Small Target Detection in Sea Clutter based on Normalized Hurst Exponent and Phase Linearity Degree

This paper aims to design a feature-based detector to detect sea-surface small targets in high-resolution sea clutter. Amplitude sequence of sea clutter is modeled as a fractal process in the time scale from 0.01 second to several seconds and Hurst exponent is used as a single feature to detect small targets. In the view of the dependency of Hurst exponents of sea clutter on sea state, the normalized Hurst exponent (NHE) is used for detection. However, the absence of phase information of radar returns in detection restricts the ability of the fractal-based detectors. Phase linearity degree (PLD) is used as the other feature to detect targets together. The radio of the NHE and PLD is used as a new test statistic to build a simple but effective thresholding detector. Through the experiments on the IPIX and CSIR radar databases, the proposed detector attains better detection performance. ^