In order to improve the speed and accuracy of terrain measurement in farmland leveling operation, a method of farmland leveling terrain measurement was proposed based on UAV LiDAR. Firstly, high-precision farmland terrain point cloud data was obtained by using LiDAR measurement system equipped with UAV. After preprocessing the collected point cloud data, a large number of noise points along the z-axis were removed by direct filtering. Then, the progressive morphological filtering algorithm was used to segment farmland surface points and non surface points. Finally, on the basis of retaining the main farmland terrain information features, the farmland terrain point cloud reduction method based on VoxelGrid filter was used to simplify the farmland terrain point cloud data. The results showed that compared with the flatness of GNSS, the accuracy was 94.681%, 91.364%, 90.588% and 90.287% respectively; compared with the maximum elevation difference of GNSS, the accuracy was 99.391%, 98.167%, 97.025% and 98.776% respectively. Compared with the height difference distribution of GNSS, the accuracy was 99.307%, 97.914%, 98.673% and 95.110% respectively. The farmland terrain measurement method proposed based on UAV LiDAR can quickly and accurately measure the farmland terrain information, and provide data support for the next farmland leveling operation.