Machine Vision
Beispiel für eine kundenspezifische Anpassung
Custom Camera Example
In one customized application, a line scan camera continuously captures gray value images of a homogeneous material that is passing beneath the camera. Defects in the material can be seen as gray value differences in the images.
Image analysis performed in the conventional way would transmit the image data to a PC and let the PC check the transmitted lines of pixels for possible gray value deviations. This requires extensive computing time and illustrates why it would be better to let the camera pre-process pixel data directly "on site". For example, the camera could examine each pixel to see if it exceeds a certain threshold or is out of a specificed range of brightness and would output a binary result rather than the full pixel data. The threshold could be manually defined by the camera user or be adjusted automatically by the camera using a histogram function and thus be adapted to variations in lighting and material.
Using this method, an application with a camera that captures pixel data at 8-bit depth will reduce the amount of data transmitted by a factor of eight. Since the connected PC must only focus on single bits, deviations will be found much faster. This greatly lowers computing time and opens the possibility of having one PC analyzing the images from several cameras. For defect classification, the lines that show where defects occur could be transmitted out of the camera at their full gray scale values.
In one customized application, a line scan camera continuously captures gray value images of a homogeneous material that is passing beneath the camera. Defects in the material can be seen as gray value differences in the images.
Image analysis performed in the conventional way would transmit the image data to a PC and let the PC check the transmitted lines of pixels for possible gray value deviations. This requires extensive computing time and illustrates why it would be better to let the camera pre-process pixel data directly "on site". For example, the camera could examine each pixel to see if it exceeds a certain threshold or is out of a specificed range of brightness and would output a binary result rather than the full pixel data. The threshold could be manually defined by the camera user or be adjusted automatically by the camera using a histogram function and thus be adapted to variations in lighting and material.
Using this method, an application with a camera that captures pixel data at 8-bit depth will reduce the amount of data transmitted by a factor of eight. Since the connected PC must only focus on single bits, deviations will be found much faster. This greatly lowers computing time and opens the possibility of having one PC analyzing the images from several cameras. For defect classification, the lines that show where defects occur could be transmitted out of the camera at their full gray scale values.
Example
These two images below illustrate an application where the scanned material is being checked for black impurities. To form the left image, unprocessed gray values (256 levels) are collected from the camera. In the right image, the camera is transmitting computed binary output.
© 2008 Basler AG | Impressum | Gesetzliche Bestimmungen
