A Local Cfar Detector Based on Gray Intensity Correlation in Sar Imagery

This paper proposes a local CFAR detector based on gray intensity correlation in SAR imagery. The new detector comprehensively uses the local SCR and the strong gray correlation in ship targets. The detector can well adapt to the changing background by using local SCR, further, the detection performance improves greatly using the 2D CFAR detection by modelling the joint gray intensity PDF (JPDF) of neighbouring pixel pairs of the clutter in the local window. By using real clutter extraction procedure in the background cell, the actual 2D joint Log-normal distribution is precisely modelled which fits the JPDF of the clutter well. Using this detector, the false alarm rate (FAR) caused by speckle and local background non-homogeneity can be greatly reduced, and ship targets too close can also be detected. Under the same probability of false alarm (PFA), the probability of detection (PD) improves greatly compared with conventional CFAR detectors.