Confirming the center frequency and begin-end time is the primary task of burst signal detection in non-cooperative communication. The correlation algorithms of burst signal detection are discussed. The morphology filtering and skeletonization algorithm is adopted to preprocess the broadband time-frequency image of burst signals and suppress noise and interference. The occupied time rates of signals are computed and the center frequency of burst signals are obtained. The Power-Law detector based on Goertzel algorithm is employed and the detection of burst signal is realized. Appropriate method is taken to estimate the begin-end time of burst signal according to the coarse estimation of signal to noise ratio. Simulation results indicate that the algorithm is more effective to extract the center frequency of burst signal and can enhance the performance of begin-end time extraction.