In order to attain stable and global beam-pattern characteristics of sparse arrays in phased-array 3D ima-ging sonar systems, an optimization method to design sparse arrays in both the far-field and the near-field conditionsis proposed. In this method, first, based on the analyses of the near-field beam pattern, a new energy function isbrought forward for the near-field sparse array optimization in practical engineering applications. Next, the functionis simplified to reduce the computation load. Then, an appropriate near-field sparse array is obtained by minimizingthe simplified energy function via an optimization process. Finally, a second-step optimization of the near-fieldsparse array is carried out to obtain an array with higher sparse ratio for far-field design. The proposed method isemployed to optimize a 100×100 rectangular planar transducer array, and the results are compared with thoseobtained by the existing methods. It is indicated that the proposed method helps to obtain sparse arrays possessingdesired beam-pattern characteristics for both the far-field and the near-field design in random working ranges, withthe sparse ratio being close to those of the existing methods.