The increasing demand for the efficiency in the fabrication of jewelry stones pushes producers to look for new and fast methods of the quality control. One possible way, how to reach satisfactory results, is to use the camera-based system and appropriate methods of image post-processing. Since the glass stones are very small, this task is quite challenging. This paper focuses on methods of quality assessment of the glass jewelry stones and proposes semi-automatic method of their classification based on stone feature detection and distance transform. This algorithm includes rough stone classification, which achieves 93.97% hits in our experiment, and classification based on the evaluation of single stone facets with 96.67% hits.