Abstract: Counting the total volume of grain in the granary, which represents an indispensable work in grain storage management, is of great importance to the macro grain regulation in China. In response to the low efficiency and accuracy of traditional way to survey and map the grain-pile volume in granary, this paper proposed a method for rapidly re-constructing the 3D model of the grain pile based on 3D laser scanner technique, and a method for accurately calculating the grain-pile volume based on constructing 3D laser monitoring system. This system employed the DME5000 of German SICK, the 3D laser range finder with its range finder installed on the grain pile and then created a Cartesian coordinate system in a granary. With the range finder scanning the surface of grain pile, distance can be calculated between the point on the grain pile and the bottom of the granary, and the original point cloud data of the information about the grain pile can be obtained after scanning some points. As noise point could not be avoided during the process of collecting those data because of the influence from environment and irregular point clouds were scattered, the standard point cloud data could be acquired by processing the original data through noise-eliminating, jointing and gridding. Median filter was used to remove noise while weighted average algorithm was applied in polygonizing. The standard point cloud data were processed with 3D triangulation in accordance with the Delaunay principle, on the basis of which three-dimensional curve surface model of grain piles were built. Then the 3D model of the grain pile could be built with the help of 3D rendering engine on the basis of Windows Presentation Foundation, a technology from Microsoft. At the same time the grain-pile volume could be calculated by means of square grid. The principle of the grid algorithm was to divide the bottom of the grain pile into a number of squares so that the total volume of the grain pile could be equal to the total volume of the columns with a square bottom. The system was firstly adopted in an experiment to calculate the volume of the grain pile in a laboratory which located in Beijing Forestry University. Due to the small volume of the grain pile for experiments, accurate measurement could be done manually. The actual volume of the grain pile was 2.205 m3. By completing scanning ten times and analyzing the data acquired in this experiment, we got the relative tolerance of only 0.318% between the actual volume of the grain pile and the one calculated by scanning the grain pile with this system, with the standard deviation of 0.0083m3. Another experiment was conducted in a granary in Cangzhou, Hebei province. The experimental result was compared with the one acquired by means of manually traditional surveying and mapping method. After ten times of surveying and mapping with the help of this system, we got the average grain volume of 22 469.97 m3, with the standard deviation of 42.92 m3. The volume calculated with traditional manual method was 22 466.50 m3, with the standard deviation reaching 70.25 m3. The experimental results showed that this system could acquire accurate and stable grain-pile volume , and definitely meet the requirement of granary management in real works. This provides an effective way to improve the efficiency and accuracy of accounting inventory in the management of granary in reality.