Font Image Research Articles

FontFusionGAN: Refinement of Handwritten Fonts by Font Fusion

Handwritten fonts possess unique expressive qualities; however, their clarity often suffers because of inconsistent handwriting. This study introduces FontFusionGAN (FFGAN), a novel method that enhances handwritten fonts by mixing them with printed fonts. The proposed approach leverages a generative adversarial network (GAN) to synthesize fonts that combine the desirable features of both handwritten and printed font styles. Training a GAN on a comprehensive dataset of handwritten and printed fonts enables it to produce legible and visually appealing font samples. The methodology was applied to a dataset of handwriting fonts, showing substantial enhancements in the legibility of the original fonts, while retaining their unique aesthetic essence. Unlike the original GAN setting where a single noise vector is used to generate a sample image, we randomly selected two noise vectors, z1 and z2, from a Gaussian distribution to train the generator. Simultaneously, we input a real image into the fusion encoder for exact reconstruction. This technique ensured the learning of style mixing during training. During inference, we provided the encoder with two font images, one handwritten and the other printed font, to obtain their respective latent vectors. Subsequently, the latent vector of the handwritten font image was injected into the first five layers of the generator, whereas the latent vector of the printed font image was injected into the last two layers to obtain a refined handwritten font image. The proposed method has the potential to improve the readability of handwritten fonts, offering benefits across diverse applications, such as document composition, letter writing, and assisting individuals with reading and writing difficulties.

Visual Normalization of Handwritten Chinese Characters Based on Generative Adversarial Networks

Compared with printed Chinese characters, handwritten Chinese characters are affected by writing coherence and anthropic factors, resulting in differences in stroke deformation, displacement, tilt, length, and thickness. This consequently affects the recognition of handwritten Chinese characters. A method based on generative adversarial networks (GANs) of Chinese character handwriting-to-printing font conversion is proposed herein, where Chinese handwriting-to-printing font conversion is represented as a font style conversion and regarded as the optimal state of normalization of handwritten Chinese characters. First, an encoder–decoder is integrated into a generative adversarial network, and a symmetric network extracts handwritten font multi-scale information. Subsequently, U-Net with skip connection is used to extract deep features and reduce the large amount of low-level information shared between the input and output. Furthermore, the skip connection reduces information loss during down-sampling. Finally, the integration loss measures the differences in the character structure and font style between the generated Chinese character image and the target font image by combining font style, encoding consistency, and [Formula: see text] losses. An experiment to convert Chinese character handwriting to multiple printing fonts is performed on the CASIA and CASIA-HWDB1.1 datasets, in which image pixel difference and character recognition accuracy are used as evaluation metrics. The obtained results show a more normalized visual writing style and contribute to recognition improvement, thereby verifying the effectiveness of this method.

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.

SkelGAN: A Font Image Skeletonization Method

In this research, we study the problem of font image skeletonization using an end-to-end deep adversarial network, in contrast with the state-of-the-art methods that use mathematical algorithms. Several studies have been concerned with skeletonization, but a few have utilized deep learning. Further, no study has considered generative models based on deep neural networks for font character skeletonization, which are more delicate than natural objects. In this work, we take a step closer to producing realistic synthesized skeletons of font characters. We consider using an end-to-end deep adversarial network, SkelGAN, for font-image skeletonization, in contrast with the state-of-the-art methods that use mathematical algorithms. The proposed skeleton generator is proved superior to all well-known mathematical skeletonization methods in terms of character structure, including delicate strokes, serifs, and even special styles. Experimental results also demonstrate the dominance of our method against the state-of-the-art supervised image-to-image translation method in font character skeletonization task.

F2PNet: font‐to‐painting translation by adversarial learning

For Chinese font images, when all their strokes are replaced by pattern elements such as flowers and birds, they become flower–bird character paintings, which are traditional Chinese art treasures. The generation of flower–bird painting requires professional painters’ great efforts. How to automatically generate these paintings from font images? There is a huge gap between the font domain and the painting domain. Although many image-to-image translation frameworks have been proposed, they are unable to handle this situation effectively. In this study, a novel method called font-to-painting network (F2PNet) is proposed for font-to-painting translation. Specifically, an encoder equipped with dilated convolutions extracts features of the font image, and then the features are fed into the domain translation module for mapping the font feature space to the painting feature space. The acquired features are further adjusted by the refinement module and utilised by the decoder to obtain the target painting. The authors apply adversarial loss and cycle-consistency loss to F2PNet and further propose a loss term, which is called recognisability loss and makes the generated painting have font-level recognisability. It is proved by experiments that F2PNet is effective and can be used as an unsupervised image-to-image translation framework to solve more image translation tasks.

DropRegion training of inception font network for high-performance Chinese font recognition

Chinese font recognition (CFR) has gained significant attention in recent years. However, due to the sparsity of labeled font samples and the structural complexity of Chinese characters, CFR is still a challenging task. In this paper, a DropRegion method is proposed to generatea large number of stochastic variant font samples whose local regions are selectively disrupted and an inception font network (IFN) with two additional convolutional neural network (CNN) structure elements, i.e., a cascaded cross-channel parametric pooling (CCCP) and global average pooling, is designed. Because the distribution of strokes in a font image is non-stationary, an elastic meshing technique that adaptively constructs a set of local regions with equalized information is developed. Thus, DropRegion is seamlessly embedded in the IFN, which enables end-to-end training; the proposed DropRegion-IFN can be used for high performance CFR. Experimental results have confirmed the effectiveness of our new approach for CFR.

Sub-Pixel Mapping Based on MAP Model and Spatial Attraction Theory for Remotely Sensed Image

There are a lot of mixed pixels in the remotely sensedimagery, which can seriously limit the utility of classification. Sub-pixel mapping (SPM) is a promising techniqueto solve this problem. It can generate a fine resolution land covermap from coarse resolution fractional images by predicting the spatiallocations of different land cover classes at sub-pixel scale.However, the accuracy and detail are always limited. Especiallywhen the scale factor is large among sub-pixels per pixel, thedata volumes are amplified and the sub-pixel distribution becomescomplex. The traditional methods are carried out only by the fractionsof land cover and the spatial dependence theory, which cannot satisfythe requirement of the SPM. For avoiding the above flaw, a newSPM method based on maximum a posteriori (MAP) model withsub-pixel/pixel spatial attraction theory aimed at the largescale factor is proposed. First, MAP is proposed to improve theresolution of the fractional images and obtain the initial sub-pixellocations; after that, the pixel swapping algorithm is used tooptimize and produce the final SPM result. In this paper, theproposed model is tested by a simple simulated font image and real remotelysensed imagery, which can both demonstrate that it can outperformtraditional algorithm with a more accurate sub-pixel scale land covermap.

Fuzzy-Based Segmentation for Variable Font-Sized Text Extraction from Images/Videos

Textual information embedded in multimedia can provide a vital tool for indexing and retrieval. A lot of work is done in the field of text localization and detection because of its very fundamental importance. One of the biggest challenges of text detection is to deal with variation in font sizes and image resolution. This problem gets elevated due to the undersegmentation or oversegmentation of the regions in an image. The paper addresses this problem by proposing a solution using novel fuzzy-based method. This paper advocates postprocessing segmentation method that can solve the problem of variation in text sizes and image resolution. The methodology is tested on ICDAR 2011 Robust Reading Challenge dataset which amply proves the strength of the recommended method.

A new pixels flipping method for huge watermarking capacity of the invoice font image.

Invoice printing just has two-color printing, so invoice font image can be seen as binary image. To embed watermarks into invoice image, the pixels need to be flipped. The more huge the watermark is, the more the pixels need to be flipped. We proposed a new pixels flipping method in invoice image for huge watermarking capacity. The pixels flipping method includes one novel interpolation method for binary image, one flippable pixels evaluation mechanism, and one denoising method based on gravity center and chaos degree. The proposed interpolation method ensures that the invoice image keeps features well after scaling. The flippable pixels evaluation mechanism ensures that the pixels keep better connectivity and smoothness and the pattern has highest structural similarity after flipping. The proposed denoising method makes invoice font image smoother and fiter for human vision. Experiments show that the proposed flipping method not only keeps the invoice font structure well but also improves watermarking capacity.

An image based design support system for web page design

This paper proposes an image based design support system for web page design, which enables the designed web page to have a unified impression in terms of text colors, background colors and text fonts. Previous approaches for web page design support often focus on the automated composition of HTML (HyperText Markup Language) codes and easy insertion of graphic figures and/or icons. However, even when the same textual information is contained in web pages, their impressions can vary depending on the utilized colors and fonts. Image scales, which are originally proposed for capturing human impression of colors, are incorporated in our approach to support web page design by reflecting user's image for web pages. Furthermore, the relationship between fonts and impressions is assessed to construct a font image scale so that colors and fonts can be uniformly dealt with in terms of their images. In our approach a user is required to register preferable pages, which are used to acquire his/her image for web pages. The acquired image is then used to modify colors and fonts so that the designed web page comes to have a consistent image. A prototype system with the proposed method has been implemented using Java and Prolog. Experiments were conducted to investigate the effectiveness of our system. The results are encouraging and show that it is worthwhile following this path.