Abstract: Airborne radar forward-looking detection has a wide range of civil and military applications, such as runway obstacle detection, low-altitude flight, autonomous landing, missile guidance. Because the Doppler bandwidth of the detection area is almost zero, DBS and SAR technique is not satisfied with the condition. Then the azimuth resolution is determined by the 3dB beam width of the antenna. Aiming at the issue of low resolution in azimuth for airborne radar forward-looking detection, this paper presented a statistical optimization super-resolution algorithm based on an improved signal model. Bayesian maximum likelihood selection is selected to regularize the super-resolution inverse problem, Poisson and Gaussian noise distribution are used as priori information for mathematical modeling, and approximation iterative solutions are chosen to achieve radar super-resolution in azimuth. Compared with other algorithms, simulations and real radar data result verify that the algorithm reduces the noise sensitivity further, effectively suppresses generating of false targets, and improves the ability of distinguishing targets with practical applications within the main lobe.