Jun Sun

• Fujitsu Ltd.

With the prevalence of state-of-the-art generative models, photorealistic synthetic images can now be easily generated. However, the generated images may replicate contents from the original training images, which can lead to potential legal issues. In this paper, we propose a novel method called Conceptual-Similar-guided Self-supervised Language-...

Temporal event localisation (TEL) has recently attracted increasing attention due to the rapid development of video platforms. Existing methods are based on either fully/weakly supervised or unsupervised learning, and thus they rely on expensive data annotation and time‐consuming training. Moreover, these models, which are trained on specific domai...

Action recognition has been a popular area of computer vision research in recent years. The goal of this task is to recognise human actions in video frames. Most existing methods often depend on the visual features and their relationships inside the videos. The extracted features only represent the visual information of the current video itself and...

Blockchain technology provides a “tamper-proof distributed ledger” for its users. Typically, to ensure the integrity and immutability of the transaction data, each node in a blockchain network retains a full copy of the ledger; however, this characteristic imposes an increasing storage burden upon each node with the accumulation of data. In this pa...

Blockchain, as a distributed ledger technology, becomes more and more popular in both industry and academia. Each peer in blockchain system maintains a copy of ledger and makes sure of data consistency through consensus protocol. Blockchain system can provide many benefits such as immutability, transparency and security. Hyperledger Fabric is permi...

Blockchain is one of the most popular distributed ledger technologies. It can solve the trust issue among enterprises. Hyperledger Fabric is a permissioned blockchain aiming at enterprise-grade business applications. However, compared to traditional distributed database solutions, one issue of blockchain based application development is the limited...

Knowledge graphs represent information in the form of entities and relation-ships between them. A knowledge graph consists of multi-relational data, having entities as nodes and relations as edges. The relation indicates a relationship between these two entities. Relation extraction is the key step to construct a knowledge graph. Conventional relat...

Recognizing characters in Chinese seal images is important when researching ancient cultural artworks because the seals may contain critical historical information. However, owing to large intraclass variance and a limited number of training samples, recognizing such characters in Chinese seals is challenging. Thus, this study proposes a graph-matc...

Latent variables and reconstruction error generated from auto encoder are the common means for anomaly detection dealing with high dimensional signals. They are exclusively typical representations of the original input, and a plenty of methods utilizing them for anomaly detection have achieved good results. In this paper, we propose a new method co...

Many recent few-shot learning methods concentrate on designing novel model architectures. In this paper, we instead show that with a simple backbone convolutional network we can even surpass state-of-the-art classification accuracy. The essential part that contributes to this superior performance is an adversarial feature learning strategy that imp...

Internet of Things (IoT) technology has become more and more popular recently. However, due to the limited resources of IoT devices and the centralized system architecture, some severe issues remain difficult to solve, such as: overload of centralized server, single point of failure, and the possibility of malicious usage of personal information. B...

Text in images contains exact semantic information and the text knowledge can be utilized in many image cognition and understanding applications. The human reading habits can provide the clues of text line structure for text line extraction. In this paper, we propose a novel human reading knowledge inspired text line extraction method based on k-sh...

Air pollution is a major problem in modern cities and developing countries. Fine particulate matter (PM2.5) is a growing public health concern and become the most serious air pollution. In this study, we formulate the PM2.5 inference problem in conventional environmental sensors as a sequence-to-sequence problem. We adopt the encoder-decoder LSTM (...

In this paper, we focus on the problem of script and handwritten/machine-printed identification of texts. We simultaneously identify the script (Chinese, English, Japanese, Korean, or Russian) and whether it is handwritten or machine-printed text by designing a dual-branch structured deep convolutional neural network (CNN). For the training stage,...

Since the large size documents are usually folded for preservation, creases will occur in the scanned images. In this paper, a crease detection method is proposed to locate the crease pixels for further processing. According to the imaging process of contactless scanners, the shading on both sides of the crease usually varies a lot. Based on this o...

Nowadays the CNN is widely used in practical applications for image classification task. However the design of the CNN model is very professional work and which is very difficult for ordinary users. Besides, even for experts of CNN, to select an optimal model for specific task may still need a lot of time (to train many different models). In order...

The handwritten string recognition is still a challengeable task, though the powerful deep learning tools were introduced. In this paper, based on TAO-FCN, we proposed an end-to-end system for handwritten string recognition. Compared with the conventional methods, there is no preprocess nor manually designed rules employed. With enough labelled dat...

Like other problems in computer vision, offline handwritten Chinese character recognition (HCCR) has achieved impressive results using convolutional neural network (CNN)-based methods. However, larger and deeper networks are needed to deliver state-of-the-art results in this domain. Such networks intuitively appear to incur high computational cost,...

Like other problems in computer vision, offline handwritten Chinese character recognition (HCCR) has achieved impressive results using convolutional neural network (CNN)-based methods. However, larger and deeper networks are needed to deliver state-of-the-art results in this domain. Such networks intuitively appear to incur high computational cost,...

Nowadays the CNN is widely used in practical applications for image classification task. However the design of the CNN model is very professional work and which is very difficult for ordinary users. Besides, even for experts of CNN, to select an optimal model for specific task may still need a lot of time (to train many different models). In order...

In camera captured document image analysis, specular highlights on glossy document surfaces often have negative effects on state-of-the-art Optical Character Recognition systems. Traditional highlight removal methods rely on either texture information from a single image or specific hardware setups to fuse complementary visual cues from multiple im...

This paper presents a method for image perspective correction using camera intrinsic parameters. This method is based on two assumptions: a) the taken picture have a rectangle area, but didn't know the rectangle area's aspect ratio; b) the camera's intrinsic parameters should obtain by picture (Intrinsic parameters can easily obtain in iPhone or An...

Scanned images of historical documents often suffer from bleed-through, which refers to the ink on one side seeping through the paper and appearing on the other side. In this paper, a new conditional random fields (CRF) based method is proposed to remove the bleed-through from the scanned images of historical images. The proposed method only requir...

Google patent: https://www.google.com/patents/CN105488515A?cl=en&dq=CN105488515A The invention provides a method for training a convolutional neural network classifier and an image processing device. According to the method for training the convolutional neural network classifier, global features and local features are extracted from an image for u...

In China, efforts are being made to preserve historical documents in the form of digital data so that they can be effectively used while being protected. Stitching technology is expected to play a role in these efforts as it can be used to divide a large historical document into multiple areas, scan those areas using a compact contactless scanner,...

Because of the various appearance (different writers, writing styles, noise, etc.), the handwritten character recognition is one of the most challenging task in pattern recognition. Through decades of research, the traditional method has reached its limit while the emergence of deep learning provides a new way to break this limit. In this paper, a...

The rapid increasing popularity of micro-blogging has made it an important information seeking channel. Keyphrase extraction is an effective way for summarizing and analyzing micro-blogging content, which can help users gain insights into internet hotspots. Existing methods for keyphrase extrac-tion usually unilaterally consider phrase frequency or...

Our FRDC_QA team participated in the QA-Lab English subtask of the NTCIR-11. In this paper, we describe our system for solving real-world university entrance exam questions, which are related to world history. Wikipedia is used as the main external resource for our system. Since problems with choosing right/wrong sentence from multiple sentence cho...

We describe a document image segmentation algorithm to classify a scanned document into different regions such as text/line drawings, pictures, and smooth background. The proposed scheme is relatively independent of variations in text font style, size, intensity polarity and of string orientation. It is intended for use in an adaptive system for do...

The rapid increasing popularity of micro-blogging has made it an important information seeking channel. Keyphrase extraction is an effective way for summarizing and analyzing micro-blogging content, which can help users gain insights into internet hotspots. Existing methods for keyphrase extraction usually unilaterally consider phrase frequency or...

In this paper, we focus on robust multi-color object detection with cluttered backgrounds and variable illumination for a target application to color chart detection. The task is characterized by a wide range of color variation combined with complex background. Arbitrary placement of the chart in the scene will further complicate the detection task...

Halftoning is a necessary technique for electrophotographic printers to print continuous tone images. Scanned images obtained from such printed hard copies are corrupted by screen like artifacts called halftone patterns. Descreening aims to recover high quality continuous tone image from the scanned image. In this paper, a two-step descreening meth...

The success of convolutional neural networks (CNNs) in the field of image recognition suggests that local connectivity is one of the key issues to exploit the prior information of structured data. But the problem of selecting optimal local receptive field still remains. We argue that the best way to select optimal local receptive field is to let CN...

Paper stitching technology can reconstruct a whole paper page from two sub-images separately scanned from a camera with limited vision field. Traditional technology usually chooses a global optimal seam, and the two sub-images are stitched along it. These methods perform well on the rigid object, but when distortion exists caused by the uneven plac...

Deep learning methods have recently achieved impressive performance in the area of visual recognition and speech recognition. In this paper, we propose a handwriting recognition method based on relaxation convolutional neural network (R-CNN) and alternately trained relaxation convolutional neural network (ATR-CNN). Previous methods regularize CNN a...

The present embodiments disclose a method of and device for identifying the direction of characters in an image block. The method includes: performing optical character recognition processing on the image block by assuming various directions as assumed character directions to obtain sub image blocks, recognized characters corresponding to the sub i...

This paper proposes one modified active shape model (MASM) method to extract book inner boundaries in the scanned book images. It assumes that both pages are included in the scanned image and the book page corners are provided. The MASM method introduces the “book shape” idea and represents the book shape as one set of landmark points which are sam...

Text detection in natural scene images plays an important role in content-based image retrieval, especially user-guided text detection for human-computer interaction. In this paper, we propose a fast and accurate text detection method with user-intention in terms of tap gesture. Firstly, a user-intention slice descriptor is designed based on the es...

Currently most electrophotographic printers use halftoning technique to print continuous tone images, so scanned images obtained from such hard copies are usually corrupted by screen like artifacts. In this paper, a new model of scanned halftone image is proposed to consider both printing distortions and halftone patterns. Based on this model, an a...

From a single low resolution image, a real-time document image super-resolution algorithm is proposed to obtain high resolution document image with sharp text boundaries. First, a highly efficient document image matting algorithm based on local linear modeling is designed to decompose the input image into text, foreground and background layers, whi...

We propose an accurate and robust coarse-to-fine text detection scheme with user-intention which captures the intrinsic characteristics of natural scene texts. In the coarse detection stage, a double edge detector is designed to estimate the symmetry of stroke and the stroke width, which help segment the foreground. Then the initial user-intention...

Non-contact imaging devices such as digital cameras and overhead scanners can convert hardcopy books to digital images without cutting them to individual pages. However, the captured images have distinct distortions. A book dewarping system is proposed to remove the perspective and geometric distortions automatically from single images. A book boun...

This paper proposed a sub-structure learning based method for handwritten Chinese text recognition. In conventional methods, a standard character recognizer is trained on character classes only. Unreliable recognition results on character segments will decrease final recognition precision. By discovering stable sub-structure patterns from real char...

Method and apparatus for processing an image including a character are disclosed. The method may include: searching in a set of characters one or more characters having highest similarities of shape to a character in the set of characters, hereinafter the character being referred to as a first character, the one or more searched characters forming...

In this paper, we present a novel efficient flicker noise reductionmethod for single images scanned by overhead line sensors. The flicker noise here is perceived as horizontal bands which are not necessarily periodic. We view the flicker pattern as the noise of row cumulative histogram along the vertical direction, and propose two novel cumulative...

A method for processing a document image includes: performing horizontal and vertical text line extraction on the document image; providing an overlapping matrix, a value of an element of the overlapping matrix indicating an overlapping relation between horizontal and vertical text lines; merging the overlapping matrix in the vertical and horizonta...

We propose a caption recognition method for multicolor characters on complex background. Caption characters are used for an efficient search on a large amount of recorded TV programs. In the caption character recognition, the caption appearance section and the area is extracted, the character strokes are extracted from the area, and recognized. Thi...

In the real world, there are a huge amount of advertisement (ad) boards to make customers have a visual awareness of the products or services easily. However, information appearing in the ad boards is so limited that customers always want to know more ad details in a convenient way. In this paper, we present an mobile-based prototype system to auto...

Highly accurate optical character recognition (OCR) of handwritten characters is still a challenging task, especially for languages like Chinese and Japanese. To improve the accuracy, we developed four techniques for enhanced recognition: character recognition based on modified linear discriminant analysis (MLDA), subspace-based similar-character d...

This paper proposes a robust single-image super resolution method for enlarging low quality camera captured text image. The contribution of this work is twofold. First, we point out the non-local reconstruction problem in neighbor embedding based super-resolution by statistical analysis on an empirical data set. Second, we introduce a local consist...

Captions in videos are important and accurate clues for video retrieval. In this paper, we propose a fast and robust video caption detection and localization algorithm to handle low quality video images. First, the stroke response maps from complex background are extracted by a stoke filter. Then, two localization algorithms are used to locate thin...

In recent years, various gesture recognition systems have been studied for use in television and video games[1]. In such systems, motion areas ranging from 1 to 3 meters deep have been evaluated[2]. However, with the burgeoning popularity of small mobile displays, gesture recognition systems capable of operating at much shorter ranges have become n...

Comparing with conventional character normalization methods not taking the discriminative information into account, this paper proposes a novel normalization method - Discriminative Normalization. Saliency regions contain most of discriminative information among similar characters. According to different types, they are enlarged in character normal...

Following the recent trend in using low level image features in classifying document images, in this paper we present a novel approach for structured document classification by matching the salient feature points between the query image and the reference images. Our method is robust to diverse training data size, image formats and qualities. Throug...

This paper presents a handwritten digit recognition method based on cascaded heterogeneous convolutional neural networks (CNNs). The reliability and complementation of heterogeneous CNNs are investigated in our method. Each CNN recognizes a proportion of input samples with high-confidence, and feeds the rejected samples into the next CNN. The sampl...