Abstract: For signal monitoring of frequency hopping in complex electromagnetic environment, a signal extraction based on weighted threshold was proposed, which could solve the difficult problem of frequency hopping signal extraction in low signal to noise ratio(SNR) condition. This paper conducts time-frequency analysis for received signal by short-time Fourier transform, which could obtain the time-frequency matrix reflecting variety of frequency over time. Then, signal feature of each segment is extracted by weighted threshold. Frequency hopping parameters could be estimated by means of histogram after clustering and screening. The screening process can also have the great capacity to delete constant frequency under strong constant frequency interference. The processing of simulation data verifies the effectiveness of algorithm in this paper.