Hybrid modeling of an OffLine Arabic Handwriting Recognition System AHRS

Meddeb, Ons; Aljawarneh, Shadi

doi:10.1109/icemis.2016.7745319

Cited by 20 publications

(7 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[ 14 , 94 ]. One or more pre-processing techniques can be used based on the degree of the text image quality [ 1 , 95 ] and according to the targeted OAHR design.…”

Section: Oahr General Frameworkmentioning

confidence: 99%

“…Feature extraction is the process of extracting the most relevant and non-redundant attributes from the text image raw data [ 1 , 3 , 14 , 92 ] and then converting them into a vector of features [ 9 , 91 , 95 ]. The different types of feature methods are classified as structural features, statistical features, and feature space transformations [ 8 , 94 , 95 ]. Structural features describe the geometrical and topological characteristics of a text image pattern, such as directions, strokes, endpoints, the intersection of line segments, number of diacritical marks, loops, types of dots, zigzag, arcs, concavities, etc.…”

Section: Oahr General Frameworkmentioning

confidence: 99%

“…Structural features describe the geometrical and topological characteristics of a text image pattern, such as directions, strokes, endpoints, the intersection of line segments, number of diacritical marks, loops, types of dots, zigzag, arcs, concavities, etc. Statistical features are extracted from the pixels’ statistical distribution, and they describe the set of characteristic measurements of a text image pattern, such as number of loops, segments, branching and crossing points, and diacritics and their positions [ 95 ], and they have low complexity and high speed [ 1 , 3 , 8 , 94 ]. Feature space transformations convert the pixel representation to a more compact form with less feature vector dimensionality [ 7 , 8 , 94 ] among the methods involving principal component analysis (PCA), Walsh Hadamard Transform, Fourier Transform, Wavelets, Hough Transform, Gabor Transform, Rapid Transform, and Karhunen Loeve Expression and moments [ 1 , 3 , 8 , 9 , 94 , 95 ].…”

Section: Oahr General Frameworkmentioning

confidence: 99%

“…The classification phase is the decision-making process of the recognition system, and its role relies on recognizing and allocating an input feature with a class label or membership scores for digits, characters, or words to the correct related defined classes [ 1 , 3 , 92 ], so that the texts in images are transformed into a computer understandable form [ 92 ], and the performance of a classifier depends on the quality of the extracted features. In the literature, the categories of numerous recognition approaches, including template matching [ 92 ] and the structural, statistical, stochastic, and hybrid approaches, are obtained by combining multiple classifiers [ 95 ]. For Arabic handwritten recognition, many classifiers have been used, such as k-nearest neighbors (kNN), HMM, SVM, and ANN [ 1 , 3 , 93 , 94 , 95 , 96 ].…”

Section: Oahr General Frameworkmentioning

confidence: 99%

“…In the literature, the categories of numerous recognition approaches, including template matching [ 92 ] and the structural, statistical, stochastic, and hybrid approaches, are obtained by combining multiple classifiers [ 95 ]. For Arabic handwritten recognition, many classifiers have been used, such as k-nearest neighbors (kNN), HMM, SVM, and ANN [ 1 , 3 , 93 , 94 , 95 , 96 ]. In recognition systems, this phase is achieved by selecting a suitable recognition method and then employing two processes: the training process, which uses extracted features to train the classifier to build the appropriate models, and the testing process, which uses the previously generated models.…”

Section: Oahr General Frameworkmentioning

confidence: 99%

See 4 more Smart Citations

Novel Deep Convolutional Neural Network-Based Contextual Recognition of Arabic Handwritten Scripts

Ahmed

Gogate

Tahir

et al. 2021

Entropy

View full text Add to dashboard Cite

Offline Arabic Handwriting Recognition (OAHR) has recently become instrumental in the areas of pattern recognition and image processing due to its application in several fields, such as office automation and document processing. However, OAHR continues to face several challenges, including the high variability of the Arabic script and its intrinsic characteristics such as cursiveness, ligatures, and diacritics, the unlimited variation in human handwriting, and the lack of large public databases. In this paper, we have introduced a novel context-aware model based on deep neural networks to address the challenges of recognizing offline handwritten Arabic text, including isolated digits, characters, and words. Specifically, we have proposed a supervised Convolutional Neural Network (CNN) model that contextually extracts optimal features and employs batch normalization and dropout regularization parameters to prevent overfitting and further enhance its generalization performance when compared to conventional deep learning models. We employed numerous deep stacked-convolutional layers to design the proposed Deep CNN (DCNN) architecture. The proposed model was extensively evaluated, and it was observed to achieve excellent classification accuracy when compared to the existing state-of-the-art OAHR approaches on a diverse set of six benchmark databases, including MADBase (Digits), CMATERDB (Digits), HACDB (Characters), SUST-ALT (Digits), SUST-ALT (Characters), and SUST-ALT (Names). Further comparative experiments were conducted on the respective databases using the pre-trained VGGNet-19 and Mobile-Net models; additionally, generalization capabilities experiments on another language database (i.e., MNIST English Digits) were conducted, which showed the superiority of the proposed DCNN model.

show abstract

“…[ 14 , 94 ]. One or more pre-processing techniques can be used based on the degree of the text image quality [ 1 , 95 ] and according to the targeted OAHR design.…”

Section: Oahr General Frameworkmentioning

confidence: 99%