Abstract: Image processing systems have been widely used in agricultural applications, but it is hard to realize the real-time processing of image and make low cost that can be accepted by users, therefore, it is necessary to reduce the costs of design and production. A modular design method of cotton image processing system based on Petri nets was proposed. Firstly, the image processing system was divided into functional modules, the top-level controller was modeled based on Petri nets; then the graphical model was transferred to the VHDL and the VHDL program was simulated and synthesized in the QuartusII software; Finally the image segmentation module was designed to introduce the parallel of the system. This design method of global asynchronous local synchronous (GALS) based on Petri nets can reduce the difficulty of design and debugging of image processing system, and decrease the timing requirement of the system.