NPEN++: An On-line Handwriting Recognition System
Stefan Jaeger, S. Manke, Alex H. Waibel · 2000
This paper presents the on-line handwriting recognition system NPen++ developed at the University of Karlsruhe and the Carnegie Mellon University. The NPen++ recognition engine is based on a Multi-State Time Delay Neural Network and yields recognition rates from 96% for a 5000 word dictionary to 93.4% on a 20,000 word dictionary and 91.2% for a 50,000 word dictionary. The proposed tree search and pruning technique reduces the search space considerably without loosing too much recognition performance compared to an exhaustive search. This allows running the NPen++ recognizer in real-time with large dictionaries. 1 Introduction This paper describes the preprocessing steps, the computation of features, the recognizer with training and testing, and the dictionary based search of the NPen++ handwriting recognition system. Section 2 begins with the description of the normalizing preprocessing steps in NPen++. Section 3 shows different features computed after preprocessing. The core...