Abstract: With the rapid development of data acquisition technology,the precision of point cloud of target object becoming more and more sophisticated.It's possible to build a high accuracy virtual simulation of the real world with the high-density data,but for its modelling and analyzing,the time-consuming is also growing.In this study,the spatial mesh model built through point cloud is treated as the studied object,and a parallel algorithm of model simplification under the control of parameter is studied.The steps are as follows:first,the K-adjacency table is established according to the topology of spatial mesh points;second,the independent set of points is calculated;third,the header element in each item of K-adjacency is taken as the center point,the projection plane is calculated with its neighbour points,then the K value is calculated;forth,the original model is transformed into the K value space,and the feature distribution of the model is studied;fifth,the model is simplified with the K value compared with the input parameter,loops until meet the condition;finally,the experiment is carried on series of spatial meshes.It's proved that the algorithm not only can simplify the model effectively,but also the degree of distortion is also under the control.