针对船舶导航雷达方位分辨率较低的缺点，提出一种基于约束优化理论的方位超分辨方法，在强目标信号环境下，将方位超分辨问题转化为约束优化问题.通过设计一种基于奇异值分解的正则化模型，使重建后的目标方位信息等同于模型的最小二乘解.在此基础上，采用内点法对模型进行迭代求解，重建目标原始方位信息.仿真结果表明，相比于现有范数正则化超分辨方法，该方法分辨效果良好，当信噪比达到10 dB时，分辨率可提高6倍；信噪比为20 dB时，分辨率可提高9倍；当信噪比为30 dB时，分辨率可提高4倍. 因此，本文方法具有良好的噪声适应能力，能够明显提升船舶导航雷达方位分辨精度.

A novel super-resolution method was proposed based on the constrained optimization theory for the weakness of angular resolution existed in marine navigation radar, which converted the super-resolution problem into constrained optimization problem under strong target signals. A norm regularization model was designed depending on singular value decomposition (SVD), and the restored angular information was equivalent to the least square solution of the model. On this basis, the interior point method was used to work out the optimal solution after iterations so that the original angular information was restored. Simulation results show that a favorable resolution performance is provided that the resolution can be enhanced accordingly 6, 9 and 4 times while the signal to noise ratio (SNR) equals 10dB, 20dB and 30dB respectively in comparison with the existing norm regularization super-resolution methods. Therefore, desirable capability of adaptive noise is gained by this proposed method, which can improve tremendously the resolution of marine navigation radar.