Research On K-maximum Visual Query Of Moving Target In 3D Obstacle Space

As an important research field of spatial database,visual query has attracted the attention of researchers since it was put forward.Up to now,there are some research directions,such as visual K-nearest neighbor query,target maximum visual query and moving target maximum visual query.With the continuous development of urban three-dimensional modeling technology,its research subject gradually expands from fixed target to moving target,and its research dimension gradually extends from two-dimensional to three-dimensional space.At present,the research on maximum visual query of moving target mostly focuses on the situation that the moving trajectory of the target is known.However,in real life,the moving trajectory of the target is not necessarily known,and users may need to make predictions according to certain conditions.Moreover,the existing maximum visual query algorithm for moving targets has some defects and limitations in I/O processing and time complexity.To solve the above problems,this paper proposes a K-max visual query algorithm for moving objects,which optimizes and improves the shortcomings and defects of previous algorithms.The main research contents of this paper are as follows.Firstly,a path prediction algorithm based on transition probability is proposed.The concept of transition probability is introduced into spatial data,the geographical distribution of obstacles and points in three-dimensional space is analyzed,and the K-step transition probability of candidate paths is calculated to reasonably predict the moving target’s route.When the target deviates from the predicted route,the algorithm is refreshed,and the path prediction result is updated.The prediction results of the algorithm provide a path basis for the next step of K-max visual query of moving target.Secondly,a moving target K maximum visual query algorithm based on grid division is proposed.From the point of view of the moving target,the algorithm calculates the position of the moving target on the specified time stamp on the predicted path,and proposes a pruning algorithm based on the maximum search ball of the target to filter out invalid candidate viewpoints and unaffected obstacles.Then,the target object is divided into reasonable grids,and a grid-based visible area quantization algorithm is proposed,which quantifies the visible areas of candidate viewpoints by the number of visible grid units.Finally,calculate the number of the de-duplicated visible grid elements of all K candidate sets,and compare them to get the result set.Finally,according to the research content,an experimental environment is set up,different data scales are set,and the k value is queried,and the existing algorithm is compared with the algorithm in this paper.It is proved that this algorithm has certain advantages in time complexity and query accuracy.