In order to apply spatial spectrum estimation to millimeter-wave radiation array system for detecting targets with supper resolution, it is needed to solve two key-problems which are weakness of thermal signals’energy emitted from targets and the decline of performance caused by array errors. To solve this problems, an array error model with low signal noise ratio (SNR) were presented. In this model, the influence of amplitude and phase errors on signal source and channel noise were considered, and the perturbation range of the eigenvalues and the error distance of the signal eigenvector space of the autocorrelation matrix received by array were derived. Then the calibration algorithm for amplitude and phase errors based on the model was proposed, which can make spatial spectrum estimation with high resolution be used in millimeter-wave radiation array system effectively. The effectiveness and correctness of the model are verified by the experiments.