Text Lines Research Articles

A hybrid text line segmentation approach for the ancient handwritten unconstrained freestyle Modi script documents

ABSTRACTIn this paper a novel approach for the segmentation of text lines in handwritten, unconstrained, and freestyle Modi script document image is presented. The approach is based on the analysis of the gray level thresholding metric to estimate the text region location. True text line locations are estimated and segmented using the estimated text region location. The connected components of each estimated text region are scrutinized to group and calculate the text line region. The approach is recursively applied on the estimated line segments until the individual text lines are not separated. This algorithm handles various type of documents as multi-skewed, non-uniform Shirorekha, variable size characters or text lines, touching and overlapping line segments therefore it is robust. The algorithm is tested and compared using ancient handwritten Modi script documents and the benchmarking freely available online datasets, respectively.

Dynamic effects of user- and marketer-generated content on consumer purchase behavior: Modeling the hierarchical structure of social media websites

User- and marketer-generated content items on social media platforms are supposed to have an impact on economic target variables, such as variables measuring consumers' purchase behavior. The position of each content item – and thus the impact on economic variables – changes with newly appearing items. We propose a hierarchy score to capture the dynamics of the content items on social media platforms. In order to mimic the reduced visibility of earlier content items, our hierarchy score computes the position of content items based on the number of text line equivalents of content items above a particular item. Employing the proposed hierarchy score in a dynamic regression framework for data of a large online store yields improved estimates and predictions compared to a variety of other models.

Multi-Oriented and Multi-Lingual Scene Text Detection with Direct Regression.

Multi-oriented and multi-lingual scene text detection plays an important role in computer vision area and is challenging due to the wide variety of text and background. In this paper, firstly we point out the two key tasks when extending CNN based object detection frameworks to scene text detection. The first task is to localize the text region by a down-sampled segmentation based module, and the second task is to regress the boundaries of text region determined by the first task. Secondly, we propose a scene text detection framework based on fully convolutional network (FCN) with a bi-task prediction module in which one is pixel-wise classification between text and non-text, and the other is pixel-wise regression to determine the vertex coordinates of quadrilateral text boundaries. Post-processing for word-level detection is based on Non-Maximum Suppression (NMS), and for line-level detection we design a heuristic line segments grouping method to localize long text lines. We evaluated the proposed framework on various benchmarks including multi-oriented and multi-lingual scene text datasets, and achieved state-of-the-art performance on most of them. We also provide abundant ablation experiments to analyze several key factors in building high performance CNN based scene text detection systems.

Ground Truth for training OCR engines on historical documents in German Fraktur and Early Modern Latin

In this paper we describe a dataset of German and Latin \textit{ground truth} (GT) for historical OCR in the form of printed text line images paired with their transcription. This dataset, called \textit{GT4HistOCR}, consists of 313,173 line pairs covering a wide period of printing dates from incunabula from the 15th century to 19th century books printed in Fraktur types and is openly available under a CC-BY 4.0 license. The special form of GT as line image/transcription pairs makes it directly usable to train state-of-the-art recognition models for OCR software employing recurring neural networks in LSTM architecture such as Tesseract 4 or OCRopus. We also provide some pretrained OCRopus models for subcorpora of our dataset yielding between 95\% (early printings) and 98\% (19th century Fraktur printings) character accuracy rates on unseen test cases, a Perl script to harmonize GT produced by different transcription rules, and give hints on how to construct GT for OCR purposes which has requirements that may differ from linguistically motivated transcriptions.

A Novel Text Line Segmentation Method Based on Contour Curve Tracking for Tibetan Historical Documents

In this paper, we proposed a novel method for text line segmentation of Tibetan historical document image with uchen script based on contour tracking. Our method is mainly to segment the text lines from the image documents using the contour curve of the text lines, which consists of three parts: First, we calculate the barycentre coordinates of the connected components for the text regions, and then the barycentre of each text line is connected in order, so that the main part of each text line is connected and a new connected component is formed; then the contour curve of the connected component is obtained using the contour tracing algorithm; Second, the contour curve and the barycentre gravity are used to assign key elements (such as the syllable point, the upper vowel, the lower vowel, and the broken strokes and so on) of the text lines, and next the candidate text lines are obtained based on these connected components; Finally, the contour tracking algorithm is used to calculate the contour curve of the candidate text lines and segment the text lines. We evaluated our text line segmentation method on the 200 document image data sets. Experimental results show that the proposed method based on contour curve tracing can accurately segment the text lines of image documents and achieve the encouraging results.

A comprehensive study of hybrid neural network hidden Markov model for offline handwritten Chinese text recognition

This paper proposes an effective segmentation-free approach using a hybrid neural network hidden Markov model (NN-HMM) for offline handwritten Chinese text recognition (HCTR). In the general Bayesian framework, the handwritten Chinese text line is sequentially modeled by HMMs with each representing one character class, while the NN-based classifier is adopted to calculate the posterior probability of all HMM states. The key issues in feature extraction, character modeling, and language modeling are comprehensively investigated to show the effectiveness of NN-HMM framework for offline HCTR. First, a conventional deep neural network (DNN) architecture is studied with a well-designed feature extractor. As for the training procedure, the label refinement using forced alignment and the sequence training can yield significant gains on top of the frame-level cross-entropy criterion. Second, a deep convolutional neural network (DCNN) with automatically learned discriminative features demonstrates its superiority to DNN in the HMM framework. Moreover, to solve the challenging problem of distinguishing quite confusing classes due to the large vocabulary of Chinese characters, NN-based classifier should output 19900 HMM states as the classification units via a high-resolution modeling within each character. On the ICDAR 2013 competition task of CASIA-HWDB database, DNN-HMM yields a promising character error rate (CER) of 5.24% by making a good trade-off between the computational complexity and recognition accuracy. To the best of our knowledge, DCNN-HMM can achieve a best published CER of 3.53%.

Online Support Shows Potential of Tech-Based MH Care, Insel Says

Back to table of contents Previous article Next article Clinical and Research NewsFull AccessOnline Support Shows Potential of Tech-Based MH Care, Insel SaysNick ZagorskiNick ZagorskiSearch for more papers by this authorPublished Online:13 Jun 2018https://doi.org/10.1176/appi.pn.2018.6b22AbstractOnline counseling networks like the Crisis Text Line or 7 Cups show how simple technological approaches can reduce delays in care and build online help communities.“There has been a dramatic shift in the service economy in the last few years,” said Thomas Insel, M.D., in the lecture he presented at APA’s 2018 Annual Meeting last month. “How people expect to get goods and services has fundamentally changed.” Technology can expand the reach of mental health services as well as offer certain advantages, says Thomas Insel, M.D.David HathcoxThe field of medicine is not exempt from these changes: he noted that companies like Google and Facebook are betting that the way that people get health care now—through traditional doctor visits and medical facilities—will be changing. Large tech firms invested over $3 billion in health care last year, looking to leverage the power of Big Data and social networks for new health care screening and delivery models.“From their point of view, why not try and bend the curve?,” Insel asked. He noted that today’s health care, especially mental health care, is constrained by imprecise diagnoses, lack of access to treatment, delays in treatment initiation, and fragmented care. These shortcomings and barriers to care are a shame, Insel continued.Adding technology to tested mental health interventions could make these treatments work more consistently and reach more people, Insel said. “I call it high tech for high touch,” he said. “It’s like putting treatment on steroids.” Insel noted the success of the Crisis Text Line, a text-based counseling service for people experiencing thoughts of suicide or self-harm. According to the company’s website, “Every texter is connected with a crisis counselor, a real-life human being trained to bring texters from a hot moment to a cool calm through active listening and collaborative problem solving.” An algorithm used by the company enables messages to be interpreted for severity and triaged to the appropriate professionals.Another company providing a similar service is 7 Cups. (Insel disclosed that he serves on the company’s board.)7 Cups is an online and text-message-based peer-support network in which people who want to talk can search for an online “listener” who meets their needs. Some encounters may involve crises but others are on a broad topics—for example, discussions of relationship problems or frustrations. Individuals who complete a listening session are eligible to participate in a short training session and become listeners themselves—thus expanding the reach of the online support community. While these “listeners” are nonprofessional, 7 Cups also offers online therapy services through messaging and voice options with licensed therapists for a modest fee (for example, $150 for a month of unlimited chatting).“This whole company consists of 15 people working without a central office, and their therapists served over 575,000 people in California last year,” Insel said. By comparison, psychotherapists working in brick-and-mortar settings in the state served around 480,000 people in same period. Insel acknowledged that 7 Cups may not necessarily be providing higher quality psychotherapy, but he pointed out that the company does routinely fire psychotherapists who don’t meet quality standards. And if online methods can deliver effective therapy, the potential could be enormous. Though Insel had referenced several criticisms about traditional mental health care, he did say that one area where psychiatry is not faltering is in the quality of treatment it provides. He doesn’t typically hear psychiatrists tell him they cannot give good care. “They say that they have a lot of barriers to providing care, but when they can give care, it works.” ■ ISSUES NewArchived

Slant Removal Technique for Historical Document Images

Slanted text has been demonstrated to be a salient feature of handwriting. Its estimation is a necessary preprocessing task in many document image processing systems in order to improve the required training. This paper describes and evaluates a new technique for removing the slant from historical document pages that avoids the segmentation procedure into text lines and words. The proposed technique first relies on slant angle detection from an accurate selection of fragments. Then, a slant removal technique is applied. However, the presented slant removal technique may be combined with any other slant detection algorithm. Experimental results are provided for four document image databases: two historical document databases, the TrigraphSlant database (the only database dedicated to slant removal), and a printed database in order to check the precision of the proposed technique.

Was es bedeutet, gegen Journalisten zu hetzen

AbstractThe problematization of journalistic work is a recurring theme in literature. This paper traces the reservations journalists see themselves confronted with according to literary representations, how criticism is solidified, and how the suspicion that journalists are making up facts is becoming more universal. This will be done by looking at a historical line of texts, ranging from Die Journalisten by Gustav Freytag to Josef Haslinger’s Opernball. There are two main criteria underlying the choice of texts: first, the figures of journalists play a central role in each of the texts; second, these texts make visible the stages of a presumed development process from the middle of the 19th century through to today in the most succinct way possible.

Driver adherence to recommendations from support systems improves if the systems explain why they are given: A simulator study

This paper presents a large-scale simulator study on driver adherence to recommendations given by driver support systems, specifically eco-driving support and navigation support. 123 participants took part in this study, and drove a vehicle simulator through a pre-defined environment for a duration of approximately 10 min. Depending on the experimental condition, participants were either given no eco-driving recommendations, or a system whose provided support was either basic (recommendations were given in the form of an icon displayed in a manner that simulates a heads-up display) or informative (the system additionally displayed a line of text justifying its recommendations). A navigation system that likewise provided either basic or informative support, depending on the condition, was also provided.Effects are measured in terms of estimated simulated fuel savings as well as engine braking/coasting behaviour and gear change efficiency. Results indicate improvements in all variables. In particular, participants who had the support of an eco-driving system spent a significantly higher proportion of the time coasting. Participants also changed gears at lower engine RPM when using an eco-driving support system, and significantly more so when the system provided justifications. Overall, the results support the notion that providing reasons why a support system puts forward a certain recommendation improves adherence to it over mere presentation of the recommendation.Finally, results indicate that participants’ driving style was less eco-friendly if the navigation system provided justifications but the eco-system did not. This may be due to participants considering the two systems as one whole rather than separate entities with individual merits. This has implications for how to design and evaluate a given driver support system since its effectiveness may depend on the performance of other systems in the vehicle.

Estela con iconografía e inscripción celtibéricas procedente de Clunia

El objetivo del trabajo es publicar una nueva estela celtibérica recuperada en Clunia. Es una pieza que contiene una inscripción celtibérica y la representación de un guerrero, combinación característica de los monumentos procedentes de esta ciudad. El texto está inscrito en signario celtibérico occidental, presenta redundancia vocálica y, en principio, es compatible con el denominado sistema dual. Se dispone en una única línea de texto delimitada por dos líneas incisas, quizá parte de una cartela. La lectura propuesta es mukuuroskiimine+ y, aunque no hay interpunciones, lo más plausible es identificar en primero término un nombre personal (Muguros) y un segundo elemento que carece de buenos paralelos. Sobre el epígrafe aparece un infante calzado con botas, pertrechado con grebas y que parece vestir un sagum, con su diestra agarra un astil que porta sobre el hombro y cuyo extremo superior se ha perdido.

Fully convolutional network with dilated convolutions for handwritten text line segmentation

We present a learning-based method for handwritten text line segmentation in document images. Our approach relies on a variant of deep fully convolutional networks (FCNs) with dilated convolutions. Dilated convolutions allow to never reduce the input resolution and produce a pixel-level labeling. The FCN is trained to identify X-height labeling as text line representation, which has many advantages for text recognition. We show that our approach outperforms the most popular variants of FCN, based on deconvolution or unpooling layers, on a public dataset. We also provide results investigating various settings, and we conclude with a comparison of our model with recent approaches defined as part of the cBAD ( https://scriptnet.iit.demokritos.gr/competitions/5/ ) international competition, leading us to a 91.3% F-measure.

A coarse-to-fine scene text detection method based on Skeleton-cut detector and Binary-Tree-Search based rectification

Scene text detection has been a long standing hot and challenging research topic in pattern recognition. In this paper a novel coarse-to-fine text detection method is proposed to solve edge-adhesion problem. In coarse detection stage, Skeleton-cut detector is proposed. At first,8-Neighborhoods-Search is applied on skeletons map to find the adhesion junctions between text and background skeletons. Then junctions in disordered skeletons are picked out by hysteresis selection and cut to separate text skeletons from background. And the text skeletons are verified through a two-stage classifier to obtain the coarse detection result. In fine detection stage, bounding boxes of all these filtered skeletons are weighted accumulated to obtain the Static Skeleton Response(SSR). Then many finer text lines candidates can be calculated through the gradient operation to the SSR’s horizontal projection. And the text rectification based on Binary-Tree-Search is proposed to find a path from text lines’ search space to the fine detection result. Experimental results on ICDAR dataset, SVT dataset and MSRA-TD500 dataset demonstrate that our algorithm achieves state of art performance in scene text detection.

Unsupervised multi-language handwritten text line segmentation

Unsupervised multi-language handwritten text line segmentation

An inscribed altar from the Khirbat Ataruz Moabite sanctuary

A cylindrical stone incense altar inscribed with seven lines of text in two separate inscriptions was discovered in a cultic context during 2010 excavations at Khirbat Ataruz in Jordan. The two short inscriptions are written in Moabite language, using an Early Moabite script datable to the late 9th or early 8th century BCE. Both inscriptions also employ Hieratic numerals. Inscription A appears to tabulate small quantities of metal, possibly for some purpose relating to the cultic context of the inscription. The longer Inscription B appears to be potentially dedicatory and/or commemorative in focus, but remains largely enigmatic. These inscriptions provide a new important historical witness to the period after the Moabite conquest and occupation of Khirbat Ataruz/Atarot described in the Mesha Inscription.

A robust proposal generation method for text lines in natural scene images

Abstract Motivated by the success of object proposal generation methods for object detection, we propose a novel method for generating text line proposals from natural scene images. Our strategy is to detect text regions which we define as part of text lines containing a whole character or transitions between two adjacent characters. We observe that, if we scale text regions to a small and fixed size, their image gradients exhibit certain patterns irrespective of text shapes and language types. Based on this observation, we propose simple features which consist of means and standard deviations of image gradients to train a Random Forest so as to detect text regions over multiple image scales and color channels. Text regions are then merged into text line candidates which are ranked based on the Random Forest responses combined with the shapes of the candidates, e.g., horizontally elongated candidates are given higher scores, because they are more likely to contain texts. Even though our method is trained on English, our experiments demonstrate that it achieves high recall with a few thousand good quality proposals on four standard benchmarks, including multi-language datasets. Following the One-to-One and Many-to-One detection criteria, our method achieves 91.6%, 87.4%, 92.1% and 97.9% recall on the ICDAR 2013 Robust Reading Dataset, Street View Text Dataset, Pan’s multilingual Dataset and Sampled KAIST Scene Text Dataset respectively, with an average of less than 1250 proposals.

CRF based text detection for natural scene images using convolutional neural network and context information

Abstract This paper presents a novel scene text detection method based on conditional random field (CRF) framework. We estimate the confidence of Maximally Stable Extremal Region (MSER) being text by leveraging convolutional neural network (CNN) to define the unary cost item. In addition, we establish the neighboring interactions for MSERs using four different features including color, shape, stroke and spatial features to define the pairwise cost item. Considering the special layout of texts appearing in natural scene images, we employ context information to recover missing text MSER candidates. Furthermore, text MSERs are grouped into candidate text lines which are verified with shape-specific classifiers by integrating gray and binary features. Experimental results on four public benchmark datasets show that the proposed method achieves the comparable performance.

Utilizing text recognition for the defects extraction in sewers CCTV inspection videos

This paper proposed a novel automated framework for analyzing and tracking sewer inspection close-circuit television (CCTV) videos. The proposed model mainly supports the off-site examination and quality management process of the videos and enables efficient revaluation of CCTV videos to extract sewer condition data. The study discusses an important module for any automated analysis and defect detection in CCTV video. It includes two main modules: text recognition and cracks extraction. In the first module, multi-frame integration (MFI) was applied to reduce the background complexity, time and computational requirements needed for the video processing. Then maximally stable extremal regions (MSER) was used on the grayscale channel and HSV channel to effectively detect all the text edges. Saturation color channel was also applied to verify the detected text line and remove false alarms. In the second module, by utilizing the text information on each frame, the operator’s operation during the inspection is simulated which would indicate valuable clues about the location and severity of the cracks. The proposed methodology was validated using a set of video provided by the Korea Institute of Construction Technology.

Locate, Combine, Contradict, Iterate: Serial Strategies for PostInternet Art

Locate, Combine, Contradict, Iterate: Serial Strategies for PostInternet Art

Print-scan invariant text image watermarking for hardcopy document authentication

In this paper, a novel contour feature-based text image watermarking scheme against print and scan processes is proposed. We employ a mathematical multiplicative transformation model to approximate the geometric invariant feature that can survive a variety of attacks during the print-scan process and thus serve as reference points for both watermark embedding and extraction. Based on the print-scan invariant, the boundary points of each character are flipped using Fourier descriptors with visual perception characteristics, so that the watermarks are embedded into the visually nonsignificant points. In the calculation process of the print-scan invariant, a certain text line serves as the benchmark line without affording additional characters for watermark adjustment. Thus, the hiding capacity is greatly improved. For the data detection, noise reduction and deskewing mechanisms are performed previously to compensate for the distortions caused by hardcopy. The watermark is then extracted by parity check of the invariant feature of connected components for soft authentication. The experimental results show that the proposed approach is not limited to a particular language, and has better robustness, watermark transparency as well as hiding capacity compared with some existing methods.