A novel comprehensive database for offline Persian handwriting recognition

Sadri, Javad; Yeganehzad, Mohammad Reza; Saghi, Javad

doi:10.1016/j.patcog.2016.03.024

Cited by 23 publications

(29 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the proposed CNN model needs 224×224 images as the input size in the first layer. There are several methods for image resizing [14]. The simplest one is the resizing of all images to 224×224, but applying this method on all images with different sizes may deform structure of handwritten images.…”

Section: Preprocessingmentioning

confidence: 99%

See 1 more Smart Citation

Holistic Persian Handwritten Word Recognition Using Convolutional Neural Network

Zohrevand

Imani

2021

IJE

View full text Add to dashboard Cite

Due to the cursive-ness and high variability of Persian script, the segmentation of handwritten words into sub-words is still a challenging task. These issues could be addressed in a holistic approach by sidestepping segmentation at the character level. In this paper, an end-to-end holistic method based on deep convolutional neural network is proposed to recognize off-line Persian handwritten words. The proposed model uses only five convolutional layers and two fully connected layers for classifying word images effectively, which can lead to a substantial reduction in the required parameters. The effect of various pooling strategies is also investigated in this paper. The primary goal of this article is to ignore handcrafted feature extraction and to attain a generalized and stable word recognition system. The presented model is assessed using two famous handwritten Persian word databases called Sadri and IRANSHAHR. The recognition accuracies were obtained at 98.6% and 94.6%, on Sadri and IRANSHAHR datasets respectively, and outperformed the state-of-the-art methods.

show abstract

Section: Preprocessingmentioning

confidence: 99%

“…• Analysing the proposed method on two popular Persian handwritten word datasets called IRANSHAHR [13] and Sadri [14]. • Analysing errors on Sadri dataset for the first time.…”

Section: Introductionmentioning

confidence: 99%

Holistic Persian Handwritten Word Recognition Using Convolutional Neural Network

Zohrevand

Imani

2021

IJE

View full text Add to dashboard Cite

show abstract

“…Since the approach is based on transfer learning, we will illustrate the source and target domains, and our transfer learning strategies in the following lines.  Source Domain: Persian Handwriting Dataset: We used a Persian handwriting dataset [17] as a source domain. This dataset consists of 115 words, where each was written and labeled differently for more than 500 different writers.…”

Section: A Feature Generation Using Transfer Learningmentioning

confidence: 99%

Learning Representations from Persian Handwriting for Offline Signature Verification, a Deep Transfer Learning Approach

Mersa

Etaati

Masoudnia

et al. 2019

2019 4th International Conference on Pattern Recognition and Image Analysis (IPRIA)

View full text Add to dashboard Cite

Offline Signature Verification (OSV) is a challenging pattern recognition task, especially when it is expected to generalize well on the skilled forgeries that are not available during the training. Its challenges also include small training sample and large intra-class variations. Considering the limitations, we suggest a novel transfer learning approach from Persian handwriting domain to multi-language OSV domain. We train two Residual CNNs on the source domain separately based on two different tasks of word classification and writer identification. Since identifying a person's signature resembles identifying one's handwriting, it seems perfectly convenient to use handwriting for the feature learning phase. The learned representation on the more varied and plentiful handwriting dataset can compensate for the lack of training data in the original task, i.e. OSV, without sacrificing the generalizability. Our proposed OSV system includes two steps: learning representation and verification of the input signature. For the first step, the signature images are fed into the trained Residual CNNs. The output representations are then used to train SVMs for the verification. We test our OSV system on three different signature datasets, including MCYT (a Spanish signature dataset), UTSig (a Persian one) and GPDS-Synthetic (an artificial dataset). On UT-SIG, we achieved 9.80% Equal Error Rate (EER) which showed substantial improvement over the best EER in the literature, 17.45%. Our proposed method surpassed state-of-the-arts by 6% on GPDS-Synthetic, achieving 6.81%. On MCYT, EER of 3.98% was obtained which is comparable to the best previously reported results.

show abstract

“…• restricting the potential uncontrolled propagation and reducing the resulting electromagnetic emissions (spatial separation, screening, filtration) [17]; • appropriately forming the sources of those emissions (i.e., by reducing their effectiveness) [2,[18][19][20]. In the first case, the levels of electromagnetic emissions are reduced to levels where any attempt to retrieve data from recorded emissions and present it in a graphical form (images) fails [21,22]. Forming a source of compromising emanations without having a significant impact on their absolute values results in losing their distinctive features (correlation with processed information), and thus, the data retrieved from the recorded compromising emanations and presented in graphical form remains illegible and useless (e.g., safe font) [23,24].…”

Section: Introductionmentioning

confidence: 99%

“…In the first case, the levels of electromagnetic emissions are reduced to levels where any attempt to retrieve data from recorded emissions and present it in a graphical form (images) fails [21,22]. Forming a source of compromising emanations without having a significant impact on their absolute values results in losing their distinctive features (correlation with processed information), and thus, the data retrieved from the recorded compromising emanations and presented in graphical form remains illegible and useless (e.g., safe font) [23,24].…”

Section: Introductionmentioning

confidence: 99%

The Effectiveness of TMDS Coding in Counteracting the Non-Invasive Acquisition of Graphic Data

Kubiak

Przybysz

2019

Electronics

View full text Add to dashboard Cite

Modern video display terminals commonly use digital video signals. Transition minimized differential signaling (TMDS) coding implemented in video signal transmission using DVI (Digital Visual Interface) standard is commonly used. The aim of the coding scheme adopted by this solution is to eliminate the constant component of the electrical signal, increase the resistance to electromagnetic (EM) interference, and reduce electronic interference between cables. Professionals and hobbyists interested in the problems relating to protecting information against electromagnetic infiltration believe that TMDS coding, in contrast to the VGA (Video Graphics Array) analogue standard, significantly improves the electromagnetic security of processed graphic information. This paper shows a comparison of the abovementioned standards in terms of information protection against electromagnetic infiltration. The paper presents the results of computer simulations and studies dealing with practical compromising emanations for DVI standard and its susceptibility to electromagnetic radiation spying. The obtained results show that the commonly expressed ideas of digital standards being fully secure are false. The obtained test results show that the level of electromagnetic protection can be increased by using appropriate pairs of colors for the text and background. This solution has to be connected with a mode that smooths the edges of graphic signs. Then, the number of frequencies in which valuable emissions exist can be limited. In this paper, pairs of colors for which the level of protection of information can be increased are shown. The authors present their analyses on the basis of the method of colors. The method is connected to possibilities of selection of smoothing modes of edges. As Windows is the most commonly used system in classified work stations (so-called TEMPEST computers), this operating system was considered from the viewpoint of the protection of processed information.

show abstract

A novel comprehensive database for offline Persian handwriting recognition

Cited by 23 publications

References 25 publications

Holistic Persian Handwritten Word Recognition Using Convolutional Neural Network

Holistic Persian Handwritten Word Recognition Using Convolutional Neural Network

Learning Representations from Persian Handwriting for Offline Signature Verification, a Deep Transfer Learning Approach

The Effectiveness of TMDS Coding in Counteracting the Non-Invasive Acquisition of Graphic Data

Contact Info

Product

Resources

About