Advances in image processing

Mariusz Stefan Stachera discusses digital image processing in pharma quality systems

In terms of popular internet definitions, digital image processing is the use of computer algorithms to perform image processing on digital images.

Digital image processing has many advantages over analogue image processing. It delivers a wider range of algorithms to be applied and helps to avoid problems, such as signal noise and distortion during processing. Furthermore, the range of images is not limited to two dimensions only but can be extended to three dimensions. That technology delivers automatic ways of image interpretation free of human potential mistakes.

Taking advantages of this possibility, the digital image analysis philosophy can be used during measurements in which a precise and automatic introduction of a sensor is required.

Until now, the unsolvable problem has been the introduction of automation of multi-hole dies measurements so that the device can find all the holes without any intervention by the operator.

The measurement process can be divided into several basic stages. The first is making a photo of the die – at this stage, the die picture is taken by the camera.

The next stage involves binarisation of die photography (analogue image conversion process to the binary image) - at this stage, the multi-colour photography of the die is transformed into a binary image (i.e. containing only two colours: black and white). In practice, it looks like the colourful image is first transformed into a grayscale, then selected pixels darker than the predetermined threshold are assigned to black “level”. Remaining pixels are assigned to white “level”. The threshold used in binarization strongly depends on the lighting of the image and other factors.

Finding holes is the next stage. This consists of finding clusters of black pixels representing holes in the image. First, the algorithm looks for the first hole in the outermost circle. When it is found, it approximates the positions of the next holes based on the number of holes selected by the user and checks if there are holes in those positions. The process is then repeated for successive circles, and after its completion, the algorithm knows the location of all holes in the image.

The fourth stage is converting the position of holes in the image to place these on the actual die.

THE LAST STAGE
The final stage is proper measurement. At this stage, the probe, is dropped to the next holes in a strictly specified order (first outer circles are measured, the order of holes on the circle dictates clockwise), and then pulled out of these. During continuous ejection, its position is measured, which is finally presented as the result of the measurement.

These days, one of the possible applications in the pharmaceutical industry has been introduced by Adamus. In its measuring unit Ti-2, the company proposed quality evaluation of multi-hole dies using digital image processing algorithms. Without these algorithms, the measurement problem of that kind would still remain unresolved.

Mariusz Stefan Stachera is with Adamus

Recent Issues