The invention provides a color image segmentation method and device based on a pulse coupling neural network. The method comprises the following steps that: S1: converting the color space of a first color image into a LAB color space, and generating a second color image; and S2: using an improved PCNN (Pulse Coupled Neural Networks) model to independently label corresponding tags on all pixels in the second color image, wherein the improved PCNN model comprises a connection control unit. By use of the method, the first color image is subjected to space conversion, the frequency spectrum input of the color image is received through the connection control unit, and the improved PCNN model is used for labeling the pixel so as to realize color image segmentation.