A characteristic bitmap coding method for vector elements based on self-adaptive gridding.

Spatial index is a key component of Geographic Information Systems (GISystems). To date, an increasing number of spatial indexes have been developed to enhance the efficiency of spatial analysis and spatial query. Approximate expressions are adopted in the foundation of spatial index construction, to assist data organisation, e.g., bounding box of vector elements can be employed to build R-tree index. However, R-tree index using such bounding box expresses elements approximately, which usually results in redundancy and excessiveness due to inherent roughness of this method. This study proposes a characteristic bitmap coding method, termed the QCODE method, which generates approximate expressions of vector elements based on self-adaptive gridding. Based on the sizes of vector elements, this method selects the grid at an appropriate level in a self-adaptive manner, discretises the vector elements into grids through a rasterisation operation, as well as compresses and encodes this information as the code of a characteristic bitmap, i.e., the QCODE. The ‘bit’ data type is used in the design of QCODE to restrict the approximate expression of vector elements into finite bytes, providing more precise filtering as compared to the case in which only bounding box is used. With its distinct characteristics, the QCODE is introduced for the improvement of R-tree index. The results of experiments show that, combined with R-tree index, this method can reduce the filtering amount of vector elements as required for spatial analysis, and accelerate the execution efficiency of the entire process of GIS spatial analysis. [ABSTRACT FROM PUBLISHER]