scispace - formally typeset
Search or ask a question
Proceedings ArticleDOI

Optical character recognition using artificial neural network

01 Jan 2017-pp 1-4
TL;DR: In this research, characters are recognized even when noise such as inclination and skewedness presents, by training the network to look for discrepancies in data and relate them using vocabulary, grammar and common recurrences that may occur after a character.
Abstract: The objective of this work is to convert printed text or handwritten characters recorded offline using either scanning equipment or cameras into a machine-usable text by simulating a neural network so that it would improve the process of collecting and storing data by human workers. Another goal is to provide an alternate, better and faster algorithm with higher accuracy to recognize the characters. In this context, we choose artificial neural network and make it much more tolerant to anomalies in the recorded image or data. Common optical character recognition tasks involve identifying simple edge detection and matching them with predefined patterns. In this research, characters are recognized even when noise such as inclination and skewedness presents, by training the network to look for discrepancies in data and relate them using vocabulary, grammar and common recurrences that may occur after a character. Images are also masked in multiple ways and processed individually to increase the confidence level of prediction.
Citations
More filters
01 Jan 2000
TL;DR: This article showed that the number of higher frequency neighbors is inhibitory in reading and also examined the influence of orthographic structure in form and repetition-priming effects, which again suggests that orthographic neighbors seem to play an inhibitory role in the selection process.
Abstract: This paper reviews recent research on the effects of “orthographic neighbors” (i.e., words that can be created by changing one letter of the stimulus item, preserving letter positions, see Coltheart et al., 1977) on reading and laboratory word identification tasks. We begin this paper with a literature review on the two basic “neighborhood” effects (neighborhood size and neighborhood frequency). This review shows that the number of higher frequency neighbors is inhibitory in reading. We also examine the influence of orthographic structure in form - and repetition-priming effects, which again suggests that orthographic neighbors seem to play an inhibitory role in the selection process. Finally, we discuss the empirical evidence in the context of current models of visual word recognition and reading.

56 citations

Proceedings ArticleDOI
01 Nov 2018
TL;DR: The purpose of this paper is to compare the algorithm of Artificial Neural Network (ANN) and Gaussian Naïve Bayes (GNB) in handwriting number recognition, and shows that GNB produces a higher level of accuracy compared to the ANN.
Abstract: Current technological developments spur the application of pattern recognition in various fields, such as the introduction of signature patterns, fingerprints, faces, and handwriting. Human handwriting has differences between one another and often is difficult to read or difficult to recognize and this can hamper daily activities, such as transaction activities that require handwriting. Even though one of the human biometric features is handwriting. The purpose of this paper is to compare the algorithm of Artificial Neural Network (ANN) and Gaussian Naive Bayes (GNB) in handwriting number recognition. Both of these algorithms are quite reliable in performing the classification process. ANN can do pattern recognition and provide good results. If the size of the training data is small, the accuracy of GNB provides good results. To recognize the handwriting pattern, the characteristics of the handwriting object are extracted using an invariant moment. The test results show that GNB produces a higher level of accuracy of 28.33% compared to the ANN of 11.67%. The resulting accuracy level is still very low. This is because the result extraction data has a small distance for each class or any number character.

4 citations

Book ChapterDOI
29 Oct 2018
TL;DR: This paper presents a survey of keylogger and screenlogger attacks to increase the understanding and awareness of their threat by covering basic concepts related to bank information systems and explaining their functioning, as it presents and discusses an extensive set of plausible countermeasures.
Abstract: Keyloggers and screenloggers are one of the active growing threats to user’s confidentiality as they can run in user-space, easily be distributed and upload information to remote servers. They use a wide number of different techniques and may be implemented in many ways. Keyloggers and screenloggers are very largely diverted from their primary and legitimate function to be exploited for malicious purposes compromising the privacy of users, and bank customers notably. This paper presents a survey of keylogger and screenlogger attacks to increase the understanding and awareness of their threat by covering basic concepts related to bank information systems and explaining their functioning, as it presents and discusses an extensive set of plausible countermeasures.

4 citations

Journal ArticleDOI
TL;DR: In this paper, the authors proposed a low power energy-efficient accelerator with real-time capabilities called iDocChip, which is a configurable hybrid hardware-software programmable system-on-chip (SoC) based anyOCR for digitizing historical documents.
Abstract: In recent years, $$\hbox {optical character recognition (OCR)}$$ systems have been used to digitally preserve historical archives. To transcribe historical archives into a machine-readable form, first, the documents are scanned, then an $$\hbox {OCR}$$ is applied. In order to digitize documents without the need to remove them from where they are archived, it is valuable to have a portable device that combines scanning and $$\hbox {OCR}$$ capabilities. Nowadays, there exist many commercial and open-source document digitization techniques, which are optimized for contemporary documents. However, they fail to give sufficient text recognition accuracy for transcribing historical documents due to the severe quality degradation of such documents. On the contrary, the anyOCR system, which is designed to mainly digitize historical documents, provides high accuracy. However, this comes at a cost of high computational complexity resulting in long runtime and high power consumption. To tackle these challenges, we propose a low power energy-efficient accelerator with real-time capabilities called iDocChip, which is a configurable hybrid hardware-software programmable $$\hbox {System-on-Chip (SoC)}$$ based on anyOCR for digitizing historical documents. In this paper, we focus on one of the most crucial processing steps in the anyOCR system: Text and Image Segmentation, which makes use of a multi-resolution morphology-based algorithm. Moreover, an optimized $$\hbox {FPGA}$$ -based hybrid architecture of this anyOCR step along with its optimized software implementations are presented. We demonstrate our results on multiple embedded and general-purpose platforms with respect to runtime and power consumption. The resulting hardware accelerator outperforms the existing anyOCR by 6.2 $$\times$$ , while achieving 207 $$\times$$ higher energy-efficiency and maintaining its high accuracy.

3 citations

Proceedings ArticleDOI
01 Jan 2019
TL;DR: A system for transforming the text inscribed in Konkani dialect into Speech by utilizing artificial neural network is demonstrated, which can be upgraded to work with letters written in different styles.
Abstract: This paper demonstrates a system for transforming the text inscribed in Konkani dialect into Speech by utilizing artificial neural network. Many visually challenged individuals use Text to Speech framework as a tool for communication. The ability to convert text to voice lessens the reliance, dissatisfaction, and feeling of defenselessness of these individuals. India is called as the land of unity and diversity, there are 22 official languages. TTS frameworks are mostly accessible in English; in any case, it has been watched that individuals feel more comfortable in hearing their own native dialect. Handwritten optical character recognition is the most challenging research zone, because of its intricacy in segmenting the character that grows on account of Devnagari Script because of Modifiers and compound characters. The document comprising Konkani text is scanned and fed to the system. In this framework the character recognition is done by utilizing Neural Network, in this manner the structure can be upgraded to work with letters written in different styles. After the characters in the Documents are viably recognized by neural network, it is composed to a text document, the entered text document is analyzed, the syllabification is accomplished in view of the phonological guidelines and the syllables are secured autonomously. At that point the syllable coordinating speech file is linked and the silence existing in the linked discourse is confined. breaks within the discourse are removed at syllable limits without diminishing the superiority of speech.

2 citations

References
More filters
Proceedings ArticleDOI
18 Jun 2003
TL;DR: This work presents an algorithm for matching handwritten words in noisy historical documents that performs better and is faster than competing matching techniques and presents experimental results on two different data sets from the George Washington collection.
Abstract: Libraries and other institutions are interested in providing access to scanned versions of their large collections of handwritten historical manuscripts on electronic media. Convenient access to a collection requires an index, which is manually created at great labor and expense. Since current handwriting recognizers do not perform well on historical documents, a technique called word spotting has been developed: clusters with occurrences of the same word in a collection are established using image matching. By annotating "interesting" clusters, an index can be built automatically. We present an algorithm for matching handwritten words in noisy historical documents. The segmented word images are preprocessed to create sets of 1-dimensional features, which are then compared using dynamic time warping. We present experimental results on two different data sets from the George Washington collection. Our experiments show that this algorithm performs better and is faster than competing matching techniques.

626 citations

Journal ArticleDOI
TL;DR: It is shown in a subset of the George Washington collection that such a word spotting technique can outperform a Hidden Markov Model word-based recognition technique in terms of word error rates.
Abstract: Searching and indexing historical handwritten collections are a very challenging problem. We describe an approach called word spotting which involves grouping word images into clusters of similar words by using image matching to find similarity. By annotating “interesting” clusters, an index that links words to the locations where they occur can be built automatically. Image similarities computed using a number of different techniques including dynamic time warping are compared. The word similarities are then used for clustering using both K-means and agglomerative clustering techniques. It is shown in a subset of the George Washington collection that such a word spotting technique can outperform a Hidden Markov Model word-based recognition technique in terms of word error rates.

368 citations

Journal ArticleDOI
TL;DR: The effect of word length (number of letters in a word) on lexical decision was reexamined using the English Lexicon Project and an unexpected pattern of results taking the form of a U-shaped curve was revealed.
Abstract: In the present study, we reexamined the effect of word length (number of letters in a word) on lexical decision. Using the English Lexicon Project, which is based on a large data set of over 40,481 words (Balota et al., 2002), we performed simultaneous multiple regression analyses on a selection of 33,006 English words (ranging from 3 to 13 letters in length). Our analyses revealed an unexpected pattern of results taking the form of a U-shaped curve. The effect of number of letters was facilitatory for words of 3–5 letters, null for words of 5–8 letters, and inhibitory for words of 8–13 letters. We also showed that printed frequency, number of syllables, and number of orthographic neighbors all made independent contributions. The length effects were replicated in a new analysis of a subset of 3,833 monomorphemic nouns (ranging from 3 to 10 letters), and also in another analysis based on 12,987 bisyllabic items (ranging from 3 to 9 letters). These effects were independent of printed frequency, number of syllables, and number of orthographic neighbors. Furthermore, we also observed robust linear inhibitory effects of number of syllables. Implications for models of visual word recognition are discussed.

286 citations


"Optical character recognition using..." refers background in this paper

  • ...The effect of word length on lexical decision is being examined by using the English lexicon project [9]....

    [...]

Journal ArticleDOI
TL;DR: An off-line handwritten alphabetical character recognition system using multilayer feed forward neural network that will be suitable for converting handwritten documents into structural text form and recognizing handwritten names is described in the paper.
Abstract: An off-line handwritten alphabetical character recognition system using multilayer feed forward neural network is described in the paper. A new method, called, diagonal based feature extraction is introduced for extracting the features of the handwritten alphabets. Fifty data sets, each containing 26 alphabets written by various people, are used for training the neural network and 570 different handwritten alphabetical characters are used for testing. The proposed recognition system performs quite well yielding higher levels of recognition accuracy compared to the systems employing the conventional horizontal and vertical methods of feature extraction. This system will be suitable for converting handwritten documents into structural text form and recognizing handwritten names.

135 citations

Proceedings Article
01 Dec 2009
TL;DR: SparseDTW as discussed by the authors exploits the existence of similarity and/or correlation between the time series to compute the dynamic time warping distance between two time series that always yields the optimal result.
Abstract: We present a new space-efficient approach, (SparseDTW), to compute the Dynamic Time Warping (DTW) distance between two time series that always yields the optimal result. This is in contrast to other known approaches which typically sacrifice optimality to attain space efficiency. The main idea behind our approach is to dynamically exploit the existence of similarity and/or correlation between the time series. The more the similarity between the time series the less space required to compute the DTW between them. To the best of our knowledge, all other techniques to speedup DTW, impose apriori constraints and do not exploit similarity characteristics that may be present in the data. We conduct experiments and demonstrate that SparseDTW outperforms previous approaches.

131 citations