Abstract: An optimal scheme is provided for the detection of moving objects observed in airborne SAR images.The keystone transform can correct the linear range migration of moving targets.After the keystone transform,moving target signatures are spread along the slow-time Doppler domain.A generalized likelihood ratio test approach is then adopted to detect moving targets and estimate their parameters in Gaussian noise.By taking into account the antenna radiation pattern,the proposed algorithm determines not only the target location,but also the two components of the velocity vector(cross-ranger and slant-range velocity).Combined with the genetic algorithm,the effective parameter estimation and high detection speed are achieved.