With the increasing demand for real-time processing and high resolution in modern communication and signal processing systems, the inherent trade-off between frequency resolution and bandwidth in traditional FFT-based spectral analysis has become increasingly evident, making it challenging to simultaneously achieve rapid processing and high-precision analysis. To address this issue, this paper proposes a real-time spectral analysis method based on the Zoom-FFT algorithm. The proposed approach leverages a localized spectral refinement technique to perform high-resolution spectral analysis while reducing computational complexity and satisfying real-time requirements. In this method, the target frequency band is first down-converted to baseband using digital down-conversion. Multistage decimation filtering, incorporating both low-pass filtering and down-sampling, compresses the data while preserving essential spectral features. Subsequently, a localized high-resolution FFT is applied to the decimated signal, which enhances the detection capability for weak signals. An overlapping frame technique is also introduced to mitigate spectral leakage and improve the spectrum update rate. The method is ultimately implemented and validated on FPGA hardware. Experimental results indicate that, at a sampling rate of 250 MHz, the proposed method achieves a frequency resolution of 1 kHz for a signal with a 50 kHz bandwidth, while the FPGA’s parallel architecture further improves data processing efficiency. This integrated approach of innovative signal processing and hardware acceleration provides an effective solution for high-real-time and high-precision spectral analysis in applications such as communication signal monitoring and radar pulse analysis.