Abstract: In order to reduce the influence of noise and interference on the detection accuracy of optical spot coordinates in semi-active laser weapons, the measurement bandwidth of pulse signal increased and the effective value of the signal was extracted. By overcoming the great time-consuming and complexity of the serial fast Fourier transform (FFT) operation, parallel FFT computing method was proposed based on multi-core and parallel architecture system on chip-field-programmable gate array (SoC-FPGA) platform and OpenCL software. By this method, the time complexity of FFT (1-D) can be reduced to 1/Q times and the better acceleration effect was obtained. After comparing the computational time-consuming experiments of three platforms (advanced risc machines, digital signal processing and SoC-FPGA), the caculating time of the proposed algorithm is to 6.0449ms (1-D 4096 points) and less than that of the other two platforms with the same number of points. The results show that parallel FFT algorithm not only meets the requirement of the real-time performance of laser semi-active seeker and achieves the effect of denoising, but also can effectively reduce the influence of noise and background light.