Abstract
Vector data compression can significantly improve efficiency of geospatial data management, visualization and data transmission over internet. Existing compression methods are either based on information theory for lossless compression mainly or based on map generalization methods for lossy compression. Coordinate values of vector spatial data are mostly represented using floating-point type in which data redundancy is small and compression ratio using lossy algorithms is generally better than that of lossless compression algorithms. The purpose of paper is to implement a new algorithm for efficient compression of vector data. The algorithm, named space division based compression (SDC), employs the basic idea of linear Morton and Geohash encoding to convert floating-point type values to strings of binary chain with flexible accuracy level. Morton encoding performs multiresolution regular spatial division to geographic space. Each level of regular grid splits space horizontally and vertically. Row and column numbers in binary forms are bit interleaved to generate one integer representing the location of each grid cell. The integer values of adjacent grid cells are proximal to each other on one dimension. The algorithm can set the number of divisions according to accuracy requirements. Higher accuracy can be achieved with more levels of divisions. In this way, multiresolution vector data compression can be achieved accordingly. The compression efficiency is further improved by grid filtering and binary offset for linear and point geometries. The vector spatial data compression takes visual lossless distance on screen display as accuracy requirement. Experiments and comparisons with available algorithms show that this algorithm produces a higher data rate saving and is more adaptable to different application scenarios.
Highlights
Spatial data acquisition efficiency and accuracy have been drastically improved due to fast development of positioning technologies including global navigation satellite systems (GNSS), Bluetooth, Wi-Fi and others equipped on portable mobile devices [1,2]
Lossy compression reduces the volume of data at the cost of certain data accuracy, and relevant methodologies can be roughly grouped into those based on map generalization algorithms and those based on high precision to low precision data type conversion
Lossless compression algorithms are mainly based on information theory, such as Huffman encoding, LZ series encoding [21,22], which are based on data redundancy evaluation and dictionary building
Summary
Spatial data acquisition efficiency and accuracy have been drastically improved due to fast development of positioning technologies including global navigation satellite systems (GNSS), Bluetooth, Wi-Fi and others equipped on portable mobile devices [1,2]. Map generalization can be implemented based on spatial to time-frequency domain transformations, such as discrete wavelet transformation (DWT) and discrete cosine transformation (DCT) [15,16,17] by converting spatial domain information to frequency domain information and filtering high frequency coefficients after quantization This type of algorithm is normally used in raster data and image compression, for example JPEG 2000 [18]. The Geohash code length or Morton code value range of encoding results depends on the level of spatial division which can be different at various parts of research area and adaptive to local complexity of spatial objects This property has been applied in geospatial data indexing [30] and point cloud data compression [31].
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have