The supporting slats of laser flatbed machines cause process reliability problems, such as tilted parts colliding with the cutting head. In order to mitigate these problems the position of the supporting points for a part to be cut must be known, before the machines numerical control program can be changed accordingly. Being able to detect the position of supporting slats accurately is necessary to do that. This work compares image processing methods to localize the supporting slats in single images. The best features are based on filters in the frequency domain and can have accuracies above 96 %.