Text Localization Research Articles

TEXT LOCALIZATION IN SCENE IMAGES BY BENDELET TRANSFORM

In an automated text recognition system, one of the prerequisites is the localization of text. It is a challenging task in scene image due to their background and non uniform size of characters in the images. In this study, an efficient text localization system using bendlet transform is presented. Among the various multi-resolution and multi-directional analysis, bendlet transform has superior property that they classify the curvature precisely. To achieve this property, it uses an addition parameter than shearlets called bending operator. The system decomposes the scene images by bendlet transform and then reconstructs using the bands which contains only the edge information. Then, a series of post processing is applied to locate the text region in a scene image. Results show the robustness of the text localization system by successfully locating the text region in the scene images with different background and non-uniform text sizes.

Text Extraction and Recognition in Natural Scene Images using Contourlet Transform and PNN

Of late, the rapid development in the technology and multimedia capability in digital cameras and mobile devices has led to ever increasing number of images or multi-media data to the digital world. Particularly, in natural scene images, the text content provides explicit information to understand the semantics of images. Therefore, a system developed for extracting and recognizing texts accurately from natural scene images, in real-time, has significant relevance to numerous applications such as, assistive technology for people with vision impairment, tourist with language barrier, vehicle number plate detection, street signs, advertisement bill boards, robotics, etc. The extraction of the texts from natural scene images is a formidable task due to large variations in character fonts, styles, sizes, text orientations, presence of complex backgrounds and varying light conditions. The main focus of this research paper is to propose a novel hybrid approach for automatic detection, localization, extraction and recognition of text in natural scene images with cluttered background. Firstly, image regions with text are detected using edge features (GLCM) extracted from Contourlet transformed image and SVM (Support Vector Machine) classifier. Secondly, horizontal projection is applied on text regions for segmenting lines and vertical projection is applied on each text line for segmenting characters. The proposed method for text extraction has produced the precision, recall, F-Score and accuracy of 98.50%, 90.85.62%, 95.00%, and 89.90%, respectively. And, these results prove that the proposed method is efficient. Further, the so extracted characters are processed for recognition using contourlet transform and Probabilistic Neural Network (PNN) classifier. The computed features are moment invariants. Only the English script is considered for the experimentation. The proposed character recognition method has accuracy of 79.07%, which is higher in comparison to accuracy of 75.15% obtained by KNN (K-Nearest Neighbors) classifier

Deep Multi-Scale Context Aware Feature Aggregation for Curved Scene Text Detection

Scene text plays a significant role in image and video understanding, which has made great progress in recent years. Most existing models on text detection in the wild have the assumption that all the texts are surrounded by a rotated rectangle or quadrangle. While there also exist lots of curved texts in the wild, which would not be bounded by a regular bounding box. In this paper, we develop a novel architecture to localize the text regions, which can deal with curved-shape scene texts. Specifically, we first design a text-related feature enhancement module by incorporating the prior knowledge of the text shape to enhance the feature representations. After that, based on the enhanced features, we employ a region proposal network to generate the candidate boxes of scene texts. For each text candidate, a pyramid region-of-interest pooling attention module is utilized to extract the fixed-size features. Finally, we exploit the box-aware context-based text segmentation module and box refinement network to obtain the location of scene text. Experiments are conducted on four challenging benchmarks ${CTW1500}$ , ${totalTEXT}$ , ${ICDAR-2015}$ and ${MLT}$ , and the experimental results have demonstrated the superiority of our model.

An algorithm to identify text marked in rock core pictures with machine learning algorithm

The aim of our project is to extract numerical descriptors or gather feature of rock samples from their images made under microscope with computer vision algorithm which can overcome the bottleneck of traditional method. It is impossible to solve it without getting the imformation of microscope magnification of images when they were made. That is why this paper is about an algorithm to identify the text used to mark microscope magnification information in rock core sample pictures. This is a sub project. As locations of text in these pictures are not fixed, font of text is not fixed, font size of text is not fixed, a method of two steps is designed to identify the text from the background. It works as follows: first, find the location of text; second, recognize the text with Convolution Neural Network. Experimental results demonstrate the potential value of the proposed approach to identify the magnification information in rock core sample pictures and to apply as a funtion to a program of our future project.

Real-time localization of multi-oriented text in natural scene images using a linear spatial filter

This paper proposes a multi-oriented text localization method in natural images suitable for real-time processing of high-definition video on portable and mobile devices. Our method is based on the connected components (CC) approach: first, CC are isolated by convolving a multi-scale pyramid with a specifically designed linear spatial filter followed by hysteresis thresholding. Next, non-textual CC are pruned employing a local classifier consisting of a cascade of multilayer perceptron (MLP) fed with increasingly extended feature vectors. The stroke width feature is estimated in linear time complexity by computing the maximal inscribed squares in the CC. Candidate CC and their neighbors are then checked using a more context aware neural network classifier that takes into account the target CC and their vicinity. Finally, text sequences are extracted in all pyramid levels and fused using dynamic programming. The main contribution of the work presented here is execution speed: the CPU-only parallel implementation of the proposed method is capable of processing 1080p HD video at nearly 30 frames per second on a standard laptop. Furthermore, when benchmarked on the ICDAR 2013 Robust Reading and on the ICDAR 2015 Incidental Scene Text data sets, our system performs more than twice faster than the state-of-the-art, while still delivering competitive results in terms of precision and recall.

Detection, Localization of Text in Images by Mser and Enhanced Swt

The number digital images and digital videos has raised highly. The text shows in the images is major thing for fully understanding the images. So it is very signifigant to recognize the text. First step of text recognization system is known as text detection. But detecting text from images is very challenging and it is receiving high amount of concentration. Text detection is improved by pre-procssing techniques like noise removal, contrast changes and etc. Recently MSER (Maximally Stable Edge Region) and SWT (Stroke Width Transformation) methods are used seperately or combined with each other for detecting and localizing text. This paper analyzes and compares technical challenges and performance of above methods and take in new and fast method for text detection and localization. The proposed method comprises an enhanced technique based on SWT which combined with MSER. The proposed method detect wording which is located in differnt directions, different angle and combined letters. Performance of exisisting and proposed methods on road signal images depicted in experimental results.

Unambiguous Scene Text Segmentation with Referring Expression Comprehension.

Text instance provides valuable information for the understanding and interpretation of natural scenes. The rich, precise high-level semantics embodied in the text could be beneficial for understanding the world around us, and empower a wide range of real-world applications. While most recent visual phrase grounding approaches focus on general objects, this paper explores extracting designated texts and predicting unambiguous scene text segmentation mask, i.e. scene text segmentation from natural language descriptions (referring expressions) like orange text on a little boy in black swinging a bat. The solution of this novel problem enables accurate segmentation of scene text instances from the complex background. In our proposed framework, a unified deep network jointly models visual and linguistic information by encoding both region-level and pixel-level visual features of natural scene images into spatial feature maps, and then decode them into saliency response map of text instances. To conduct quantitative evaluations, we establish a new scene text referring expression segmentation dataset: COCO-CharRef. Experimental results demonstrate the effectiveness of the proposed framework on the text instance segmentation task. By combining image-based visual features with language-based textual explanations, our framework outperforms baselines that are derived from state-of-the-art text localization and natural language object retrieval methods on COCO-CharRef dataset.

The Effect of an Innovative Vision Simulator (OrCam) on Quality of Life in Patients with Glaucoma

Purpose:The aim of this study was to determine the impact of OrCam on vision-related quality of life of patients with legal blindness and end-stage glaucoma. OrCam is a device comprised of a camera and earpiece attached to the user’s eyeglass that translates written text to speech.Methods:In this prospective, observational study, a total of 27 participants were recruited and tested. All participants were legally blind. Participants were provided with the OrCam device during the study period and trained how to operate it. National Eye Institute Visual Functioning Questionnaire-25 (NEI-VFQ-25) and other questionnaires were given to participants at the baseline visit and at the end of the 4-week study for comparison. Additional questions were surveyed weekly during the study to monitor progress.Results:At the final visit after 1 month, most participants, 74.1% ( n = 20/27), reported an increase in their overall quality of life. Participants were highly satisfied with the device; 88.9% ( n = 24/27) reported being “very satisfied” or “satisfied” with the device. The NEI-VFQ-25 near vision subscale score improved from 29.9 to 37.2 ( p = .045).Discussion:OrCam allowed participants with legal blindness to read independently when in controlled settings, subsequently improving their quality of life. However, users had more difficulty using the device when the location of text was not readily apparent (e.g., street signs, billboards), and users with some degree of vision had an easier time aligning the device with the text.Implications for practitioners:This device may offer participants with visual impairments the opportunity to read independently.

Text-mining BHL: towards new interfaces to the biodiversity literature

The taxonomic literature is one of the largest resources of information on biodiversity, both current and in the past. Unlike many scientific disciplines this literature remains perpetually relevant as successive taxonomic work builds upon those earlier foundations. Projects such as the Biodiversity Heritage Library (BHL) have greatly increased access to that literature, as have numerous independent digitisation efforts by museums, herbaria, and publishers. But the focus of this access has been human readers, with limited use of text mining tools, mostly focussed on extracting taxonomic names. This talk explores other kinds of data that can be extracted from text on BHL and elsewhere, focusing on taxonomic names, geographic localities and specimen codes in the context of the BioStor project (https://biostor.org, Page 2011). The problem of finding taxonomic names in text has been well studied (e.g., Akella et al. 2012), and new BHL content is continuously indexed by names. Despite this, there is only weak linkage between taxonomic name databases and BHL. Even projects that create these links (e.g., BioNames, Page 2013) do not enable links in the reverse direction. In other words, a BHL reader is unaware whether the appearance of a name on a page is the first publication of that name, nor are they told of the fate of a name in subsequent research. The absence of these links reduces the value of BHL to working taxonomists. In addition to taxonomic names, a typical taxonomic paper often contains specimen codes. Extracting these from text and linking them to digital representations, such as occurrence records in GBIF, opens up the possibility to provide detailed provenance for occurrence data, as well as citation-based metrics for the utility of natural history collections. Taxonomic papers are also often rich in geographic information. A simple method for extracting locality information from text is to search for latitude and longitude coordinates, and BioStor currently does this. To date some 83,000 individual point localities have been extracted (Fig. 1 ). These are used to provide a simple geographic search interface in BioStor, and are also harvested by JournalMap (Karl et al. 2013). But these localities are not linked to the original location in the source text, nor are they linked to any associated specimens, so they cannot be interpreted as occurrences that could be harvested by GBIF. If the goal is to contribute to GBIF then we need tools that can parse locality information and link that to associated specimens. A general framework for handling data on taxonomic names, specimens, and geographic localities in text is to treat them as annotations (Batista-Navarro et al. 2017). By modelling annotations using the Web Annotation Data Model (https://www.w3.org/TR/annotation-model/ ) we can incorporate these annotations into biodiversity knowledge graphs (Page 2016). We can also combine these annotations with new standards for describing digitised content, such as the International Image Interoperability Framework (IIIF, https://iiif.io). The implications of this approach for developing new interfaces to the biodiversity literature will be discussed.

Text extraction from natural scene images using Renyi entropy

Nowadays, development in machine vision incorporated with artificial intelligence surpasses the ability of human intelligence and its application expands exponentially with the increasing number of electronic gadgets in our day-to-day life. The explosive revolution in multimedia research leads to the need for expanding the utility of texts in a machine vision environment to promote web search operation. Hence, extracting text from images forms the core aspect of information retrieval-based intelligent system. This article is aimed towards extracting text from unconstrained environments. Here, the significance of the CIE-Lab colour space is analysed over text localisation assisted through Renyi entropy-based thresholding. The proposed algorithm is tested on the MSRA Text Detection 500 dataset (MSRA-TD500) and Street View Text (SVT) datasets, which are challenging datasets. Authors’ proposed Renyi entropy-based text localisation algorithm is successful in identifying blurred texts, texts with different font characteristics and multi-lingual texts with manifold orientations from complex background natural scenes.

An Enhanced MSER Pruning Algorithm for Detection and Localization of Bangla Texts from Scene Images

Text detection and localization have great importance for content based image analysis and text based image indexing. The efficiency of text recognition depends on the efficiency of text localization. So, the main goal of the proposed method is to detect and localize text regions with high accuracy. To achieve this goal, a new and efficient method has been introduced for localization of Bangla text from scene images. In order to improve precision and recall as well as f-measure, Maximally Stable Extremal Region (MSER) based method along with double filtering techniques have been used. As MSER algorithm generates many false positives, we have introduced double filtering method for removing these false positives to increase the f-measure to a great extent. Our proposed method works at three basic levels. Firstly, MSER regions are generated from the input color image by converting it into gray scale image. Secondly, some heuristic features are used to filter out most of the false positives or non-text regions. Lastly, Stroke Width Transform (SWT) based filtering method is used to filter out remaining non-text regions. Remaining components are then grouped into candidate text regions marked by bounding box over each region. As there is no benchmark database for Bangla text, the proposed method is implemented on our own prepared database consisting of 200 scene images of Bangla texts and has got prominent performance. To evaluate the performance of our proposed approach, we have also tested the proposed method on International Conference on Document Analysis and Recognition( ICDAR) 2013 benchmark database and have got a better result than the related existing methods.

LOCALIZATION OF WEBSITE VERBAL CONTENT INTO ENGLISH AS A MARKETING STRATEGY

Цель. На сегодняшний день необходимость локализации онлайн-ресурсов обусловлена глобализирующей тенденцией развития интернет-коммуникации. Стратегии локализации варьируются в зависимости от характера реализуемых посредством интернет-платформ продуктов и услуг, однако в любом случае локализованные версии предоставляют потенциальным клиентам доступ к достоверным источникам информации о мировых брендах. Целью настоящего исследования является изучение феномена локализации с лингвистической точки зрения применительно к вербальному контенту международных сайтов. В статье предпринимается попытка обоснования необходимости дифференции англоязычного контента для каждой из англоязычных стран, включенных в структуру локализованных версий сайтов, и выявить вербальные средства, за счет которых осуществляется данный процесс.Методология проведения работы. Основу исследования образуют методы контекстуального, сравнительного и прагматического анализа.Результаты. В результате проведенного исследования были описаны механизмы создания локализованного контента в рамках одного сайта и в пространстве одного языка – английского. В статье делается вывод о значимости локализации сайта и текста, в частности, на английский язык и о необходимости дифференциации между двумя уровнями, на которых может осуществляться локализация – уровень текста и интертекстуальный уровень.Область применения результатов. Результаты исследования могут быть применены в сфере осуществления масштабных проектов локализации различных видов международных сайтов.

Text localization in camera captured images using fuzzy distance transform based adaptive stroke filter

Localization of text from camera captured images with complex background is now-a-days a growing demand of modern IT enable service. Most of the current text localization techniques are sensitive to text features like color, size, style and also to the background clutter. Among all the methods proposed in different literatures, Stroke Filter is much more effective in localization of text. The effectiveness of traditional stroke filter is limited because of its fixed width and is capable of segmenting strokes/texts of predefined range of width. The proposed method uses Fuzzy Distance Transform based adaptive stroke filter which can effectively localize text regions from camera captured images with complex background. The method is applied by experiment on a database containing 600 images and the visual response of text segmentation is quite impressive. To get the accuracy of the proposed method, it is applied on a set of 16 test images and the segmentation result is compared with the ground truth images resulting in a recall, precision and f-measure values of 96.65%, 87.77% and 91.89% respectively.

An augmented reality sign-reading assistant for users with reduced vision.

People typically rely heavily on visual information when finding their way to unfamiliar locations. For individuals with reduced vision, there are a variety of navigational tools available to assist with this task if needed. However, for wayfinding in unfamiliar indoor environments the applicability of existing tools is limited. One potential approach to assist with this task is to enhance visual information about the location and content of existing signage in the environment. With this aim, we developed a prototype software application, which runs on a consumer head-mounted augmented reality (AR) device, to assist visually impaired users with sign-reading. The sign-reading assistant identifies real-world text (e.g., signs and room numbers) on command, highlights the text location, converts it to high contrast AR lettering, and optionally reads the content aloud via text-to-speech. We assessed the usability of this application in a behavioral experiment. Participants with simulated visual impairment were asked to locate a particular office within a hallway, either with or without AR assistance (referred to as the AR group and control group, respectively). Subjective assessments indicated that participants in the AR group found the application helpful for this task, and an analysis of walking paths indicated that these participants took more direct routes compared to the control group. However, participants in the AR group also walked more slowly and took more time to complete the task than the control group. The results point to several specific future goals for usability and system performance in AR-based assistive tools.

Arabic Cursive Text Recognition from Natural Scene Images

This paper presents a comprehensive survey on Arabic cursive scene text recognition. The recent years’ publications in this field have witnessed the interest shift of document image analysis researchers from recognition of optical characters to recognition of characters appearing in natural images. Scene text recognition is a challenging problem due to the text having variations in font styles, size, alignment, orientation, reflection, illumination change, blurriness and complex background. Among cursive scripts, Arabic scene text recognition is contemplated as a more challenging problem due to joined writing, same character variations, a large number of ligatures, the number of baselines, etc. Surveys on the Latin and Chinese script-based scene text recognition system can be found, but the Arabic like scene text recognition problem is yet to be addressed in detail. In this manuscript, a description is provided to highlight some of the latest techniques presented for text classification. The presented techniques following a deep learning architecture are equally suitable for the development of Arabic cursive scene text recognition systems. The issues pertaining to text localization and feature extraction are also presented. Moreover, this article emphasizes the importance of having benchmark cursive scene text dataset. Based on the discussion, future directions are outlined, some of which may provide insight about cursive scene text to researchers.

Implementation of Multi-Agent based Digital Rights Management System for Distance Education (DRMSDE) using JADE

The main objective of Distance Education (DE) is to spread quality education regardless of time and space. This objective is easily achieved with the help of technology. With the development of World Wide Web and high-speed internet the quality of DE is improved because now Digital Content (DC) can be easily and in no time distributed to many learners of different locations in text, audio and video formats. But, the main obstacle in digital publishing is the protection of Intellectual Property Rights (IPR) of DC. Digital Rights Management (DRM) that manages rights over any digital creation is the only solution to this problem. In this paper, we have made an attempt to implement a Digital Rights Management System for Distance Education known as DRMSDE. We have identified that Multi-Agent System (MAS) based technology is very popular for such type of implementations. Keeping that in mind, we have chosen one of the most popular Multi-Agent based tools, namely JAVA Agent Development Framework (JADE), for our system. This paper presents an overview and the system architecture for the proposed implementation.

Heuristic Algorithm for Generalized Function Matching

The problem of generalized function matching can be defined as follows: given a pattern p = p1 ⋯ pm and a text t = t1 ⋯ tn, find a mapping f : ∑p→∑t⁎; and all text locations i such that f(p1)f(p2) ⋯ f(pm)=ti ⋯ tj, a substring of t.By modifying the restrictions of the matching function f, one can obtain different matching problems, many of which have important applications. When f : ∑p → ∑t we are faced with problems found in the well-established field of combinatorial pattern matching. If the single character constraint is lifted and f : ∑p→∑t⁎ we obtain generalized function matching as introduced by Amir and Nor (JDA 2007). If we further constrain f to be injective, then we arrive at generalized parametrized matching as defined by Clifford etal. (SPIRE 2009).There are a number of important applications for pattern matching in computational biology, text editors and data compression, to name a few. Therefore, many efficient algorithms have been developed for a wide variety of specific problems including finding tandem repeats in DNA sequences, optimizing embedded systems by reusing code etc.In this work we present a heuristic algorithm illustrating a practical approach to tackling a variant of generalized function matching where f : ∑p → ∑t+ and demonstrate its performance on human-produced text as well as random strings.

Topological descriptors of philosophical text

The article considers the problem of topological descriptors of a philosophical text. Topological descriptors act as discursive points for the localization of the philosophical text, which allow us to describe its role and place in the production of modern humanitarian knowledge. Three perspectives of the analysis of the philosophical text in the contemporary sociocultural situation are outlined: the perspective of historical and philosophical research, the perspective of cultural research and the perspective of modern humanitarian knowledge. The historical and philosophical perspective from the point of view of topology is described through genre differences between philosophical texts, and also taking into account the differences between philosophical schools, trends, styles of thinking in intellectual history. The perspective of cultural research includes a philosophical text in a wide context of political, social and cultural transformations of modern society, and it becomes possible to talk about social topology. The perspective of modern humanitarian knowledge is analyzed from the perspective of the rehabilitation of philosophical discourse, the need for detailed textual work for the full inclusion of philosophy in the production process of significant social and cultural knowledge. As an original methodological approach, the article describes topological analytics, which is actively used by modern researchers to solve a whole range of issues of localization, description and understanding of the role of a philosophical text in modern humanities. Conclusions are drawn about the prospects of further research in this direction.

Can the similarity index predict the causes of retractions in high-impact anesthesia journals? A bibliometric analysis.

Background:The overall similarity index (OSI) and highest similarity scores (HSSs) from a single source might help to predict the potential reasons for the retraction from the anesthesia journals.Methods:Retracted publications, from five highest impact anesthesia journals, were retrieved from the MEDLINE and journal archives and analyzed using a plagiarism detection software (iThenticate) and manually verified for citation characteristics, OSI, HSS, and the presence, extent, and location of the duplicate text. The validity of the OSI including and excluding quotations and references and the HSS in predicting the potential reasons for retraction were tested using the receiver operating characteristic curves.Results:Of the total 138 retracted original and corresponding articles identified, 131 articles were analyzed. Most of them had the HSS more than 40% arising from a single source. Extensive degree of plagiarism (OSI score >35%) was identified through the main text of all analyzed retracted articles. The areas under the curves indicate that the OSI including and excluding quotations and bibliography and the HSS had reasonable ability to predict plagiarism and fabrication with a perfect sensitivity rate and low specificity but were weaker at distinguishing ethical misconduct or inconsistent or erroneous contents.Conclusions:The study highlights the presence of significant plagiarism in the retracted anesthesia publications irrespective to the reasons for retraction. The high OSI and the HSS could be useful tools to identify the potential manuscripts with high risks for plagiarism and fabrication.

A Detection and Verification Model Based on SSD and Encoder-Decoder Network for Scene Text Detection

Text detection in natural scene image is challenging due to text variation in size, orientation, color and complex background, contrast, and resolution. In this paper, we focus on the long text detection in complex background. In order to deal with multi-scale text variation and exploit the recognition result to enhance the detection performance, we propose a detection and verification model based on SSD and encoder-decoder network for scene text detection. First, we present a text localization neural network based on SSD, which incorporates a text detection layer into the standard SSD model and can detect horizontal texts, especially long and dense Chinese texts in natural scenes more effectively. Second, a text verification model based on the encoder-decoder network is designed to recognize and verify the initial detection results, in order to eliminate non-text areas that are falsely detected as text areas. A series of experiments have been conducted on our constructed horizontal text detection dataset, which is composed of the horizontal text images in ICDAR 2017 Competition on Reading Chinese Text in the Wild (RCTW 2017) and some scene images taken by cameras. Compared with previous approaches, experimental results show that our method has achieved the highest recall rate of 0.784 and competitive precision rate in text detection, indicating the effectiveness of our proposed method.