research papers
A method is presented to automatically locate a crystal and its holder for centring on a goniometer spindle and alignment with an X-ray beam. Here, a novel algorithm that has been developed and tested with the images of users' crystals saved in an annotated database is described. The algorithm improves on the difficult situations that are commonly observed and poorly handled by the first-generation crystal-centring algorithms. These include highly transparent crystals, bad cryocooling or lens effects arising from the geometry of the drop. Most crystals have polyhedral shapes and a number of straight edges, which yield useful information. In this method, crystal detection relies on a feature-scoring system in which line extraction has the highest weight. Here, the image processing and calculations implemented in the program C3D are described. This program is designed to operate with a client program that controls specific diffractometer hardware. In order to select the best detection conditions, C3D provides various functionalities adapted to various hardware configurations.