This paper focuses on the use of slant normalization in this context. Let us consider the system in Figure 1(b). In this example, the numeral strings are slantnormalized in order to reduce the overlap between adjacent numerals. To this end, a slope ( 2 ) is estimated from the whole numeral string image. The string recognition is done by matching numeral HMMs against the normalized string by means of dynamic programming. The construction of these models is presented in Figure 1(a). They are trained from isolated numerals, which are also slant-normalized. However, a different slope ( 1 ) is used, which is estimated from the numeral image. Although the same slant normalization method may be used to estimate 1

In this study we evaluate different HMM topologies in terms of recognition of handwritten numeral strings by considering the framework of the Level Building Algorithm (LBA). By including an end-state in a left-to-right HMM structure we observe a significant improvement in the string recognition performance since it provides a better definition of the segmentation cuts by the LBA. In addition, this end-state allows us the use of a two-step training mechanism with the objective of integrating handwriting-specific knowledge into the numeral models to obtain a more accurate representation of numeral strings. The contextual information regarding the interaction between adjacent numerals in strings (spaces, overlapping and touching) is modeled in a pause model built into the numeral HMMs. This has shown to be a promising approach even though it is really dependent on the training database.

In this paper a two-stage HMM-based method for recognizing handwritten numeral strings is extended to work with handwritten numeral strings of unknown length. We have proposed a Bayesian-based string length predictor (SLP) to estimate the number of digits in a string taking into account its width in pixels. The top 3 decisions of the SLP module are used to control the maximum number of levels to be searched by the Level Building (LB) algorithm. On 12,802 handwritten numeral strings and 2,069 touching digit pairs, this strategy has shown a small loss (0.91%) in terms of recognition performance compared to the results when the string length is considered as known.