The work aims to design a PVA crack image detection system based on FPGA with respect to the problems in the production of PVA, to achieve the timely feedback of product quality in production. The hardware components of the system were built combined with the advantages of FPGA, and the functional modules matching with such components were designed, including image acquisition, data storage, image processing, image display and so on. By optimizing the gray level transformation, image segmentation and other traditional image processing algorithms, Xilinx ISE software was used to simulate the design on the SystemGenerator software platform. The collected PVA crack images were processed. The feasibility of the design scheme was verified by identifying the PVA quality defects. Simulation and experimental results showed that FPGA could use high-speed data parallel processing to perform image processing and apply to the image recognition of PVA crack. This system provides a new route for PVA crack automatic recognition and it has certain market value.