Small Target Detection In Sea Clutter Based On Fractal Theory

Sea clutter refers to radar backscatter wave from sea surface, and it seriouslyinterferes in detection performance of targets within sea clutter. So it is significant tostudy sea clutter for radar system design, radar signal processing and targets detectionwithin sea clutter. Fractal is a branch of nonlinear science, which has comprehensiveapplications on a great many research areas. It has been researched how to use theFractal theory on sea clutter processing and small targets detection in this paper.Theoretical and experimental researches are combined in this dissertation. On onehand, the latest fractal theories are used to analyze and processing sea clutter. On theother hand, IPIX real-life data is used to verify the effectiveness of the theories andmethods.This paper mainly makes a study of the different fractal parameters between seaclutter and target echo in time domain and fractional Fourier transform domain, andanalyzes the quality of several fractal parameters based on IPIX real-life data, in theend, detects the small target in sea clutter with the improved method. It mainlycontains the following work:1. Study the fractal features of sea clutter in time domain. Analyze the differentcharacters between the sea clutter and target echo with box fractal dimension,fractional Brownian motion model and high order fractal characteristicsrespectively. Compare the qualities of different methods.2. Study the fractal features of sea clutter in fractional Fourier transform domain.The possibility was verified that sea clutter is somehow fractal. A method fordetection of sea surface target was presented using fractal dimensiondifference in fractional Fourier transform domain, and is compared with themethod in time domain. At last, analyze the fractal character of sea clutter in FRFT domain by introducing extended fractal theory.3. Extract multiscale Hurst parameters to form characteristic vector, and combinewith high order fractal characteristic parameter to establish Bayes targetdetection system based on Hurst exponent in FRFT domain. The experimentalresults show that this algorithm has better detection performance than othermethods.