Unconstrained handprint recognition using limited lexicon
1993
A word recognition system has been developed at NIST to read free-formatted text paragraphs containing handprinted characters.The system has been developed and tested using samples of handprint from NIST Special Database 1 .This database of binary images contains 2,100 different writers' printings of the Preamble to the U. S. Constitution.Each writer was asked to print these sentences in an empty 70mm by 175mm box.The Constitution box contains no guidelines for the placement and spac- ing of the handprinted text, nor are there guidelines to instruct the writer where to stop printing one line and to begin the next.While the layout of the handprint in these paragraphs is unconstrained, a limited-size lexicon may be applied to reduce the complexity of the recognition application.The Preamble contains 38 unique words comprised of 35 unique upper-case and lower-case letters, ignoring punctuation marks.The system is divided into four general components.1) The Constitution box is located within a full-page image, and the handprint within the box is isolated.2) The subimage containing the handprinted text is segmented using coimected component labeling, and the resulting blobs are sorted into correct reading order.3) The segmented blobs are classified using feature-based nemal network recognition.4) Words are parsed from the line-ordered classifications using the lexicon to locate word boundaries and to correct classification and segmentation errors.These components have been combined into an end-to-end hybrid system that executes across a UNIX file server and a massively parallel SIMD computer.The recognition system achieves a word error rate of 49% across all 2,100 printings of the Preamble (109,096 words).This performance is achieved with a neural network char- acter classifier that has a substitution error rate of 14% on its 22,823 training patterns.This demonstrates the power of using a limited-size lexicon to parse words from a less than optimal character classifier.This paper discusses the word recognition system in detail.