The track regularity with high accuracy is crucial important to protect the dynamic safety in high speed and heavy haul railways, normally, the inertial measurement unit (IMU)is capable to provide the precise solution for track irregularity measurement. However, to achieve high accuracy, the most existing track irregularity measurement methods seriously depended on the fiber-optic IMU, it is very expensive, also with larger size and weight, while the growing availability of low-cost and small size MEMS-IMU can yield acceptable accuracy for track irregularity by using multiple MEMS-IMUs fusion and particular geometric constraints. For these reasons, this paper proposes an accurate measurement method for track irregularity that combines the multiple MEMS-IMUs and special geometric constraints, which can be obtained noticeable measuring performances compared with fiber-optic IMU. To achieve this, we first construct the geometric constraint models for multiple MEMS-IMUs and establish their optimal installation configuration. Subsequently, an adaptive federal Kalman filter (AFKF) is developed to fuse the multiple local MEMS-IMUs information into the global filter, which can adaptive adjustment each MEMS-IMU sub-filter fusion factor and feedforward factor to obtain the highly estimation accuracy for track irregularity. Finally, the test results from both laboratory and field environment prove that the measurement accuracy of track irregularity using our proposed method are very close to the fiber-optic IMU method, and indicate a significant step forward in the track irregularity with the low-cost and small size MEMS-IMU.