Saeeda Naz

Evaluation of cursive and non-cursive scripts using recurrent neural networks

Abstract Character recognition has been widely used since its inception in applications involved processing of scanned or camera-captured documents. There exist multiple scripts in which the languages are written. The scripts could broadly be divided into cursive and non-cursive scripts. The recurrent neural networks have been proved to obtain state-of-the-art results for optical character recognition. We present a thorough investigation of the performance of recurrent neural network (RNN) for cursive and non-cursive scripts. We employ bidirectional long short-term memory (BLSTM) networks, which is a variant of the standard RNN. The output layer of the architecture used to carry out our investigation is a special layer called connectionist temporal classification (CTC) which does the sequence alignment. The CTC layer takes as an input the activations of LSTM and aligns the target labels with the inputs. The results were obtained at the character level for both cursive Urdu and non-cursive English scripts are significant and suggest that the BLSTM technique is potentially more useful than the existing OCR algorithms.

Offline cursive Urdu-Nastaliq script recognition using multidimensional recurrent neural networks

Optical Character Recognition of cursive scripts remains a challenging task due to a large number of character shapes, inter- and intra-word overlaps, context sensitivity and diagonality of text. This paper presents an implicit segmentation based recognition system for Urdu text lines in Nastaliq script. The proposed technique relies on sliding overlapped windows on lines of text and extracting a set of statistical features. The extracted features are fed to a multi-dimensional long short term memory recurrent neural network (MDLSTM RNN) with a connectionist temporal classification (CTC) output layer that labels the character sequences. Experimental study of the proposed technique is carried out on the standard Urdu Printed Text-line Images (UPTI) database which comprises 10,000 text lines in Nastaliq font. Evaluations under different experimental settings realize promising recognition rates with a highest character recognition rate of 96.40%.

Segmentation techniques for recognition of Arabic-like scripts: A comprehensive survey

Arabic script based text recognition system has been a popular field of research for many years that can be used in the learning and teaching process to the students and educators how to read and understand educational contents of Arabic script. The challenging nature of Arabic script recognition has attracted the attention of researchers from both industry and academic circles but these efforts have not achieved good results until now. Segmentation of Urdu script when written in Nasta’liq writing style is very difficult task due to the complexity of writing style as compare to Naskh writing style. Good segmentation is one of the reasons for high accuracy. Character segmentation has been a critical phase of the OCR process. The higher recognition rates for isolated characters as compare to results of words or connected character well illustrate the importance of segmentation. Current study investigates the recent work for character segmentation and challenges for segmentation for Arabic script based languages.

Challenges in Baseline Detection of Arabic Script Based Languages

In this chapter, we present baseline detection challenges for Arabic script based languages and targeted Nastaliq and Naskh writing style. Baseline is an important step in the OCR as it directly affects the rest of the steps and increases the performance and efficiency of character segmentation and feature extraction in OCR process. Character recognition on Arabic script is relatively more difficult than Latin text due to the nature of Arabic script, which is cursive, context sensitive and different writing style. In this paper, we provide a comprehensive review of baseline detection methods for Urdu language. The aim of the chapter is to introduce the challenges during baseline detection in cursive script languages for Nastaliq and Naskh script.

Efficient leukocyte segmentation and recognition in peripheral blood image

BACKGROUND Blood cell count, also known as differential count of various types of blood cells, provides valuable information in order to assess variety of diseases like AIDS, leukemia and blood cancer. Manual techniques are still used in diseases diagnosis that is very lingering and tedious process. However, machine based automatic analysis of leukocyte is a powerful tool that could reduce the human errors, improve the accuracy, and minimize the required time for blood cell analysis. However, leukocyte segmentation is a challenging process due to the complexity of the blood cell image; therefore, this task remains unresolved issue in the blood cell segmentation. OBJECTIVE The aim of this work is to develop an efficient leukocyte cell segmentation and classification system. METHODS This paper presents an efficient strategy to segment cell images. This has been achieved by using Wiener filter along with Curvelet transform for image enhancement and noise elimination in order to elude false edges. We have also used combination of entropy filter, thresholding and mathematical morphology for obtaining image segmentation and boundary detection, whereas we have used back-propagation neural network for leukocyte classification into its sub classes. RESULTS As a result, the generated segmentation results are fruitful in a sense that we have overcome the problem of overlapping cells. We have obtained 100%, 96.15%, 92.30%, 92.30% and 96.15% accuracy for basophil, eosinophil, monocyte, lymphocyte and neutrophil respectively.

Urdu Nastaliq recognition using convolutionalrecursive deep learning

Recent developments in recognition of cursive scripts rely on implicit feature extraction methods that provide better results as compared to traditional hand-crafted feature extraction approaches. We present a hybrid approach based on explicit feature extraction by combining convolutional and recursive neural networks for feature learning and classification of cursive Urdu Nastaliq script. The first layer extracts low-level translational invariant features using Convolutional Neural Networks (CNN) which are then forwarded to Multi-dimensional Long Short-Term Memory Neural Networks (MDLSTM) for contextual feature extraction and learning. Experiments are carried out on the publicly available Urdu Printed Text-line Image (UPTI) dataset using the proposed hierarchical combination of CNN and MDLSTM. A recognition rate of up to 98.12% for 44-classes is achieved outperforming the state-of-the-art results on the UPTI dataset.

Robust Optical Recognition of Cursive Pashto Script Using Scale, Rotation and Location Invariant Approach.

The presence of a large number of unique shapes called ligatures in cursive languages, along with variations due to scaling, orientation and location provides one of the most challenging pattern recognition problems. Recognition of the large number of ligatures is often a complicated task in oriental languages such as Pashto, Urdu, Persian and Arabic. Research on cursive script recognition often ignores the fact that scaling, orientation, location and font variations are common in printed cursive text. Therefore, these variations are not included in image databases and in experimental evaluations. This research uncovers challenges faced by Arabic cursive script recognition in a holistic framework by considering Pashto as a test case, because Pashto language has larger alphabet set than Arabic, Persian and Urdu. A database containing 8000 images of 1000 unique ligatures having scaling, orientation and location variations is introduced. In this article, a feature space based on scale invariant feature transform (SIFT) along with a segmentation framework has been proposed for overcoming the above mentioned challenges. The experimental results show a significantly improved performance of proposed scheme over traditional feature extraction techniques such as principal component analysis (PCA).

Curvelet Based Offline Analysis of SEM Images

Manual offline analysis, of a scanning electron microscopy (SEM) image, is a time consuming process and requires continuous human intervention and efforts. This paper presents an image processing based method for automated offline analyses of SEM images. To this end, our strategy relies on a two-stage process, viz. texture analysis and quantification. The method involves a preprocessing step, aimed at the noise removal, in order to avoid false edges. For texture analysis, the proposed method employs a state of the art Curvelet transform followed by segmentation through a combination of entropy filtering, thresholding and mathematical morphology (MM). The quantification is carried out by the application of a box-counting algorithm, for fractal dimension (FD) calculations, with the ultimate goal of measuring the parameters, like surface area and perimeter. The perimeter is estimated indirectly by counting the boundary boxes of the filled shapes. The proposed method, when applied to a representative set of SEM images, not only showed better results in image segmentation but also exhibited a good accuracy in the calculation of surface area and perimeter. The proposed method outperforms the well-known Watershed segmentation algorithm.

Deep Learning for Medical Image Processing: Overview, Challenges and the Future

The health care sector is totally different from any other industry. It is a high priority sector and consumers expect the highest level of care and services regardless of cost. The health care sector has not achieved society’s expectations, even though the sector consumes a huge percentage of national budgets. Mostly, the interpretations of medical data are analyzed by medical experts. In terms of a medical expert interpreting images, this is quite limited due to its subjectivity and the complexity of the images; extensive variations exist between experts and fatigue sets in due to their heavy workload. Following the success of deep learning in other real-world applications, it is seen as also providing exciting and accurate solutions for medical imaging, and is seen as a key method for future applications in the health care sector. In this chapter, we discuss state-of-the-art deep learning architecture and its optimization when used for medical image segmentation and classification. The chapter closes with a discussion of the challenges of deep learning methods with regard to medical imaging and open research issue.

Deep learning based isolated Arabic scene character recognition

The technological advancement and sophistication in cameras and gadgets prompt researchers to have focus on image analysis and text understanding. The deep learning techniques demonstrated well to assess the potential for classifying text from natural scene images as reported in recent years. There are variety of deep learning approaches that prospects the detection and recognition of text, effectively from images. In this work, we presented Arabic scene text recognition using Convolutional Neural Networks (ConvNets) as a deep learning classifier. As the scene text data is slanted and skewed, thus to deal with maximum variations, we employ five orientations with respect to single occurrence of a character. The training is formulated by keeping filter size 3 × 3 and 5 × 5 with stride value as 1 and 2. During text classification phase, we trained network with distinct learning rates. Our approach reported encouraging results on recognition of Arabic characters from segmented Arabic scene images.