Fast System for the Retrieval of Ornamental Letter Image M. Delalandre 1, J.M. Ogier 2, J. Lladós 1 1 CVC, Barcelona, Spain 2 L3i, La Rochelle, France L3i Meeting L3i, La Rochelle, France Thursday 13th July 2007

Plan Introduction Our System Works in Progress Conclusions and Perspectives

Introduction (1/3) Old Printed Graphics Old books of XV° and XVI° centuries Bartolomeo (1534)Alciati (1511)Laurens (1621) figure ornamental letter headline 41% ornamental letter 59% others Graphics type 63% textual 37% graphical Foreground pixel [Jour’05] 4755 (3.4 per page)Graphics 1385Page 46Book Graphical parts CESR Database 1. Old printed graphics 2. Needs of Historian People 3. Problematic & Approaches

Vascosan 1555Marnef 1576 (1) Wood plug tracking Printing house plug exchange copy Introduction (2/3) Needs of Historian People (2) User-driven historical metadata acquisition Metadata file Metadata file Metadata file Metadata file Without retrieval With retrieval more faster reduce error Wood plug (bottom view) Retrieve similar printings Plug 1 Plug 2 Plug 3 Printing 1 Printing 2 Meeting with people of CESR (Tours) 1. Old printed graphics 2. Needs of Historian People 3. Problematic & Approaches

Introduction (3/3) Problematic & Approaches DB query results result Noise Offset Shape complexity Accuracy Scalability several hundred of classes several thousand of images Real-time Problematic 1. Old printed graphics 2. Needs of Historian People 3. Problematic & Approaches Existing Approaches descriptors Fast Accurate To scalar [Loncaric’98]  Hough, Radon, Zernike, Hu, Fourrier  scaled and orientation invariant  fast  local (character, symbol, digit) To image [Gesu’99]  template matching, Hausdorff distance  no scaled and orientation invariant  slow  global (scene) Template matching versus scalar descriptors Accuracy and low complexity are needed for our problematic, Key idea: to work on a compressed representation of image Key Idea

Plan Introduction Our System Works in Progress Conclusions and Perspectives

Our System (1/4) System Overview Query Compression Centering and Comparison R1 R2 R3 Checking Image Database

Our System (2/4) Checking Compression Centering and Comparison Checking Several image providers Several digitalization tools Long time process Human supervised Complex plate-form … Digitalization problems 250 to 350Resolutions UncompressCompression TiffFormats grayModel MpSize 2038Files Results QUEID Engine Base accepted rejected Parameters Checking

Our System (3/4) Compression Compression image foreground background both Results Compression Centering and Comparison Filtering

Centering x2x2 x2x2 x2x2 x1x1 x1x1 x1x1 x2x2 x2x2 x1x1 line (y) image 1 line (y+d y ) image 2 x stack reference while x 2  x 1 handle image 2 while x 1  x 2 handle image 1 Comparison Our System (4/4) Centering and Comparison Compression Centering and Comparison Filtering Results Max Mean Min Time s Size k.pixel Max Mean Min Time s Size k.run image database query image 7 times faster

Plan Introduction Our System Works in Progress Conclusions and Perspectives

Mean query of 40 s, how to reduce again without using a lossless compression and to loose accuracy ? Level 1 : image sizes Level 2 : black, white pixels Level 3 : RLE comparison Our first system Works in Progress (1/3) How to process the distance curve ? 11 22 if  1 -  2 < 0 push x, cluster while  1 -  2 < 0 next Using a basic clustering algorithm ‘elbow criteria’ query 1 st Level 2 sd Level To use a system approach using different level of operator (from more speed to more accurate) to select image to compare Our key idea Speed Depth

From 4% to 59%, how to reduce the variability ? To work on a better selection criteria seems ambiguous … Works in Progress (2/3) To add an intermediate operator between scalar and image data Our key idea Selection results 59%Max 24%Mean 4%Min Selection % 4 times faster

Base IHM Retrieve engine control display retrieve Labels driven labelling Bench1Bench2 To produce Example of query result Same plug Next plug Query First results seem good, but how to get the ground truth and to evaluate our system? Works in Progress (3/3) To use our engine to produce benchmark database Our key idea

Plan Introduction Our System Works in Progress Conclusions and Perspectives

Conclusions QUEID to check image database Speedup of comparison (  30 times faster) RLE compression (   7 times faster) Image selection (   4 times faster) Perspectives To add operator to reduce the variability of selection process and to speed again the process To extend our system to do performance evaluation Conclusions and Perspectives