Manual morphological analysis of bee wing images is time-consuming, so we developed a procedure for automated processing of images. The algorithm enables the extraction of individual wing images from a larger image, image enhancement, wing venation segmentation, skeletonization, intersection detection and landmark identification. Morphometric features were calculated from landmark coordinates and used to classify Carniolan honey bee colonies. We showed that the procedure enables automated data preparation for morphological analysis and that the features are useful for distinguishing selected colonies within the same subspecies.
|
