Detection of Protein Crystallizations under Dynamic Environment

We propose an algorithm for automatic detection of protein crystallizations in photomicrographs involving many kinds of crystal growth, extragenous materials, and lighting effects, etc. Our method can improve the detection rate of plate crystals and simplify the tuning of discrimination parameters for screening objects in a photomicrograph. We define simple crystal certainty as circumference, area, and complexity of crystal outlines to discriminate whether it is a crystal or not. It is easy to set parameters by using not only numerical features but also morphological ones