The invention discloses a stockpiling surface prediction method based on mine field environment perception. Firstly, mine field environment sensing is carried out, stockpiling and walls are classified, and stockpiling surface prediction is completed on the basis. The mine field environment sensing comprises the steps of point cloud preprocessing, registration, fusion, horizontal calibration, removal of useless points on the ground and above a stockpile, clustering, classification and extraction, and environment point cloud data is collected through a 3D laser radar; the environment sensing method comprises the steps of point cloud preprocessing, registration, fusion, horizontal calibration, removal of useless points on the ground and the like, clustering, classification and extraction. The stockpiling surface prediction model is established based on a Kd tree and K nearest neighbor search. According to the method, effective fusion of point cloud registration and segmentation methods is realized, the stockpiling points in the mine field environment can be effectively extracted and utilized, the stockpiling surface prediction model established based on the real stockpiling point cloud can be used for three-dimensional excavation volume calculation, and the accuracy of a calculation result can be realized.