compile features - done -- Found PkgConfig: /usr/bin/pkg-config (found version "0.29.1") -- Found OpenCV: /usr/local (found...
Read more

Testing image capture with high speed cameras

Image
In industrial systems, objects may be moving along conveyor belts or carried on pick and place machines. A camera based inspection solution for such situations needs to take into accoun that the object in question may not be stationery, or at a fixed distance from the camera. Ergo, the captured image may not contain the object in focus, or not have the object in view at all. One option would be to stop the conveyor belt to take a photo of the object:   However, stopping a conveyor belt is not always possible. You would need cameras that can capture a clear image even when objects are moving. Enthusiasts naiively assume that one can take a video and perform analysis on individual frames of the video. Unless the video has the correct frame rate and good resolution, the solution would not work. Assume that you need to implement an industrial imaging solution to monitor items moving on a conveyor belt. You should consider specifications such as these: Distance from camera t...
Read more