Data Compression

DATA COMPRESSION:

Data compression of Network GIS refers to compression of geospatial data within a network GISso that volume of data transmitted across the network can be reduced. Typically, a properly chosen compression algorithm can reduce data size to 5~10% of original for images , and 10~20% for vector and textual data . Such compression ratios result in significant performance improvement.

Data compression algorithms can be categorized into lossless and lossy. Bit streams generated by lossless compression algorithm can be faithfully recovered to the original data. If loss of one single bit may cause serious and unpredictable consequences in original data (for example, text and medical image compression) lossless compression algorithm should be applied. If data consumers can tolerate distortion of original data to a certain degree, lossy compression algorithms are usually better because they can achieve much higher compression ratios than lossless ones

Data compression of network GIS is similar to other data compression algorithms on distributed computing platforms. Image compression algorithms such as JPEG had been applied since the first Web-based GIS emerged in 1993. However, the compression of vector data is introduced much later, such as the Douglas-Peuker algorithm and the work done in 2001 by Bertolotto and Egenhofer.

1 SCIENTIFIC FUNDAMENTALS

Data compression originates from information theory, which concentrates on systematicresearch on problems arising when analog signals are converted to and from digital signals and digital signals are coded and transmitted via digital channels. One of the most significant theoretical results in information theory is the so-called source coding theorem, which asserts that there exists a compression ratio limit that can only be approached but never be exceeded by any compression algorithms. For most practical signals it is even very difficult to obtain compression algorithms whose performance is near this limit. However, compression ratio is by no means the unique principal in the development of compression algorithm. Other important principals include fast compression speed, low resource consumption, simple implementation, error resilience, adaptability to different signals, etc.