Abstract: Optical Character recognition is an eminent topic of research in modern times. An exhaustive research is going on character recognition of different languages. These languages mainly include English and Devanagiri in India. A lot of research has been carried out already in the English language. Devanagiri consists of over 120 regional languages, which is a current topic of research nowadays. Character recognition is done on two types of documents typed and handwritten. In case of Indian and other languages OCR systems are not yet able to recognize the characters successfully with 100 percent reliability because of variation in scripts, quality, size, font, and style. Various algorithms are being developed now days for increasing the reliability of these characters for accurate recognition.

Abstract: Few-shot font generation (FFG), which aims to generate a new font with a few examples, is gaining increasing attention due to the significant reduction in labor cost. A typical FFG pipeline considers characters in a standard font library as content glyphs and transfers them to a new target font by extracting style information from the reference glyphs. Most existing solutions explicitly disentangle content and style of reference glyphs globally or component-wisely. However, the style of glyphs mainly lies in the local details, i.e. the styles of radicals, components, and strokes together depict the style of a glyph. Therefore, even a single character can contain different styles distributed over spatial locations. In this paper, we propose a new font generation approach by learning 1) the fine-grained local styles from references, and 2) the spatial correspondence between the content and reference glyphs. Therefore, each spatial location in the content glyph can be assigned with the right fine-grained style. To this end, we adopt cross-attention over the representation of the content glyphs as the queries and the representations of the reference glyphs as the keys and values. Instead of explicitly disentangling global or component-wise modeling, the cross-attention mechanism can attend to the right local styles in the reference glyphs and aggregate the reference styles into a fine-grained style representation for the given content glyphs. The experiments show that the proposed method outperforms the state-of-the-art methods in FFG. In particular, the user studies also demonstrate the style consistency of our approach significantly outperforms previous methods.

Abstract: In our age of ubiquitous digital displays, adults often read in short, opportunistic interludes. In this context of Interlude Reading, we consider if manipulating font choice can improve adult readers’ reading outcomes. Our studies normalize font size by human perception and use hundreds of crowdsourced participants to provide a foundation for understanding, which fonts people prefer and which fonts make them more effective readers. Participants’ reading speeds (measured in words-per-minute (WPM)) increased by 35% when comparing fastest and slowest fonts without affecting reading comprehension. High WPM variability across fonts suggests that one font does not fit all. We provide font recommendations related to higher reading speed and discuss the need for individuation, allowing digital devices to match their readers’ needs in the moment. We provide recommendations from one of the most significant online reading efforts to date. To complement this, we release our materials and tools with this article.

Abstract: Generating a new font library is a very labor-intensive and time-consuming job for glyph-rich scripts. Few-shot font generation is thus required, as it requires only a few glyph references without fine-tuning during test. Existing methods follow the style-content disentanglement paradigm and expect novel fonts to be produced by combining the style codes of the reference glyphs and the content representations of the source. However, these few-shot font generation methods either fail to capture content-independent style representations, or employ localized component-wise style representations, which is insufficient to model many Chinese font styles that involve hyper-component features such as inter-component spacing and “connected-stroke”. To resolve these drawbacks and make the style representations more reliable, we propose a self-supervised cross-modality pre-training strategy and a cross-modality transformer-based encoder that is conditioned jointly on the glyph image and the corresponding stroke labels. The cross-modality encoder is pre-trained in a self-supervised manner to allow effective capture of cross- and intra-modality correlations, which facilitates the content-style disentanglement and modeling style representations of all scales (strokelevel, component-level and character-level). The pretrained encoder is then applied to the downstream font generation task without fine-tuning. Experimental comparisons of our method with state-of-the-art methods demonstrate our method successfully transfers styles of all scales. In addition, it only requires one reference glyph and achieves the lowest rate of bad cases in the few-shot font generation task (28% lower than the second best).

Abstract: Automatic font generation remains a challenging research issue due to the large amounts of characters with complicated structures. Typically, only a few samples can serve as the style/content reference (termed few-shot learning), which further increases the difficulty to preserve local style patterns or detailed glyph structures. We investigate the drawbacks of previous studies and find that a coarsegrained discriminator is insufficient for supervising a font generator. To this end, we propose a novel Component-Aware Module (CAM), which supervises the generator to decouple content and style at a more fine-grained level, i.e., the component level. Different from previous studies struggling to increase the complexity of generators, we aim to perform more effective supervision for a relatively simple generator to achieve its full potential, which is a brand new perspective for font generation. The whole framework achieves remarkable results by coupling component-level supervision with adversarial learning, hence we call it Component-Guided GAN, shortly CG-GAN. Extensive experiments show that our approach outperforms state-of-the-art one-shot font generation methods. Furthermore, it can be applied to handwritten word synthesis and scene text image editing, suggesting the generalization of our approach.

Abstract: The font size effect is a metamemory illusion in which larger-font items produce higher judgments of learning (JOLs) but not better memory, relative to smaller-font items. We conducted meta-analyses to determine what is currently known about how font size affects JOLs and memory accuracy. In addition, we implemented both univariate and multivariate meta-regressions to isolate the moderators of JOL effects and memory effects. The results revealed a small-to-moderate effect of font size on JOLs. There was also a small but significant effect of font size on memory. This suggests that JOLs and memory accuracy both increase with font size, rather than being completely dissociated. Moreover, JOL-memory dissociation only occurred when font size ranged between very small and intermediate. Our working explanation is that the memory effects of font size are tied to (dis)fluency, but its JOL effects are not. Some boundary conditions were identified for font size effects on both JOLs and memory. Specifically, larger font sizes only reliably increased both JOLs and memory accuracy (a) when font sizes ranged from intermediate to very large, (b) when study materials were unrelated word lists, (c) when JOLs were solicited immediately after encoding, and (d) when study time was relatively brief.The online version contains supplementary material available at 10.1007/s11409-021-09287-3.

Abstract: We develop a method to automatically segment a font’s glyphs into a set of overlapping and intersecting strokes with the aim of generating artistic stylizations. The segmentation method relies on a geometric analysis of the glyph’s outline, its interior, and the surrounding areas and is grounded in perceptually informed principles and measures. Our method does not require training data or templates and applies to glyphs in a large variety of input languages, writing systems, and styles. It uses the medial axis, curvilinear shape features that specify convex and concave outline parts, links that connect concavities, and seven junction types. We show that the resulting decomposition in strokes can be used to create variations, stylizations, and animations in different artistic or design-oriented styles while remaining recognizably similar to the input font.

Abstract: Topic of exhaustive study for about past decades has been carried out in machine imitation of human reading. a small number of investigates have been accepted on the detection of cursive font writing like Arabic texts for its individual challenge and difficulty .In this work, a novel technique for automatic Arabic font recognition is proposed to demonstrate an suitable recognition rate for multi fonts styles and multi sizes of Arabic word images. The scheme can be classified into a number of steps. First, segmenting Arabic line into words depending on the vertical projection and dynamic threshold then we implicated each Arabic word as a class by ignoring segmenting the word into characters .Second ,normalizing step, the size of Arabic word images varies from each other .The system converts the images that contribution into a new size that is divisible by "N" without remainder, to decrease the difficulty of feature extraction and recognition of the system that may allow images from different resources, Third, feature extraction step which is based on apply the ratio of vertical sliding strips as a features. Finally, multi class support vector machine (one versus one technique)is used as a classifier .This method was estimated on off line printed fonts, five Arabic fonts, (Andalus, Arial, Simplified Arabic, Tahoma and Traditional Arabic) were used and the average recognition rate of all fonts was 95.744%.

Abstract: The amount of text people need to read and understand grows daily. Software defaults, designers, or publishers often choose the fonts people read in. However, matching individuals with a faster font could help them cope with information overload. We collaborated with typographers to (1) select eight fonts designed for digital reading to systematically compare their effectiveness and to (2) understand how font and reader characteristics affect reading speed. We collected font preferences, reading speeds, and characteristics from 252 crowdsourced participants in a remote readability study. We use font and reader characteristics to train FontMART, a learning to rank model that automatically orders a set of eight fonts per participant by predicted reading speed. FontMART’s fastest font prediction shows an average increase of 14–25 WPM compared to other font defaults, without hindering comprehension. This encouraging evidence provides motivation for adding our personalized font recommendation to future interactive systems.

Abstract: Previous works on font generation mainly focus on the standard print fonts where character's shape is stable and strokes are clearly separated. There is rare research on brush hand-writing font generation, which involves holistic structure changes and complex strokes transfer. To address this issue, we propose a novel GAN-based image translation model by integrating the skeleton information. We first extract the skeleton from training images, then design an image encoder and a skeleton encoder to extract corresponding features. A self-attentive refined attention module is devised to guide the model to learn distinctive features between different domains. A skeleton discriminator is involved to first synthesize the skeleton image from the generated image with a pre-trained generator, then to judge its realness to the target one. We also contribute a large-scale brush handwriting font image dataset with six styles and 15,000 high-resolution images. Both quantitative and qualitative experimental results demonstrate the competitiveness of our proposed model.

Abstract: Font detection and similar font suggestions are essential in computer vision, document analysis, pattern recognition, web development, and core work for graphics designers and UX-UI engineers. Even though Bangla is one of the most communicated languages globally and the growing popularity of Bangla in online publications and social media platforms, there has been a noticeable demand for Bangla fonts. However, there is no appreciable work for font detection in Bangla, unlike other high resource languages like English, Chinese, Hindi, Arabic, Spanish, and German. This work represents a model to recognize Bangla fonts from images using the transfer learning method. In Image Processing and Classification, resources are insufficient for such work in Bangla, so it is required to build as much raw data as possible to train the model. Therefore, as part of the work, 6500 raw images of five different fonts are used, and with augmentation 26000 image data are created to train and 2600 images to validate the model. Three transfer learning models, which are VGG-16, VGG-19, and Xception are applied. Among them, VGG-16 archives the highest accuracy of 96.23%. This paper is the first publicly available work on Bangla font recognition using the Transfer Learning approach to the best of our knowledge.

Abstract: Considerable research has demonstrated the positive effects of handwritten font styles on product attachment and word-of-mouth behavior. However, few studies examined whether these positive effects can be mitigated or even reversed. The purpose of this study is to fill this knowledge gap by identifying several boundary conditions (communal orientation, message type, and hotel type) for the positive effects of handwritten font styles. We conducted two quasi-experimental studies. In Study 1 (n = 125), the positive effect of handwritten font styles on attitude toward a hotel was not observed among individuals with a low communal orientation. In Study 2 (n = 245), the handwritten (vs. machine-written) font styles in the sustainability messages of a luxury hotel reduce warmth of the hotel. Hospitality managers should use handwritten font styles carefully depending on hotel type, message type, and customer characteristics.

Abstract: Recently, the artistic font generation has made significant progress. However, existing methods typically treat the style of artistic font as a whole. Their performance is usually limited to the artistic fonts with complex style elements in glyph and text effect. To solve these problems, this paper presents a disentangled style encoding network, termed DSE-Net, to synthesize artistic fonts. In order to obtain the disentangled text effect features, we introduce a perspective transformation network. We propose a cross-layer fusion mechanism to improve the artistic fonts' structure and texture according to their different representations in CNN. Notably, encoding different style elements for artistic font generation is a new task, so there is no publicly-accessible dataset. Therefore, a new dataset, termed SSAF, has been constructed. Extensive experiments demonstrate that our model significantly outperforms the state-of-the-art methods, with more fine-grained text effect and accurate stroke details.

Abstract: Textual information within scene images is very important in computer vision applications such as image retrieval based on its content, Tourist translator, and Navigation systems assistant. This paper presented a scene text recognition system based on YOLO. The main steps of the suggested methodology are: text detection and localization (using YOLOv5), text segmentation (using Morphological processing as a new method), features extraction (using MSERs), character segmentation and word segmentation (using bounding boxes with graph), and finally character recognition (using OCR). In this work we create a new dataset model that includes most of the text challenges such as Font type, Font size, Font color, and Font. The proposed system gives higher performance for detection, localization, and recognition when using dataset containing many challenges, the results were 80%, 96%, and 87.6 for precision, recall, and F-score respectively. Comparing with other similar works it was better. The accuracy of (OCR) is more than 99%.

Abstract: Text extraction process can play a vital role for detecting valuable information from a selected image. This text extraction process involves text detection, localization, marking, tracking, extraction, enhancement and finally recognition task. It is a difficult task to detect these text characters, because of their variation of size, style, font, orientation, alignment, contrast, color and textured background. There is a growing demand of information detection, indexing and retrieval from various multimedia documents nowadays. Several methods have been developed for extraction of text from an image. This article proposes a novel method for image to text extraction. In this paper, we are presenting a multiresolution morphology based text segmentation process suitable for various types of non-text elements like drawing, pictures, halftones or etc. For image processing, python library OpenCV is used and for text extraction Tessaract is used. Python Imaging Library (PIL) is capable to handle the opening and manipulation of images in many formats in Python. Also we are in testing of such an application that can give output in every language correctly.

Abstract: This article explores typographic placemaking by comparing the design and public launch of two city fonts: TilburgsAns (2016) and Dubai Font (2017). Building on recent work on semiotic technology and graphic ideology, the authors examine how these fonts’ visual features and the promotional discourses surrounding their launch are utilized for placemaking, and how this is facilitated and constrained by technology and ideology. The results show that the two projects of typographic placemaking build on a similar repertoire of semiotic technology, but make different use of it. The authors sustain that this difference is explained by the political aims of the two projects, on the one hand, and their economic and organizational scale, on the other. A postcolonial perspective further underlines their geopolitically and historically different preconditions.

Abstract: Chinese characters have been created into many font styles, such as official script, running script, and regular script. Other than that, some famous calligraphers, such as Ouyang Xun and Yan Zhenqing, have produced fonts with their style. Being able to detect and recognize these font styles quickly and accurately has essential applications in graphic design, page layout, handwriting identification, and other use cases. Distinguishing between font styles requires professional knowledge, which almost inevitably leads to errors for unprofessional people. Therefore, this paper presents a sword-like model based on a convolutional neural network with a sword structure to recognize font styles for Chinese characters. This model includes 15 convolutional layers. For each layer, we gradually increase the number of convolutional kernels to better extract the classification features of the input image. This paper uses four downsampling layers in the model. For each downsampling operation, the length and width of the image become half of their original values while the number of channels gradually increases, leading to a sword-like shape. As a result, we name our model as SwordNet. We also created a Chinese font dataset called the Nankai Chinese Font Style dataset and made it available on Github. Using the above dataset, we compared the accuracy of our model with six other state-of-the-art network models. The experiments showed that SwordNet could achieve an average recognition accuracy of 99.03% in multiple experiments, while the other six models can only achieve accuracy up to 94.91%. So we can conclude that SwordNet could perform better in font style recognition than other models.

Abstract: What happens if we see a suitable font for our design work but we do not know its name? Visual Font Recognition (VFR) systems are used to identify the font typeface in an image. These systems can assist graphic designers in identifying fonts used in images. A VFR system also aids in improving the speed and accuracy of Optical Character Recognition (OCR) systems. In this paper, we proposed the first publicly available datasets in the field of Persian font recognition and employed Convolutional Neural Networks (CNN) to address the Persian font recognition problem. The results show that the proposed pipeline obtained 78.0% top-1 accuracy on our new datasets, 89.1% in the IDPL-PFOD dataset, and 94.5% in the KAFD dataset. Furthermore, the average time spent in the entire pipeline for one sample of our proposed datasets is 0.54 and 0.017 seconds for CPU and GPU, respectively. We conclude that CNN methods can be used to recognize Persian fonts without the need for additional pre-processing steps such as feature extraction, binarization, normalization, etc.

Abstract: Making a new font requires graphical designs for all base characters, and this designing process consumes lots of time and human resources. Especially for languages including a large number of combinations of consonants and vowels, it is a heavy burden to design all such combinations independently. Automatic font generation methods have been proposed to reduce this labor-intensive design problem. Most of the methods are GAN-based approaches, and they are limited to generate the trained fonts. In some previous methods, they used two encoders, one for content, the other for style, but their disentanglement of content and style is not sufficiently effective in generating arbitrary fonts. Arbitrary font generation is a challenging task because learning text and font design separately from given font images is very difficult, where the font images have both text content and font style in each image. In this paper, we propose a new automatic font generation method to solve this disentanglement problem. First, we use two stacked inputs, i.e., images with the same text but different font style as content input and images with the same font style but different text as style input. Second, we propose new consistency losses that force any combination of encoded features of the stacked inputs to have the same values. In our experiments, we proved that our method can extract consistent features of text contents and font styles by separating content and style encoders and this works well for generating unseen font design from a small number of reference font images that are human-designed. Comparing to the previous methods, the font designs generated with our method showed better quality both qualitatively and quantitatively than those with the previous methods for Korean, Chinese, and English characters. e.g., 17.84 lower FID in unseen font compared to other methods.

Abstract: Styles of typefaces or fonts are often associated with specific impressions, such as heavy, contemporary, or elegant. This indicates that there are certain correlations between font shapes and their impressions. To understand the correlations, this paper realizes a shared latent space where a font and its impressions are embedded nearby. The difficulty is that the impression words attached to a font are often very noisy. This is because impression words are very subjective and diverse. More importantly, some impression words have no direct relevance to the font shapes and will disturb the realization of the shared latent space. We, therefore, use DeepSets for enhancing shape-relevant words and suppressing shape irrelevant words automatically while training the shared latent space. Quantitative and qualitative experimental results with a large-scale font-impression dataset demonstrate that the shared latent space by the proposed method describes the correlation appropriately, especially for the shape-relevant impression words.

Abstract: AbstractWearable devices have a rapid development and popularity in the last decade, and the human-computer interaction issues in this field have become prominent. In general, the screen size of wearable devices is much smaller. Besides, the usage scenario is usually outdoors under the state of moving. Consequently, it is difficult to apply the existing standards of other devices, such as laptops, on them. In this study, we used a smartwatch as the interaction device, and 45 users participated in this study. Two experiments were designed to investigate the influence of the key factors (font size, contrast, and weight) that influenced legibility of text (numbers and characters) on behavioral performance and subjective experience for two age groups (people aged 18–40 vs. people aged over 40). The results showed that the font size had the greatest influence on legibility. For people over 40, the very small size and the small size text could not be clearly recognized. For the young people aged 18–40, the font size had a subtle influence on performance. However, too large font size reduced the preference for message reading. Contrast is another variable that influence legibility. Low contrast needs to be avoided in the design. The influence of weight is relatively small, and very thin weight is worse than medium and very thick weight. In general, the method used in this study had high ecological validity and we hope to provide useful suggestions for the design of smartwatch.KeywordsText-legibilityUser experienceWearable devicesWalkingAge

Abstract: Creating a complete stylized font library that helps the audience to perceive information from the text often requires years of study and proficiency in the use of many professional tools. Accordingly, automatic stylized font generation in a deep learning-based fashion is a desirable but challenging task that has attracted a lot of attention in recent years. This paper revisits the state-of-the-art methods for stylized font generation and presents a taxonomy of the deep learning-based stylized font generation. Despite the notable performance of the existing models, stylized multilingual font generation, the task of applying specific font style to diverse characters in multiple languages has never been reported to be addressed. An efficient and economical method for stylized multilingual font generation is essential in numerous application scenarios that require communication with international audiences. We propose a solution for few-shot multilingual stylized font generation by a fast feed-forward network, Multilingual Font Generation Network (MF-Net), which can transfer previously unseen font styles from a few samples to characters from previously unseen languages. Following the Generative Adversarial Network (GAN) framework, MF-Net adopts two separate encoders in the generator to decouple a font image's content and style information. We adopt an attention module in the style encoder to extract both shallow and deep style features. Moreover, we also design a novel language complexity-aware skip connection to adaptive adjust the structural information to be preserved. With an effective loss function to improve the visual quality of the generated font images, we show the effectiveness of the proposed MF-Net based on quantitative and subjective visual evaluation, and compare it with the existing models in the scenario of stylized multilingual font generation. The source code is available on https://github.com/iamyufan/MF-Net.

Abstract: Text legibility is an important and influential aspect of multimedia, not only on usual multimedia reading devices (displays, tablets, and mobile phones) but also in three-dimensional virtual environments (VE). In VEs, text can be used for many purposes in accordance with the category of experience (gaming, architectural visualization, exhibition, education). Font type used can deteriorate the VE experience if the user has to strain his eyes so that he may correctly read what has been written. The purpose of this research is to examine different fonts in order to determine which properties of a typeface are more suitable for VE. Four different fonts have been tested in 10 font sizes at three viewing distances (5, 10, and 15 meters). Times New Roman as a representative of serif fonts, Helvetica as a non-serif font, Caveat as a representative of handwritten fonts, and Roboto as a font for digital use, initially developed for the Android operating system. Different typographic characteristics of fonts significantly affect readability in VE. Additionally, was explored how the color of the text and the background color influence text readability in 20 combinations. It is known that the relationship between text color and background color affects readability. Since VR is a relatively new medium, we wanted to determine which relationship between colors of text and background is the best and whether it follows some rules present on the web. The colors of the text were black and white, and the colors of the background were black, white and gray, red, green, pastel green, blue, pastel blue, yellow, pastel orange, and pastel pink. Results regarding text color/background color have shown that most participants preferred black text on white background, black text on a gray background, and white text on a pastel orange background. On the contrary, participants rated white text on a yellow background as the least preferable combination, accompanied by white text on a pastel green background.