A method is provided for processing an acquired polarimetric synthetic aperture radar (SAR) image of a region to identify a candidate pixel that correlates to a target representation. The polarimetric SAR image is composed of a plurality of pixels, and the candidate pixel corresponds to a position in the region that contains a candidate object. The process includes deconstructing J parameter components each sample, obtaining acquired values for the image from select parameter components, acquiring reference values that characterize said parameter components for the target representation, determining distance values each reference value and each acquired value, comparing the distance values against a classification criterion to determine whether the candidate pixel conforms to the target representative. The process may further include extracting N sub-apertures from the polarimetric SAR image, and combining the distance values for the sub-apertures together to obtain distance summations for comparison. The process may additionally include multiplying the distance summations with their corresponding weighting factors to obtain weighted results and combining these to produce a weighted log-likelihood function that identifies whether the pixel conforms to the target. Determining a difference may further include subtracting reference values from their respective acquired values to respectively obtain set of differences, assigning normalized defaults to the differences in response to the difference having a specified relation to parameter thresholds, and otherwise normalizing the difference, and determining natural logs of unity minus said each difference to obtain their distance values.
展开▼
