Document image analysis
Mending broken handwriting with a macrostructure analysis method to improve recognition
Pattern Recognition Letters
Document Image Quality: Making Fine Discriminations
ICDAR '99 Proceedings of the Fifth International Conference on Document Analysis and Recognition
Prediction of OCR accuracy using simple image features
ICDAR '95 Proceedings of the Third International Conference on Document Analysis and Recognition (Volume 1) - Volume 1
Image Analysis and Mathematical Morphology
Image Analysis and Mathematical Morphology
Hi-index | 0.00 |
We present a method for automatically selecting thebest filter to treat poor quality printed documents usingimage quality assessment. We introduce five qualitymeasures to obtain information about the quality of theimages, and morphological filters to improve theirquality. A training set of 370 images was used to developthe system. Experimental results on the test set show asignificant improvement in the recognition rate from73.24% using no filter at all to 93.09% after applying afilter that was automatically selected.