symbol representation in map image compression university of joensuu department of computer science...
TRANSCRIPT
Symbol Representation in Map Image Compression
UNIVERSITY OF JOENSUUDEPARTMENT OF COMPUTER SCIENCEFINLAND
Alexander Akimov and Pasi Fränti
ACM Symposium on Applied Computing (SAC’04)March 2004, Nicosia, Cyprus
Application of map images
V e c to r m a p d a ta b a s e
S e rv e r s id e
1 . V e c to r m a pra s te ris a tio n
2 . C o m p re s s io n
C lie n t s id e
Server side
Operations:• Rasterization• Compression
Benefits:• Independent on
vector formats• Large map
databases at server side
• Low cost client side applications
Client side
=
+
Operations:• Decompression• Viewing
Map image compression
Text rasterizationHershey vector font
2
4 5
1 3
(0,0)+(5,10)+(8,0)+(–1, –1) + (2,5)+ (6,5) =
Text rotationText is stored as bitmaps:Same text after rotation consists from different bitmaps
Text rotation process
Text control data
MISS file structure
Text layer
Decoding
• Generic region decoding
• Text region decoding
1. Decoding of symbols bitmaps
2. Decoding of control data
Experimental results
Statistical data of the text rasterization
124101 201401 431204 431306
Size of dictionary 344 804 718 822
Number of symbols 2791 5440 9222 17726
Size of compressed blocks
Strip data 9930 13238 9222 17726
X sizes 351 733 609 736
Y sizes 453 706 633 731
Strip lengths 165 368 245 387
X coordinates 1068 2173 1529 2707
Y coordinates 672 1270 914 1530
Bitmap’s indexes 3332 5574 4249 7045
Conclusions
• Improved map image compression
... but where was the intelligence, computational logic, or image analysis?
• Intelligence was that we avoided all AI and IA operations at the client side.