A new compound eye model with wider fields of view and higher agility was introduced to track the low-flying targets in a complex background. The struction and preparation of the compound eye model were described and its imaging channels were traced by Zamax to evaluated imaging characteristics.The calibration and detection of the compound eye were introduced,then LM neural network calibration algorithm was trained to build the relationship between object points and corresponding image points. This calibration algorithm provides an accurate direction angle prediction from their corresponding image points, and it is easy to integrate into the system. Preliminary experimental results for neural network calibration were presented and evaluated, which shows that the residual errors between actual and measured direction angles are around 10-3~10-4 rad. In detection simulation experiments, several points were calculated and results show that the errors between actual and calculated coordinates of position are within 3%. This is a good result for the compound eye sensor that sacrifices the spatial resolution to improve the angle resolution.