The paper first gives an overview on the technical advances in the field of character and document recognition, decade by decade. Then, it highlights key technical developments especially for Kanji (Chinese character) recognition in Japan. Technical issues around post address recognition are then discussed, which have promoted advanced techniques including information integration. Robustness design principles are introduced. Finally, future prospects are discussed.

Gabor filter feature has been applied to character recognition but was not compared with the best direction feature: gradient feature. In this paper, we propose a principled method for implementing Gabor filters for character feature extraction and compare the recognition performances of Gabor feature and gradient feature on three databases. The results show that Gabor filters with low orientation sensitivity and broad frequency band favor recognition accuracy. The Gabor feature performs comparably or better than the gradient feature on two of the three databases, but is inferior on the rest one.

In integrated segmentation and recognition of character strings, the underlying classifier is trained to be resistant to noncharacters. We evaluate the performance of state-of-the-art pattern classifiers of this kind. First, we build a baseline numeral string recognition system with simple but effective presegmentation. The classification scores of the candidate patterns generated by presegmentation are combined to evaluate the segmentation paths and the optimal path is found using the beam search strategy. Three neural classifiers, two discriminative density models, and two support vector classifiers are evaluated. Each classifier has some variations depending on the training strategy: maximum likelihood, discriminative learning both with and without noncharacter samples. The string recognition performances are evaluated on the numeral string images of the NIST Special Database 19 and the zipcode images of the CEDAR CDROM-1. The results show that noncharacter training is crucial for neural classifiers and support vector classifiers, whereas, for the discriminative density models, the regularization of parameters is important. The string recognition results compare favorably to the best ones reported in the literature though we totally ignored the geometric context. The best results were obtained using a support vector classifier, but the neural classifiers and discriminative density models show better trade-off between accuracy and computational overhead.

In character string recognition integrating segmentation and classification, high classification accuracy and resistance to noncharacters are desired to the underlying classifier. In a previous evaluation study, the modified quadratic discriminant function (MQDF) proposed by Kimura et al. was shown to be superior in noncharacter resistance but inferior in classification accuracy to neural networks. This paper proposes a discriminative learning algorithm to optimize the parameters of MQDF with aim to improve the classification accuracy while preserving the superior noncharacter resistance. We refer to the resulting classifier as discriminative learning QDF (DLQDF). The parameters of DLQDF adhere to the structure of MQDF under the Gaussian density assumption and are optimized under the minimum classification error (MCE) criterion. The promise of DLQDF is justified in handwritten digit recognition and numeral string recognition, where the performance of DLQDF is comparable to or superior to that of neural classifiers. The results are also competitive to the best ones reported in the literature.

Directional features have been successfully used for the recognition of both machine-printed and handwritten Kanji characters for the last decade. This paper attempts to explain why the directional features are effective. First, the advances of directional features and related methods are briefly reviewed. Then the properties that the similarity measure should hold are discussed and simulation experiments of directional pattern matching are conducted to validate the properties. This analysis is expected to inspire the design of new and more effective features. 1.

Nonlinear normalization (NLN) by line density equalization has been popularly used in handwritten Chinese character recognition (HCCR). To overcome the intensive computation of local line density and the excessive shape distortion of NLN, we tested some alternative methods based on global transformation, including a moment-based linear transformation and two nonlinear methods based on quadratic curve fitting. The alternative methods are simpler in computation and the transformed images have more natural shapes. In experiments of HCCR on large databases, the alternative methods have yielded comparable or higher accuracies to the traditional NLN.

This paper describes a handwritten character string recognition system for Japanese mail address reading on a very large vocabulary. The address phrases are recognized as a whole because there is no extra space between words. The lexicon contains 111,349 address phrases, which are stored in a trie structure. In recognition, the text line image is matched with the lexicon entries (phrases) to obtain reliable segmentation and retrieve valid address phrases. The paper first introduces some effective techniques for text line image preprocessing and presegmentation. In presegmentation, the text line image is separated into primitive segments by connected component analysis and touching pattern splitting based on contour shape analysis. In lexicon matching, consecutive segments are dynamically combined into candidate character patterns. An accurate character classifier is embedded in lexicon matching to select characters matched with a candidate pattern from a dynamic category set. A beam search strategy is used to control the lexicon matching so as to achieve real-time recognition. In experiments on 3,589 live mail images, the proposed method achieved correct rate of 83.68 percent while the error rate is less than 1 percent.

We propose a new method to recognize the machine-printed monetary amount based on two-dimensional segmentation and bottom-up parsing. In conventional segmentation-based methods, the system segments the image only along the direction of the character line. This new method segments the image both horizontally and vertically, and extracts candidates of character segments correctly if there are many noises or characters are fragmented. A parsing module detects the optimal sequence of candidate segments using linguistic knowledge. In our method, a context-free grammar describes the linguistic constraints in the monetary amounts. We devised a new bottom-up parsing technique that interprets the results of character classification of the two-dimensionally segmented sub-images. We tested the validity of the new method using 1,314 images, and found that it improves the recognition accuracy significantly.

Proposes a handwritten character string recognition method for Japanese mail address reading on very large vocabulary. The recognition is performed by classification-embedded lexicon matching based on over-segmentation. The lexicon contains 111,349 address phrases and is represented in a trie structure. In recognition, the input text line image is matched with all lexicon entries by beam search to obtain reliable character segmentation and retrieve valid phrases. A classifier is embedded in lexicon matching to select from a dynamic set the characters matched with a candidate pattern. The beam search and the character classification jointly enable accurate phrase identification in real time. In experiments on 3,589 live mail images, the proposed method achieved correct rate of 83.68% with error rate less than 1%.

Presents a concept of an information capturing camera, which is an enhanced digital/video camera that can recognize characters in captured images and show relevant information, such as a translated version of the words in the image into another language. Also discussed are technical issues and possible approaches. Technical issues include detection of the text image regions, perspective distortion normalization, binarization of character images in unknown colors, recognition of omnifont and decorative characters, word recognition and noun compound translation. Finally, an international collaboration on the development of the required technology is proposed.

Describes a method for recognizing Japanese handwritten characters, including touching ones. The touching characters are segmented by cutting the connected components in a pre-segmentation process. The proposed segmentation method consists of the following steps: (1) identification of the touching components by checking the size of candidate character patterns; (2) estimation of the breakpoint and cutting of the components; and (3) generation of candidate character patterns by merging the components with adjacent patterns. The proposed method has the feature that the touching components can be identified even if they are comprised of relatively small patterns. The touching components are cut by using stroke width analysis and projection image analysis. This cut-and-merge procedure was able to segment 134 (88.2%) address lines correctly out of 152 Japanese address lines having touching portions, and could generate correct candidate character patterns.

A method for street number matching in Japanese address recognition is presented. This method uses the knowledge about the number expression formats, which is represented by the expression patterns of street numbers. The expression patterns are composed of character class symbols which represent numerals and delimiter characters. By matching the expression patterns against the character candidates in the lattice which is the result of character recognition, the class symbol lattice is generated. In the class symbol lattice, the numeral and delimiter character candidates are replaced by the corresponding class symbol. An experiment using 6,136 expression patterns showed that 93% of street numbers on the test mail-pieces were correctly recognized.

A method of character segmentation of Japanese handwritten characters has been developed. It is effective especially in character recognition with the over-segmentation process. The method is based on the credibility measurement of each presegmented pattern for being a true character by analyzing peripheral features such as gaps between patterns and widths and heights of patterns. A heuristic statistical method is applied to the analysis. Experimental results show that the developed method increases the accuracy of character segmentation. It is effective especially when no linguistic feedback to segmentation is available and/or the character classifier ability is not high enough. In such cases, the recognition accuracy is increased from 30% to 70% by the new segmentation method.

This paper investigates three fundamental techniques in OCR error correction for Japanese texts using morphological analysis: (1) an optimal method for candidate word extraction from a candidate character lattice, (2) optimal word entries for Japanese verb inflection analysis, and (3) a new method of word matching cost calculation which is more suitable to be used with linguistic criteria. Comparative evaluation shows that the combination of these techniques requires 84% less computation, captures 2.6% more candidate words, reduces the chart parsing computation by 20%, and attains 25% higher error correction rate than a commonly used method.

Sign language is one means of communication for hearing-impaired people. Words and sentences in sign language are mainly represented by hands' gestures. In this report, we show a sign language translation system which we are developing. The system translates Japanese sign language into Japanese and vice versa. In this system, hand shape and position data are inputted using DataGlove. Inputted hand motions are recognized and translated into Japanese sentences. Japanese text is translated into sign language represented as 3-D computer-graphic animation of sign language gestures.

Presents a new postprocessing method for recognizing handwritten Kanji names with Furigana, which shows the reading (pronunciation) of the Kanji in Kana characters, in real-time. Errors and rejects of Kanji OCR are automatically corrected to meaningful characters with the help of a Kanji-Furigana dictionary. Also, this method allows names to be written in any style. The method consists of Kanji word matching and Furigana verification. For Kanji word matching, Kanji name candidates are extracted. In Furigana verification, Kana candidate characters recognized by Kana OCR are compared with Furigana searched from the dictionary, using the Kanji name candidates. The correct name is selected from the results of Kanji and Kana word matching, plus the spacing between the first and second names. As result, 93.9% of the Kanji OCR's errors and rejects are corrected in real-time.>

A pattern-oriented segmentation method for optical character recognition that leads to document structure analysis is presented. As a first example, segmentation of handwritten numerals that touch are treated. Connected pattern components are extracted, and spatial interrelations between components are measured and grouped into meaningful character patterns. Stroke shapes are analyzed and a method of finding the touching positions that separates about 95% of connected numerals correctly is described. Ambiguities are handled by multiple hypotheses and verification by recognition. An extended form of pattern-oriented segmentation, tabular form recognition, is considered. Images of tabular forms are analyzed, and frames in the tabular structure are extracted. By identifying semantic relationships between label frames and data frames, information on the form can be properly recognized.>

An algorithm to normalize the skew of document images is proposed. The skew angle is detected in two stages. In the first stage, connected regions in an image are extracted and some feature parameters are extracted for each region. In the second stage, the Hough transform is calculated for the parameters, and the angle which gives the minimum of the transform is estimated as the skew angle. In experiments using CCITT standard documents, a detection accuracy of less than 0.1 degrees is obtained for printed documents. When graphical elements are included in the documents in addition to printed characters, the accuracy deteriorates to 0.2 degrees .>

A fast algorithm for component labeling of binary images is proposed. Component labeling is an important method for separation of objects in document image understanding, especially in character and picture extraction. The proposed algorithm is based on sorting and tracking of runs and label propagation to the connected runs. Each scan line is partitioned into small segments, i.e. blocks and runs on the scan line are sorted according to the block sequence to speed up tracking of runs. It was ascertained by experiment that the processing time is 16.7 times faster than that of the conventional pixel-based labeling method.>