- Research Article
7
- 10.1016/s1088-467x(97)00003-6
Machine Recognition of Hand-Printed Chinese Characters
- Jan 01, 1997
- Intelligent Data Analysis
- Adnan Amin + 1 more +1
Machine Recognition of Hand-Printed Chinese Characters
This paper proposes a new cognitive model, called "Segmentron", for Chinese character description and recognition. The cognitive model consists of three layers: (1) perceptual extraction for skeleton feature, (2) affine description in convex space, and (3) entropic evolutionary classification for optimization. With the model mentioned above, the correct classification result can be given and recognized characters can also be found efficiently. The model is practical for Chinese character description and recognition and can be applied in the computer system for Chinese character recognition.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Machine Recognition of Hand-Printed Chinese Characters
Machine Recognition of Hand-Printed Chinese Characters
Hybrid model for Chinese character recognition based on Tesseract-OCR
Optical character recognition (OCR) is an important way to input information into a computer. And text information can be extracted by OCR from an image. Currently, the accuracy rate of Chinese OCR can also be improved. This study proposes a hybrid Chinese character recognition model based on the characteristics of Chinese. Before the OCR engine works, the model first filters the interference information in the image. Then the model adjusts the aspect ratio of the character. After an image is identified by OCR, single character recognition result is obtained. Then the result is checked and corrected on the phrase level. The experimental results show that the hybrid model improves the accuracy rate of Chinese OCR. Through image processing, the correct rate of recognition by the Tesseract-OCR engine is increased by about 12%, and the natural language processing improves the accuracy of the recognition result by about 5%.
Read moreChapter 6 Automatic Activation of Linguistic Information in Chinese Character Recognition
Chapter 6 Automatic Activation of Linguistic Information in Chinese Character Recognition
Chinese character recognition : studies of complexity effect on recognition efficiency, spatial frequency characteristics, crowding and expertise
Chinese characters are used by about one-fifth of the world population. Each character can generally be analyzed and represented at three orthographic levels including single stroke, component and whole character. Regardless of the number of strokes a Chinese character has, it always occupies a square area. Such special spatial layout leads to a great variation in stroke density, i.e., visual complexity, among Chinese characters. Here in this thesis, the effects of visual complexity on Chinese character recognition were examined to investigate the visual processing of Chinese characters in both central and peripheral vision. \n \nEmpirical findings from four psychophysical studies will be reported. First, the efficiency of detecting and recognizing Chinese characters of different complexity levels was studied. Human recognition of a more complex Chinese character was found to be generally less efficient than that of a simpler one. The relationship between efficiency and complexity had a log-log slope of approximately -1, suggesting that the basic features for recognizing a single Chinese character might be less complex than the simplest object used, which is a single stroke. Analysis results of the detection and recognition efficiency were consistent with a two-stage processing model of Chinese character recognition—feature detection followed by feature integration. Results in peripheral vision suggested that the feature integration process was adversely influenced by internal crowding. \n \nSecond, the spatial frequency characteristics of recognizing Chinese characters of different complexity levels were examined. The results showed that the peak tuning frequency for recognizing a more complex Chinese character was higher and the bandwidth of the spatial tuning functions was narrower than for recognizing a simpler character. In addition, the effects of size on such spatial frequency characteristics depended on the complexity level. The comparison of human performance against the corresponding performance of a CSF-limited ideal observer model implied that the processing mechanism of recognizing simple and complex characters was different. Moreover, similar results were observed among native Chinese and non-Chinese readers, suggesting that expertise might not influence front end processing properties such as the spatial frequency characteristics of Chinese character recognition. \n \nThird, the importance of utilizing spatial frequency information of different orientations (i.e., horizontal versus vertical) in Chinese character recognition was investigated. Results showed higher efficiency in utilizing vertical than horizontal stroke information for Chinese character recognition among native Chinese readers whereas non-Chinese readers used both vertical and horizontal stroke information similarly. \nFourth, the effects of the target-flanker similarity in complexity on crowding in Chinese character recognition were examined. The results suggested that the behavior of feature integration process in peripheral vision might be consistent with summation theory, which stresses on the relationship between the processing capacity of neurons and the processing demand of stimuli. The “excessive feature integration explanation” of crowding was also discussed in the context of the current findings. \n \nBuilding on previous research in the literature, results from the four studies in this thesis were synthesized in a proposal of a visual information processing model for Chinese character recognition.
Read moreChinese children's character recognition: Visuo‐orthographic, phonological processing and morphological skills
Tasks tapping visual skills, orthographic knowledge, phonological awareness, speeded naming, morphological awareness and Chinese character recognition were administered to 184 kindergarteners and 273 primary school students from Beijing. Regression analyses indicated that only syllable deletion, morphological construction and speeded number naming were unique correlates of Chinese character recognition in kindergarteners. Among primary school children, the independent correlates of character recognition were rime detection, homophone judgement, morpheme production, orthographic knowledge and speeded number naming. Results underscore the importance of some dimensions of both phonological processing and morphological awareness for both very early and intermediate Chinese reading acquisition. Although significantly correlated with character recognition in younger (but not older) children, visual skills were not uniquely associated with Chinese character reading at any grade level. However, orthographic skills were strongly associated with reading in primary school but not kindergarten, suggesting that orthographic skills are more important for literacy development as reading experience increases.
Read moreDesign and implementation of Handwritten Character Recognition System Based on Deep Learning
With the development of technology, it is more and more common to store data in electronic form. In order to facilitate the conversion of paper forms into electronic information storage, this project has implemented a handwritten Chinese character recognition system for table pictures, which reduces the pressure of manual input of handwritten Chinese character information. This project preprocesses the scanned image, extracts the handwritten text information from the image, and cuts the handwritten Chinese characters. In order to improve the accuracy of single character cutting, this project uses the combination of vertical projection and single character length width ratio to determine the cutting position. ResNext50 model was used for model training, and the two models were trained to recognize numbers, letters and handwritten Chinese characters respectively. The accuracy of the Chinese character recognition model is more than 90%, and the accuracy of the number and letter recognition model is 98%. According to the input content, the proper noun list is used to correct the recognition results and improve the accuracy of recognition. Find the specific noun with the highest similarity by calculating Levenshtein distance. The method proposed in this paper effectively completes the separation and recognition of handwritten Chinese characters in table images.
Read moreHandwritten Chinese character recognition
The recognition of Chinese characters has been an area of great interest for many years, and a large number of research papers and reports have already been published in this area. There are several major problems with Chinese character recognition: Chinese characters are distinct and ideographic, the character size is very large and many structurally similar characters exist in the character set. Thus, classification criteria are difficult to generate. This article presents a technique for the recognition of hand-printed Chinese characters using backpropagation algorithm. The article also discusses Chinese character recognition using preprocessing and projection and zoning feature extraction. The system was tested with 100 characters (each character has 160 samples), and it needs 7 output units from the network.
Read moreAdaptive recognition of Chinese characters: imitation of psychological process in machine recognition
This paper is focused on imitation of human psychological process in machine recognition of Chinese characters. Some results of research on human Chinese character recognition have been discussed and unified into a compound mechanism with an adaptive and self-developing nature. A machine imitation model has been proposed for Chinese character recognition with different routines. By some simplification but with the crucial feature of the model being retained, an experimental system for handprinted Chinese character recognition based on the novel concept has been built. Experimental results have shown that the associated routines continuously improve their performance during their work even after supervised training is halted. The routine of the global pattern approach eventually learns most of the classes and the recognition process gradually shifts from the subpattern approach to the global pattern approach.
Read moreThe characteristics of event related potential in the cognitive course of individual Chinese characters
The difference between the Chinese and English character recognition process and the characteristics of Chinese character recognition process are investigated by analyzing the ERP difference between the matched and mismatched ending strokes of a single Chinese character. First, the P300 are observed in the mismatched ERP waveforms, which don't exist in the matched ERP waveforms (P<0.05). These phenomena demonstrate that the recognition of the pictograph Chinese characters are not only affected by the semanteme, but also closely related to its image of the figure. Secondly, the amplitude of the N400 in the mismatched ending strokes of Chinese character is obviously higher than that in the matched groups (P<0.05). The N400 is distributed in the other areas except the right side of temporal lobe and is higher in the left side. However, the amplitude of N400 in the waveforms has no obvious distinction between the cerebral hemispheres, where the amplitude of N400 are obtained by distracting the ERP of the mismatched individual Chinese character by the ERP of the matched group. These results show that the difference between cerebral hemispheres is attributed to the formal and phonic processing, not due to the semantic processing, which verify that the left and right cerebral hemisphere are both used in the Chinese character recognition, which is very different from the English character recognition, where the left side cerebral hemisphere is used with high percentage.
Read moreCharacter recognition with neural networks
Character recognition with neural networks
An improved method on Chinese character recognition
Genetic algorithm (GA) is an effective method to optimize system parameters. Chinese character recognition post-processing (CCRP) based on GA gets good result, but the better results are obtained after the method is improved. GA relies on the Chinese character itself and context information. In improved method, the parameters of single character recognition (SCR) model are conditional probabilities that can be converted into posterior probabilities by theoretic analysis. On the basis of SCR output analysis, posterior probabilities of candidates are obtained by statistical method. Experiment shows that post-processing performance depends on efficient evaluation of posterior probability, besides proper method. The character recognition average rate of the tested texts is from 96.93% to 97.98%, from 96.93% to 98.47% after improvement, and improves 0.49%.
Read moreFrom Unit to Gestalt: Perceptual Dynamics in Recognizing Chinese Characters
From Unit to Gestalt: Perceptual Dynamics in Recognizing Chinese Characters
Chinese character recognition via orthogonal moments
To select a suitable feature vector extracted from the interested character for the purpose of classification is essential in the design of a character recognition system. Moment descriptors have been developed as features in pattern recognition since Hu[14] first introduced the moment method. Describing a character with moments means that global properties of the character are used rather than local properties. This nature makes the method of moments a proper candidate for Chinese character recognition system. In this paper, new Legendre moment spaces for Chinese character recognition are proposed which provide significant improvements in terms of Chinese character recognition.
Read moreOptical Chinese character recognition system using a new pipelined matching and sorting very large scale integration
A VLSI implementation of the optical Chinese character recognition (OCCR) system with pipelined and parallel structure is presented. We also propose an efficient method for performing block classification and character segmentation as well as an effective and adaptive feature extraction algorithm for recognizing multifont printed Chinese characters. With the complex and huge amount of data involved in Chinese characters, their recognition requires numerous complex computations. Therefore, to improve the recognition efficiency for practical applications, a VLSI chip is designed and fabricated. To preserve a certain degree of flexibility so that various recognition algorithms can be implemented with the system, only the most time-consumig parts are implemented into the VLSI circuit. By combining the VLSI technology and the effective Chinese character recognition algorithm, a practical OCCR system with high speed, high-recognition rate, and accumulated learning capability is developed. Based on the experimental results, the VLSI chip can process up to 200 characters/s, which is one hundred times faster than the original software algorithm. The recognition rates of three different test conditions are also given.
Read moreEfficient mobile phone Chinese optical character recognition systems by use of heuristic fuzzy rules and bigram Markov language models
Efficient mobile phone Chinese optical character recognition systems by use of heuristic fuzzy rules and bigram Markov language models
Read more