Font Style Research Articles

An unsupervised font style transfer model based on generative adversarial networks

An unsupervised font style transfer model based on generative adversarial networks

Last Teen Pixels for Arabic Font Size and Style Recognition

Arabic printed script segmentation and recognition techniques change from font to other i.e. each font has particular properties calligraphic and structural which differ with other. Majority of segmentation system suffer in word or sub word segmentation into characters because they consider one algorithm to segment all kind of Arabic printed font, style and size. The goal of this work is to prepare a system of word or sub word Optical Font Arabic Recognition (OFAR) for different font size and style of Arabic printed script, in order to integrate it in global Arabic Optical Character Recognition (AOCR) to choose preferred and good segmentation algorithm. APTI database was used to extract last ten pixels for each word or sub word to build new database of last 10 pixels for each word; OFAR is based upon this new database and our extraction approach called Pixels Continuity (PC) algorithm in different matrix direction and some histogram statistics to extract 20 features. Three KNN classifiers with K=5 and three different distances using Cityblock, Euclidean and Correlation based upon majority-vote are used to evaluate the system robustness. This classifier is compared in the first time with Back propagation Neural Network and Steerable Pyramid (SP) algorithm to re cognize three font families, then in the second time with Gaussian Mixture Models (GMMs) to recognize font and size. The average recognition results obtained was 99.55% about font and size and 98.17% for font, size and style recognition.

Willingness to pay for traditional ready-to-eat food packaging: examining the interplay between shape, font and slogan

PurposeThis paper investigates how to design traditional ready-to-eat food packaging by manipulating its shape, font and slogan to promote consumer buying intention and willingness to pay (WTP).Design/methodology/approachTwo package designs, interplaying the square shapes (vertical vs horizontal), font formality (formal vs less formal/casual) and slogans (ethno-positioning vs short and catchy), were created and tested in an online experiment with 483 participants.FindingsThe results revealed that the differed package designs elicited relatively equal levels of attractiveness to consumers and their buying intentions. However, the designs significantly differentiated consumers' WTP. The results further show the significant direct and indirect effects on WTP of packaging attractiveness when it is moderated by package designs and mediated by buying intention. However, the varied package designs did not have significant direct or indirect impacts on WTP when mediated by packaging attractiveness and buying intention.Practical implicationsThe vertical square-shaped packaging, written in a less formal font and highlighting a short and catchy slogan, is more financially promising for marketers to get a better price for local food products perceived as hedonic and bought impulsively on casual occasions. Furthermore, besides prominent package design elements such as font, color, size, material and picture, packaging designers or marketing practitioners should consider other supporting factors, such as shelf efficiency.Originality/valueThis study complements existing studies of “cue utilization theory” and “packaging design theory” by demonstrating the importance of extrinsic packaging cues, such as shape, font style and slogan, in improving consumers' WTP for traditional food products. The study also fills some gaps in the literature by exploring the direct, mediating and moderating relationships between package design, packaging attractiveness, buying intention and WTP, especially in an emerging market such as Indonesia.

Unpaired font family synthesis using conditional generative adversarial networks

Automatic font image synthesis has been an extremely active topic in recent years. Various deep learning-based approaches have been proposed to tackle this font synthesis task by considering it as an image-to-image translation problem in a supervised setting. However, all such approaches mainly focus on one-to-one font mapping, i.e., synthesizing a single font style, making it difficult to handle more practical problems such as the font family synthesis, which is a one-to-many mapping problem. Moreover, this font family synthesis is more challenging because it is an unsupervised image-to-image translation problem, i.e., no paired dataset is available during training. To address this font family synthesis problem, we propose a method that utilizes a single generator to conditionally produce various font family styles to form a font family. To the best of our knowledge, our proposed method is the first to synthesize a font family (multiple font styles belonging to a font), instead of synthesizing a single font style. More specifically, our method is trained to learn a font family by conditioning on various styles, e.g., normal, bold, italic, bold-italic, etc. After training, given an unobserved single font style (normal style font as an input), our method can successfully synthesize the remaining styles (e.g., bold, italic, bold-italic, etc.) to complete the font family. Qualitative and quantitative experiments were conducted to demonstrate the effectiveness of our proposed method.

Bright English Textbook Development for Primary School Grade 6 in Batu City

Based on Indonesia’s current curriculum, English is no longer compulsory for primary school subject; it is only required to be taught starting at the secondary school level. However, in Batu city, the local government treats English as compulsory local content subject for their area development, which certainly raises several challenges for English teachers in Batu due to the lack of supplementary books and other supporting activities. Since 2019, the supplementary book writing project has been initiated, however, the draft still needs to be evaluated before being published, and this is the focus of this project. The study was projected to describe the development process of the Bright English textbook. This study employed a descriptive-analytical approach, involving primary school teachers, students, and an expert. Questionnaires were distributed to teachers and students to evaluate the overall display and content of the book, whereas an evaluation sheet was delivered to the expert to rate the language level and content. Adapting Richards (2001)’s material development theory, three stages of development were present; training, implementation, and evaluation-publication. The training stage facilitated teachers to better comprehend the textbook structures and varied activities relevant to Batu's wisdom and values, while the implementation stage encouraged more interactive and fun collaborative learning plans through synchronous offline and online meetings. The evaluation and publication phase perfected the textbook in various aspects such as its layout, font style selection, picture, instructions, number of pages, language mechanics, uniformity of graphic design, and placement inconsistency.

DESIGN TRENDS IN CHILDREN'S BOOK ILLUSTRATION - AN EXAMPLE OF LATGALIAN ABC DESIGN

Children's book illustrations have always been an area that allows artists to make the most of their creative potential. Modern design trends for children's books are changing both in the world and in Latvia. The most relevant of these are minimalism and color purity, simplicity and quality. Research also highlights the choice of illustration and font style in favor of retro style and handwriting. Picture books on the topics of kindness and love are popular in terms of content. Activity books are useful for interactive communication. The aim of the article is to develop the concept of graphic design of the ABC of Latgalian language by studying design trends in children's book illustration and the supply and demand of children's Latgalian literature in Latvia. Research methods: theoretical - research and analysis of literature and Internet resources, empirical - questionnaire, analysis of analogues.

Few-shot Font Generation with Localized Style Representations and Factorization

Automatic few-shot font generation is a practical and widely studied problem because manual designs are expensive and sensitive to the expertise of designers. Existing few-shot font generation methods aim to learn to disentangle the style and content element from a few reference glyphs, and mainly focus on a universal style representation for each font style. However, such approach limits the model in representing diverse local styles, and thus makes it unsuitable to the most complicated letter system, e.g., Chinese, whose characters consist of a varying number of components (often called ``radical'') with a highly complex structure. In this paper, we propose a novel font generation method by learning localized styles, namely component-wise style representations, instead of universal styles. The proposed style representations enable us to synthesize complex local details in text designs. However, learning component-wise styles solely from reference glyphs is infeasible in the few-shot font generation scenario, when a target script has a large number of components, e.g., over 200 for Chinese. To reduce the number of reference glyphs, we simplify component-wise styles by a product of component factor and style factor, inspired by low-rank matrix factorization. Thanks to the combination of strong representation and a compact factorization strategy, our method shows remarkably better few-shot font generation results (with only 8 reference glyph images) than other state-of-the-arts, without utilizing strong locality supervision, e.g., location of each component, skeleton, or strokes. The source code is available at https://github.com/clovaai/lffont.

FontRL: Chinese Font Synthesis via Deep Reinforcement Learning

Automatic generation of Chinese fonts is a valuable but challenging task in areas of AI and Computer Graphics, mainly due to the huge amount of Chinese characters and their complex glyph structures. In this paper, we propose FontRL, a novel method for Chinese font synthesis by using deep reinforcement learning. Specifically, we first train a deep reinforcement learning model to obtain the Thin-Plate Spline (TPS) transformation that is able to modify the reference stroke skeleton in a mean font style into the skeleton of a required style for each stroke of every unseen Chinese character. Afterwards, we utilize a CNN model to predict the location and scale information of these strokes, and then assemble them to get the skeleton of the corresponding character. Finally, we convert each synthesized character skeleton into the glyph image via an image-to-image translation model. Both quantitative and qualitative experimental results demonstrate the superiority of the proposed FontRL compared to the state of the art. Our code is available at https://github.com/lsflyt-pku/FontRL.

Kurdish Text Segmentation using Projection-Based Approaches

An optical character recognition (OCR) system may be the solution to data entry problems for saving the printed document as a soft copy of them. Therefore, OCR systems are being developed for all languages, and Kurdish is no exception. Kurdish is one of the languages that present special challenges to OCR. The main challenge of Kurdish is that it is mostly cursive. Therefore, a segmentation process must be able to specify the beginning and end of the characters. This step is important for character recognition. This paper presents an algorithm for Kurdish character segmentation. The proposed algorithm uses the projection-based approach concepts to separate lines, words, and characters. The algorithm works through the vertical projection of a word and then identifies the splitting areas of the word characters. Then, a post-processing stage is used to handle the over-segmentation problems that occur in the initial segmentation stage. The proposed method is tested using a data set consisting of images of texts that vary in font size, type, and style of more than 63,000 characters. The experiments show that the proposed algorithm can segment Kurdish words with an average accuracy of 98.6%.

Mobile Application Development to Measure Smartphone Screen Readability

This study aimed to develop an application which can be used to measure the response time needed by subjects to find a set of predetermined “words” in order to determine the level of readability of smartphone screen. The development steps began with concept development, interface design and software testing and verification. The results of this study is a software which can be used for typography experiment in which the subject is asked to find a predetermined word. The software could be used for the experiment of 4 typography factors i.e font style (FS), font size (FZ), contrast (CT), and screen rotation (SR). This software could be used for the 5x3x2x2 factorial experiment with 60 combinations. The response time of subject for each treatment combination has been computed and saved in the smartphone. Finally, the response time could be used to determine the best combination of factor and level factor for readability of smartphone screen.

Decoupled Representation Learning for Character Glyph Synthesis

Character glyph synthesis is still an open challenging problem, which involves two related aspects, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i.e.</i> , font style transfer and content consistency. In this paper, we propose a novel model named FontGAN, which integrates the character structure stylization, de-stylization and texture transfer into a unified framework. Specifically, we decouple character images into style representation and content representation, which offers fine-grained control of these two types of variables, thus improving the quality of the generated results. To effectively capture the style information, a style consistency module (SCM) is introduced. Technically, SCM exploits category-guided Kullback-Leibler divergence to explicitly model the style representation into different prior distributions. In this way, our model is capable of implementing transformations between multiple domains in one framework. In addition, we propose content prior module (CPM) to provide content prior for the model to guide the content encoding process and alleviates the problem of stroke deficiency during structure de-stylization. Benefiting from the idea of decoupling and regrouping, our FontGAN suffices to achieve many-to-many translation tasks for glyph structure. Experimental results demonstrate that the proposed FontGAN achieves the state-of-the-art performance in character glyph synthesis.

Selection of Malay Language Reading Material Based on Readability Factors

In this paper the author discuss the methods that can applied to select reading material that is appropriate to the reading ability of students. The old assumption that books or materials that written for a certain level is appropriate for all students in that particular level, needs to be changed. This is because students who isin a certain level have different reading abilities. Therefore, this paper will introduce the understanding of readability factors as one way of selecting appropriate reading material for students. The level of readability for reading materialsare influenced by a variety of factors, including the reader, environment and reading material. This article focuses on reading material factors such as clear font size and style, illustration and colour, vocabulary and sentence structure. Further, this article will present text difficulty analysis through the analysis of vocabularies. This paper conclude that vocabulary factor should not be overlooked and teachers must understand this because reading materials can be difficult when vocabulary that appear in the reading material does not fit to the level of the reader.

Few-Shot Image Synthesis using Noise-Based Deep Conditional Generative Adversarial Nets

In recent years research on automatic font generation with machine learning mainly focus on using transformation-based methods, in comparison, generative model-based methods of font generation have received less attention. Transformation-based methods learn a mapping of the transformations from an existing input to a target. This makes them ambiguous because in some cases a single input reference may correspond to multiple possible outputs. In this work, we focus on font generation using the generative model-based methods which learn the buildup of the characters from noise-to-image. We propose a novel way to train a conditional generative deep neural model so that we can achieve font style control on the generated font images. Our research demonstrates how to generate new font images conditioned on both character class labels and character style labels when using the generative model-based methods. We achieve this by introducing a modified generator network which is given inputs noise, character class, and style, which help us to calculate losses separately for the character class labels and character style labels. We show that adding the character style vector on top of the character class vector separately gives the model rich information about the font and enables us to explicitly specify not only the character class but also the character style that we want the model to generate.

Use of PowerPoint Presentation by the Teachers for Instruction in Selected Dental Colleges of Bangladesh

This descriptive type of cross-sectional study was conducted to find out the views of undergraduate dental college teachers about the PowerPoint presentations as instructional media used by teachers in dental colleges of Bangladesh. The study period was from January 2018 to December 2018. The study was carried out among the teachers of all 4 phases of BDS course of 2 government and 2 non-government dental colleges of Bangladesh. Out of 4 dental colleges, 2 were within Dhaka and 2 were outside Dhaka. The sample size was 100 for teachers. Convenience sampling technique was adopted for data collection. Self-administered semi-structured questionnaires using five points Likert scale were used for collections of data from teachers. The study revealed that 72% teachers consider Eye appealing appearance for choosing PowerPoint slides and 58% teachers did not use master slide. For selection of font style, 67% teacher agreed they consider readability, 88% agreed they consider visibility from last bench of classroom to select font size but 61% teacher disagree that they consider background for selecting font color. The study recommended, during the preparation of a PowerPoint presentation by teachers some areas like using the master slide, background, the number of words and lines need attention.&#x0D; Bangladesh Journal of Medical Education Vol.12(1) 2021: 10-15

AUTOMATIC NUMBER PLATE RECOGNITION FOR DIFFERENT FONTS AND NON-ROMAN SCRIPT

According to survey taken the total number of vehicles in [1] India were 260 million. Therefore, there is a need to develop Automatic Number Plate Recognition (ANPR) systems [1] in India because of the large number of vehicles travelling on the roads. [1] It would also help in proper tracking of the vehicles, traffic examining, finding stolen vehicles, supervising parking toll and imposing strict actions against red light breaching. Automatic number plate recognition is image processing technique for finding number plate from image and extracting characters from detected number plate. ANPR in India has always been challenging due to different lighting conditions, changes in fonts, shapes, angles, letters size, number of lines and padding between lines, different languages used. In our project we proposed a model that can detects number plate with considering all irregularities. this system uses Computer vision and machine learning technology in order to detect number plate from image. In our proposed system number plate can be of different fonts and non-roman script. For identification of characters from number plate we use OCR (Optical character recognition) technique. OCR involves two parts: Character segmentation and Character Recognition. This OCR system can be used to extract characters of different fonts and non-roman script. The Quality of OCR depends on the quality of image, image contrast, text font style and size. To improve quality of OCR we can use image processing technique to enhance quality of image.

Collaborative Online Professional Development Design

COVID-19 has impacted not only higher education teaching practice, but also professional development (PD) such as F2F workshop sessions for supervisor training have had to pivot to online modes. This presentation overviews the process of designing an academic supervision training workshop as an online module that can be used as a design framework for other professional development and training contexts. We used a design-based research methodology (McKenney &amp; Reeves, 2019) encompassing three stages.&#x0D; Stage 1 Evaluation and Analysis&#x0D; We formed a collaborative design team of two academics and a professional staff member and met regularly over Zoom as due to COVID-19 we were all working from home, and across countries (Australia and New Zealand). We began with an evaluation of the existing introductory supervision online module to identify key design elements and refinements that we could integrate into the new refresher module. This was followed by a collaborative definition of the scope of the new PD module that was focused upon providing a ‘Refresher Course’ for experienced higher degree research supervisors. We decided to focus the content of the module upon authentic user-generated scenarios from highly experienced academics across the university. User reflection was designed through an optional link to contribute to an institutional Wiki page to provide tips and comments from user experiences in supervision. The analysis of users needing an informative but short time investment in the refresher module highlighted the need for collating resources that would be referenced throughout the module to be made available at the end of the module as a downloadable interactive PDF resource for users.&#x0D; Stage 2 Collaborative Prototype Design&#x0D; An initial storyboard module prototype based upon learning object and instructional design principles (Boyle, 2003; Author1, 2007; Reigeluth et al., 2016) was created. While we began with big ideas based upon the rapid prototype of how to make the new module interactive and appealing to the users there were significant design restraints imposed by the institutional Course Builder online development platform, that provides a basic Web 1.0 approach to online module design and delivery. This meant that user interaction options were limited, file size uploads were small, and display dimensions fixed. Therefore, Vimeo was used to host and embed the video vignettes in the module pages through an institutional license providing a range of privacy and customisation options. We were able to negotiate these technical design hurdles, albeit with downgraded expectations of user interaction within our new module design.&#x0D; Stage 3 Evaluation and Redesign&#x0D; User feedback on the initial prototype was very positive and the user-generated video vignettes and funky iPad created graphics were highlights. Feedback also highlighted the need for additional user instructions, a more unified design across the module and implementation of accessibility features such as closed captioning for the video vignettes. This was achieved through maintaining a consistent font and paragraphing style and creating a new embedded player style format for the Vimeo video host platform. Closed captions were added by manually transcribing and timecoding the videos on Vimeo.

Challenges of Deep Learning-based Text Detection in the Wild

The reported accuracy of recent state-of-the-art text detection methods, mostly deep learning approaches, is in the order of 80% to 90% on standard benchmark datasets. These methods have relaxed some of the restrictions of structured text and environment (i.e., "in the wild") which are usually required for classical OCR to properly function. Even with this relaxation, there are still circumstances where these state-of-the-art methods fail. Several remaining challenges in wild images, like in-plane-rotation, illumination reflection, partial occlusion, complex font styles, and perspective distortion, cause exciting methods to perform poorly. In order to evaluate current approaches in a formal way, we standardize the datasets and metrics for comparison which had made comparison between these methods difficult in the past. We use three benchmark datasets for our evaluations: ICDAR13, ICDAR15, and COCO-Text V2.0. The objective of the paper is to quantify the current shortcomings and to identify the challenges for future text detection research.

Separating Content from Style Using Adversarial Learning for Recognizing Text in the Wild

Scene text recognition is an important task in computer vision. Despite tremendous progress achieved in the past few years, issues such as varying font styles, arbitrary shapes and complex backgrounds etc. have made the problem very challenging. In this work, we propose to improve text recognition from a new perspective by separating the text content from complex backgrounds, thus making the recognition considerably easier and significantly improving recognition accuracy. To this end, we exploit the generative adversarial networks (GANs) for removing backgrounds while retaining the text content . As vanilla GANs are not sufficiently robust to generate sequence-like characters in natural images, we propose an adversarial learning framework for the generation and recognition of multiple characters in an image. The proposed framework consists of an attention-based recognizer and a generative adversarial architecture. Furthermore, to tackle the issue of lacking paired training samples, we design an interactive joint training scheme, which shares attention masks from the recognizer to the discriminator, and enables the discriminator to extract the features of each character for further adversarial training. Benefiting from the character-level adversarial training, our framework requires only unpaired simple data for style supervision. Each target style sample containing only one randomly chosen character can be simply synthesized online during the training. This is significant as the training does not require costly paired samples or character-level annotations. Thus, only the input images and corresponding text labels are needed. In addition to the style normalization of the backgrounds, we refine character patterns to ease the recognition task. A feedback mechanism is proposed to bridge the gap between the discriminator and the recognizer. Therefore, the discriminator can guide the generator according to the confusion of the recognizer, so that the generated patterns are clearer for recognition. Experiments on various benchmarks, including both regular and irregular text, demonstrate that our method significantly reduces the difficulty of recognition. Our framework can be integrated into recent recognition methods to achieve new state-of-the-art recognition accuracy.

Investigating the relationship of semiotics associated with colour and font of Google logo with brand perception

Congruency between meanings exuded by logo elements and viewer's emotion would result in positive brand perception. Google logo has experimented with different colours and text styles since its inception. This study in first step examined semiotics reflected by all different versions of colour and font of Google. The findings of a qualitative study suggested that different colours and font evoke different emotions. Secondly, influence of semiotics on brand aesthetics and brands' ability to capture attention aspects of brand perception was studied by conducting an experimental study. Lastly, in step 3 combined effects of identified colours and font style were judged on two aspects of brand perception taken together. Results of MANOVA revealed Google logo in multi-colour with sans serif font style had highest contribution in evoking perception of aesthetics and making logo appealing through semiotics of modern, innovative, playful, diverse, action and attractive.

Using the Kano model to analyze the user interface needs of middle-aged and older adults in mobile reading

As society ages and mobile technology develops, the percentage of middle-aged and older adults using a smart phone for mobile reading is increasing. However, existing mobile reading apps have not been designed with middle-aged and older users in mind. This study investigated 14 mobile reading apps targeting users in Taiwan and evaluated middle-aged and older adults’ needs for user interfaces. Usability interface comprises the aspects of font style, navigation, page turning, and text to speech, whereas ease-of-understanding interface comprises the aspects of infographics, category, and item information. This study designed a questionnaire and obtained 187 valid responses from middle-aged and older adults. The Kano model was used to analyze the participants’ needs. Research results indicated that font style was the only one-dimensional attribute, whereas the remaining interfaces were attractive attributes. The participants reported that the two design interfaces of font style list and adjustable line spacing button met their aging visual needs. This study also found that Line Today and Facebook were the most popular reading apps for middle-aged and older adults in Taiwan. Although the platforms’ interfaces did not completely meet the preferences of most middle-aged and older adults, using these interfaces provided a pleasant experience for them.