Automatic detection method of bacteria

In the bacteria images automatic analysis system, bacteria detection is a vital step. According to the characteristics of sample images, this paper proposed a method based on color space conversion, self-organizing feature map (SOFM) neural network and morphological operations. First, the conversion formula is built to convert RGB color space to CMYK color space. Second, according to the information of M channel in the CMYK color space, SOFM is used to reduce gray level and then the binary image can be got. Third, morphological operation is used to distinguish and count single bacterium, mass bacteria, and color drop. With abundant samples, experimental results indicate our method can extract the objects both accurately and availably.