In accordance with the requirements of a real-time spectral detection system, the Fourier transform spectral theory is utilized to obtain the spectral information to be detected. First, the basic operation principle of the real-time spectral detection system is presented and a radix-2 FFT algorithm is optimized. Then, the design of the hardware structure of the algorithm is described in detail, the address generation and control process designed by an Algorithm State Machine (ASM) is illustrated emphatically and the hardware design of the FFT module is completed by using a XC3S400 chip with an IP core produced by Xilinx Inc. on an ISE9.1 software development platform. Finally, the Testbench program written in VerilogHDL language is used to simulate the functions of the FFT module on ModelsimSE6.3f, a third-party simulation software. The simulation result is compared with the result calculated by Matlab in theory. It shows that the FPGA hardware is designed correctly. It takes about 8.6μm for the chip operating at 100MHz to implement 256-point 16 bit radix-2 FFT data, which can meet the real-time requirement for spectral detection.