Computer Vision Based Quality Control Using Python

Pooja Juyal is manager at Samtel Avionics Ltd


How the system works

The blister-packed medical pills are manufactured through automated processes. These processes are generally carried out using a conveyor system. The packs are put on the conveyor belt, inspected and packed. There you can mount the camera to continuously monitor the packs for quality control. The source code for the system runs in an infinite loop. But for actual implementation, you need to adapt it and make it trigger based, so that processing happens only when the blister pack arrives under the camera.

The camera captures an image of the pack and converts it into one that shows hue difference from your predefined value. This means that pills will be shown as black because only at their places there will be no difference in hue from the predefined value. This image is inverted, so pills appear white and the rest of the surrounding area turns black.

The objective is to pre-process the image so that the pills area appears as blobs to get detected. The blobs are easy to count and the blob count indicates the number of pills in the package. This number is compared with the desired number of pills for each package, thus missing pills are automatically detected. The software beeps if the number of detected pills matches the desired value. You can also program it to do the opposite.

The steps to run the software are reproduced below for better understanding:

[stextbox id=”info”]Run Python IDLE->File->New File[/stextbox]

Paste the program or use the file Quality
Enter right values for:
1. Camera (number?)
2. peakHue (that you have found using the second program
3. Threshold through hit-and-trial method
4. minsize for blobs detection. This depends on the placement of your camera. Use hit-and-trial method to get the correct value
5. le == ?? (put the number of pills desired in your packs)

Now click Run followed by Run Module.

You will immediately see a screen with various processed images such as the one shown in Fig. 1. If the number of detected pills is equal to the desired value, the software will produce a beep sound and show OK on the console. Otherwise, it will show Fail on the console.





Please enter your comment!
Please enter your name here