Abstract: The image analysis technology based on morphological differences has become an important method of algae recognition in recent years. A skeleton similarity matching method was proposed to recognize microscopic images of diatom cells. The gray surface vector model of image was established to make segmentation by keeping obscure seta; The skeleton was used to represent the Chaetoceros, and it was decomposed hierarchically by competition strategy. The Chaetoceros object was represented by skeleton tree by forming the rachis elements and branches into it. The similarity mode for microscopic images of Chaetoceros was established by defining the topological and geometric difference. Experimental results show that this algorithm can achieve the better recognition of several kinds of Chaetoceros.

Key words: microscopic image of diatom cell, orientation angle model of object, skeleton decomposition, skeleton tree