In this study, we adopt a criterion of Barker code to generate a high-resolution image from sparse flight samples to establish a three-dimensional (3-D) imaging model of airborne array SAR. Under the condition of motion error, we utilize the Modified Uniformly Redundant Arrays (MURA) modulation and 3-D Back Projection (BP) algorithm to obtain 3-D complex image pairs under each flight. Based on interferometry and Compressed Sensing (CS) in frequency domain, the array deformation error compensation is realized. The phases of 3-D complex image formed by the echo corresponding to negative MURA modulation are referred to perform phase compensation on each single-pass complex image to restore the image phase relation of each flight. Coherent accumulation of each complex image is implemented to realize high-resolution 3-D imaging under sparse flight sampling. Simulation analysis and experimental data verify the feasibility of the proposed method.