Pen Tip Motion Prediction for Handwriting Drawing Order Recovery using Deep Neural Network

Zhao, Bocheng; Yang, Minghao; Tao, Jianhua

doi:10.1109/icpr.2018.8546086

Cited by 15 publications

(31 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Table 2 lists the ms-DOR DTW results of the proposed DOR for Cl.1 to Cl.5, where Φ(Cl.i) is the characters' number contained in set Cl.i. We can see from Table 2 the proposed DOR model outperforms [9] on ms-DOR accuracy at least 2.2% and outperforms [22] at least 19.8% for Cl.1. For other levels of Chinese characters, the proposed DOR model also yields obviously better accuracies compared with [9,22] on the same Chinese handwriting dataset OLHWDB 1.1.…”

Section: Multi-stroke Dor For Chinese Charactersmentioning

confidence: 88%

“…Using the results given by [9] as a baseline, we continue to discuss the performance of each step in DOR. The models, incuding Im2Seq + Seq2Order + PbC/IbC obtain obvious higher performances than those of [22] and baseline. The ms-DOR accuracies of Im2Seq + Seq2Order and Im2Seq+Seq2Order+IbC gradually increase 0.8%, 2.2% on Cl.1 and 2.6%, 3.9% on Cl.2 respectively from SEN+DEN [9].…”

Section: Multi-stroke Dor For Chinese Charactersmentioning

confidence: 91%

“…These two datasets are widely recognized online handwriting database. Being similar to [9,22], all images are normalized into 48×48 resolution, and the strcutrue of Img2Seq-CNN is similar to the DEN proposed in [9] and we adopt the structure proposed in [23] as the GRU-RNN in Seq2Order step using MindSpore framework [24].…”

Section: Experiments Settingmentioning

confidence: 99%

See 2 more Smart Citations

Drawing Order Recovery from Trajectory Components

Yang

Zhou

Sun

et al. 2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

In spite of widely discussed, drawing order recovery (DOR) from static images is still a great challenge task. Based on the idea that drawing trajectories are able to be recovered by connecting their trajectory components in correct orders, this work proposes a novel DOR method from static images. The method contains two steps: firstly, we adopt a convolution neural network (CNN) to predict the next possible drawing components, which is able to covert the components in images to their reasonable sequences. We denote this architecture as Im2Seq-CNN; secondly, considering possible errors exist in the reasonable sequences generated by the first step, we construct a sequence to order structure (Seq2Order) to adjust the sequences to the correct orders. The main contributions include: (1) the Img2Seq-CNN step considers DOR from components instead of traditional pixels one by one along trajectories, which contributes to static images to component sequences; (2) the Seq2Order step adopts image position codes instead of traditional points' coordinates in its encoder-decoder gated recurrent neural network (GRU-RNN). The proposed method is experienced on two well-known open handwriting databases, and yields robust and competitive results on handwriting DOR tasks compared to the state-of-arts.

show abstract

Section: Multi-stroke Dor For Chinese Charactersmentioning

confidence: 88%

Section: Multi-stroke Dor For Chinese Charactersmentioning

confidence: 91%

See 1 more Smart Citation

Drawing Order Recovery from Trajectory Components

Yang

Zhou

Sun

et al. 2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Bhunia et al [16] used a CNN and RNNbased Encoder-Decoder network for handwriting trajectory recovery. Attempts were also made using neural networks to identify graph features [17] and for sequential stroke prediction using regression CNNs [18].…”

Section: Related Workmentioning

confidence: 99%

Modality Conversion of Handwritten Patterns by Cross Variational Autoencoders

Sumi

Iwana

Hayashi

et al. 2019

2019 International Conference on Document Analysis and Recognition (ICDAR)

View full text Add to dashboard Cite

This research attempts to construct a network that can convert online and offline handwritten characters to each other. The proposed network consists of two Variational Auto-Encoders (VAEs) with a shared latent space. The VAEs are trained to generate online and offline handwritten Latin characters simultaneously. In this way, we create a crossmodal VAE (Cross-VAE). During training, the proposed Cross-VAE is trained to minimize the reconstruction loss of the two modalities, the distribution loss of the two VAEs, and a novel third loss called the space sharing loss. This third, space sharing loss is used to encourage the modalities to share the same latent space by calculating the distance between the latent variables. Through the proposed method mutual conversion of online and offline handwritten characters is possible. In this paper, we demonstrate the performance of the Cross-VAE through qualitative and quantitative analysis.

show abstract

“…To cope with this, Qiao and Yasuhara [7] search for the best writing paths between handwritten strokes and the pre-defined template strokes, but it still suffers from excessively distorted and intersected strokes. Recently, Zhao et al [8,9] proposed a CNN-based model (named DEN), which predicts the probability of the next stroke point position from previous frames of part-drawn handwriting images. Despite its ability of modeling instant writing states from offline handwritten character images, DEN ignores the ambiguous zones caused by the intersections of different strokes.…”

Section: Introductionmentioning

confidence: 99%

Handwriting Trajectory Recovery from Off-Line Multi-Stroke Characters by Deep Ordering Prediction and Heuristic Search

Wang

Liu

2021

2021 IEEE International Conference on Multimedia and Expo (ICME)

View full text Add to dashboard Cite

Stroke order recovery from off-line multi-stroke characters is a great challenge due to the ambiguity in intersection and connection among strokes. In this paper, we propose a novel framework for handwriting trajectory recovery from off-line handwritten characters by deep neural network (DNN) based ordering prediction and heuristic search, where several DNN modules are designed to extract stroke skeleton, ambiguous zones and starting points, respectively. Then, the ordering matrix M o among all the stroke segments is calculated by a pointer network (Ptr-Net). Besides, a convolutional neural network (CNN) is used to measure the time adjacency between two arbitrary segments. Based on these necessary measurements, the final writing order is decided via searching for the optimal permutation by heuristic A * search. Experiments on handwriting images synthesized from the public online handwriting datasets CASIA-OLHWDB1.1, ICDAR13-Online and UNIPEN show that the proposed method yields superior performance on Chinese and English/Arabic handwriting.

show abstract

Pen Tip Motion Prediction for Handwriting Drawing Order Recovery using Deep Neural Network

Cited by 15 publications

References 16 publications

Drawing Order Recovery from Trajectory Components

Drawing Order Recovery from Trajectory Components

Modality Conversion of Handwritten Patterns by Cross Variational Autoencoders

Handwriting Trajectory Recovery from Off-Line Multi-Stroke Characters by Deep Ordering Prediction and Heuristic Search

Contact Info

Product

Resources

About