Yoshimasa Kimura

Affine-invariant recognition of gray-scale characters using global affine transformation correlation

Describes a technique of gray-scale character recognition that offers both noise tolerance and affine-invariance. The key ideas are twofold. First is the use of normalized cross-correlation as a matching measure to realize noise tolerance. Second is the application of global affine transformation (GAT) to the input image so as to achieve affine-invariant correlation with the target image. In particular, optimal GAT is efficiently determined by the successive iteration method using topographic features of gray-scale images as matching constraints. We demonstrate the high matching ability of the proposed GAT correlation method using gray-scale images of numerals subjected to random Gaussian noise and a wide range of affine transformation. Moreover, extensive recognition experiments show that the achieved recognition rate of 94.3 percent against rotation within 30 degrees, scale change within 30 percent, and translation within 20 percent of the character width along with random Gaussian noise is sufficiently high compared to the 42.8 percent offered by simple correlation.

Affine-invariant correlation of gray-scale characters using GAT iteration

Introduces a new technique of affine-invariant correlation of gray-scale characters by reinforcing correlation-based matching in two ways. The first method is the use of normalized cross-correlation as a matching measure based on definite canonicalization in order to realize robustness against image degradation. The second method is the application of an iterative global affine transformation (GAT) to the input image, so as to realize the maximal affine-invariant correlation with the target image. The advantages and effectiveness of the proposed method are both shown theoretically and demonstrated through preliminary experiments using gray-scale images of numerals subject to a wide range of affine transformations and random Gaussian noise.

Toward robust handwritten Kanji character recognition

This paper addresses the problem of establishing a robust methodology for handwritten Kanji character recognition. First, existing techniques in Japan are critically surveyed to clarify the state-of-the-art and remaining problems. Second, new and promising approaches are discussed in the light of the general field of pattern recognition technology. Third, as a promising challenge to distortion-tolerant shape matching, adaptive or category-dependent shape normalization of handwritten characters using global/local affine transformation (GAT/LAT) as a general deformation model is described.

Analysis and evaluation of dictionary learning on a handy type pen-input interface for personal use

We present the learning characteristics of personal dictionaries used for handy type pen-input interfaces. The personal dictionary grows by acquiring misrecognized input patterns whenever they are generated. Personal character pattern data is collected using a handy type tablet. The correct recognition rate of the data, which consists of 756 characters with 391 different characters, is improved from 79.9% to 96.2% for data used in learning, and from 78.0% to 95.4% for unlearnt data. On the contrary, the decrease of recognition rate by this learning is 0.1% –0.2%, which is negligible. The results show that the learning yields a handy type pen-input interface that is sufficient for practical use, moreover, this paper clarifies the learning characteristics against changesin writing condition, the number of objective categories, the number of registered patterns in the personal dictionary, and the formation process of the personal dictionary. Analysis and evaluation of the results gives useful instruction on the design of recognition systems.

Combining statistical pattern recognition approach with neural networks for recognition of large-set categories

We present a two-stage hierarchical system consisting of a statistical pattern recognition (SPR) module and artificial neural network (ANN) to recognize a large number of categories including similar category sets. In the first stage, the SPR module performs classification. If the first candidate does not belong to a pre-determined similar category set, the first candidate is accepted as the final result; otherwise, the first candidate is sent to the ANN module. In the second stage, ANN performs classification for similar categories to select a correct candidate from the predetermined candidate set designated by the first candidate. The new scheme offers improved system performance by sharing tasks between SPR and ANN according to the degree of classification difficulty and forming specialized ANNs for each similar category. The system achieves higher performance for the recognition of 3,201 handprinted characters than a traditional system constructed with just the SPR module.

Affine-invariant gray-scale character recognition using GAT correlation

This paper describes a new technique of gray-scale character recognition that offers both noise-tolerance and affine-invariance. The key ideas are twofold. First is the use of normalized cross-correlation to realize noise-tolerance. Second is the application of global affine transformation (GAT) to the input image so as to achieve affine-invariant correlation with the target image. In particular, optimal GAT is efficiently determined by the successive iteration method. We demonstrate the high matching ability of the proposed method using gray-scale images of numerals subjected to random Gaussian noise and a wide range of affine transformation. The achieved recognition rate of 92.1% against rotation within 30 degrees, scale change within 30%, and translation within 20% of the character width is sufficiently high compared to the 42.0% offered by simple correlation.

Optimization of key character set for word recognition using genetic algorithm

In the context of word recognition, for a word set with high similarity between words, when using word matching based on key characters, a method is proposed for optimal selection of key characters so as to reduce the average number of candidate words while ensuring a high probability of success in the search for the right word. With the proposed method, the success probability of the word search and the average number of candidate words are calculated for all combinations of key characters, and the optimal combination is then selected. To implement this principle, a new method was developed to calculate the average number of candidate words. To make the proposed principle practicable in terms of computation requirements, a method for seeking an approximate solution using genetic algorithms was proposed. The validity of the proposed method was confirmed experimentally in the recognition of personal names written with katakana characters.

Estimation of word matching accuracy in recognition of poor‐quality characters

A method is proposed for estimation of word matching accuracy in recognition of poor-quality characters. With the proposed method, all kinds of characters that might be recognized from a word are considered as networked sets, and among these sets, the sum of the occurrence probabilities is calculated for elements that cause errors in word matching. This principle makes it possible to treat all errors in word matching, which is adequate in estimation of accuracy in the case of poor-quality characters. However, results of word-level character recognition are presented as sets of candidate characters, which means an enormous number of combinations and, hence, extensive computation. In this connection, the number of combinations is materially restricted by ignoring information related to categories that are irrelevant to word matching. The proposed method makes possible an estimation of word-matching accuracy in case of frequent occurrence of poor-quality characters (rejected or misread characters), which was impossible using conventional methods. The proposed estimation method was applied to recognition of product codes, and proved efficient in accuracy estimation for poor-quality characters.