Sea-Surface Small Target Detection Based on Four Features Extracted by FAST Algorithm

On account of current algorithm and parameter design difficulties low detection accuracy in feature extractions small target detections sea clutter environment, this paper proposes a correspondingly improved four extraction method by FAST. After the short-time Fourier transform is applied, time–frequency distribution spectrogram original data generated. Candidate points (CFP) are first extracted FAST algorithm, then four-feature implemented with DBSCAN combined. The distinction enhanced through optimization. Upon construction four-dimensional vectors, XGBoost classifier classifies detects these vectors. genetic optimizes hyperparameters updates decision threshold real time to control method’s false alarm rate. IPIX dataset employed for experimental verification. Verification results confirm that proposed has better performance than several other currently used methods. 7% 13.8% when observation set at 0.512 s 1.024 s, respectively.