Method For Writer Identification Research Articles

A new method for writer identification based on historical documents

Abstract Identifying the writer of a handwritten document has remained an interesting pattern classification problem for document examiners, forensic experts, and paleographers. While mature identification systems have been developed for handwriting in contemporary documents, the problem remains challenging from the viewpoint of historical manuscripts. Design and development of expert systems that can identify the writer of a questioned manuscript or retrieve samples belonging to a given writer can greatly help the paleographers in their practices. In this context, the current study exploits the textural information in handwriting to characterize writer from historical documents. More specifically, we employ oBIF(oriented Basic Image Features) and hinge features and introduce a novel moment-based matching method to compare the feature vectors extracted from writing samples. Classification is based on minimization of a similarity criterion using the proposed moment distance. A comprehensive series of experiments using the International Conference on Document Analysis and Recognition 2017 historical writer identification dataset reported promising results and validated the ideas put forward in this study.

An Unsupervised Writer Identification Based on Generating Clusterable燛mbeddings

The writer identification system identifies individuals based on their handwriting is a frequent topic in biometric authentication and verification systems. Due to its importance, numerous studies have been conducted in various languages. Researchers have established several learning methods for writer identification including supervised and unsupervised learning. However, supervised methods require a large amount of annotation data, which is impossible in most scenarios. On the other hand, unsupervised writer identification methods may be limited and dependent on feature extraction that cannot provide the proper objectives to the architecture and be misinterpreted. This paper introduces an unsupervised writer identification system that analyzes the data and recognizes the writer based on the inter-feature relations of the data to resolve the uncertainty of the features. A pairwise architecture-based Autoembedder was applied to generate clusterable embeddings for handwritten text images. Furthermore, the trained baseline architecture generates the embedding of the data image, and the K-means algorithm is used to distinguish the embedding of individual writers. The proposed model utilized the IAM dataset for the experiment as it is inconsistent with contributions from the authors but is easily accessible for writer identification tasks. In addition, traditional evaluation metrics are used in the proposed model. Finally, the proposed model is compared with a few unsupervised models, and it outperformed the state-of-the-art deep convolutional architectures in recognizing writers based on unlabeled data.

Funnelling Ensemble Method for Writer Identification (Fem-Wi)

Funnelling Ensemble Method for Writer Identification (Fem-Wi)

A Deep Learning Method for Chinese writer Identification with Feature Fusion

Writer identification is one of the research hotspots of computer vision and pattern recognition, and it is of great significance in the fields of judicial authentication, file security protection, historical document analysis, and so on. However, many problems are still challenging due to the different writing sources, the common features of learning local features, and the implicit features of handwriting style. This paper uses the fusion of depth features and manual features to obtain handwriting style features from handwriting pictures. Firstly, the handwriting picture is pre-treatment and divided into small pixel blocks, and then the depth feature and LPQ feature information are extracted from each pixel block, and the depth features are reduced by PCA, and then the local features are encoded into global features through the Vlad algorithm. So far, the depth global feature and the LPQ global feature of a page of handwriting pictures are obtained, and the two features are combined as the global feature vector of the page of handwriting. And our method has achieved good results on the CASIA-HWDB data set.

Deep adaptive learning for writer identification based on single handwritten word images

There are two types of information in each handwritten word image: explicit information which can be easily read or derived directly, such as lexical content or word length, and implicit attributes such as the author’s identity. Whether features learned by a neural network for one task can be used for another task remains an open question. In this paper, we present a deep adaptive learning method for writer identification based on single-word images using multi-task learning. An auxiliary task is added to the training process to enforce the emergence of reusable features. Our proposed method transfers the benefits of the learned features of a convolutional neural network from an auxiliary task such as explicit content recognition to the main task of writer identification in a single procedure. Specifically, we propose a new adaptive convolutional layer to exploit the learned deep features. A multi-task neural network with one or several adaptive convolutional layers is trained end-to-end, to exploit robust generic features for a specific main task, i.e., writer identification. Three auxiliary tasks, corresponding to three explicit attributes of handwritten word images (lexical content, word length and character attributes), are evaluated. Experimental results on two benchmark datasets show that the proposed deep adaptive learning method can improve the performance of writer identification based on single-word images, compared to non-adaptive and simple linear-adaptive approaches.

Writer identification using machine learning approaches: a comprehensive review

Handwriting is one of the most common types of questioned writing encountered and frequently attracts the attention in litigation. Contrary to the physiological characteristics, handwriting is a behavioral characteristic thus no two individuals with mature handwriting are exactly alike or an individual cannot produce the others writing exactly. Writing behavior and individualities are examined for similarities for both specimen and questioned document, thus, it is very efficient and effective strategy for biometrics. In this paper, we present a comprehensive review of writer identification methods and intend to provide taxonomy of dataset, feature extraction methods, as well as classification (conventional and deep learning based) for writer identification. For ease of reader, we grouped the discussion into English, Arabic, Western and Other languages from script prospective, whereas, from algorithm and methods perspective, we grouped the discussion with respect to implementation steps sequence. In the end, we highlighted the challenges and open research issues in the field of writer identification. Finally, we also suggest future direction.

Text-independent writer identification using convolutional neural network

The text-independent approach to writer identification does not require the writer to write some predetermined text. Previous research on text-independent writer identification has been based on identifying writer-specific features designed by experts. However, in the last decade, deep learning methods have been successfully applied to learn features from data automatically. We propose here an end-to-end deep-learning method for text-independent writer identification that does not require prior identification of features. A Convolutional Neural Network (CNN) is trained initially to extract local features, which represent characteristics of individual handwriting in the whole character images and their sub-regions. Randomly sampled tuples of images from the training set are used to train the CNN and aggregate the extracted local features of images from the tuples to form global features. For every training epoch, the process of randomly sampling tuples is repeated, which is equivalent to a large number of training patterns being prepared for training the CNN for text-independent writer identification. We conducted experiments on the JEITA-HP database of offline handwritten Japanese character patterns. With 200 characters, our method achieved an accuracy of 99.97% to classify 100 writers. Even when using 50 characters for 100 writers or 100 characters for 400 writers, our method achieved accuracy levels of 92.80% or 93.82%, respectively. We conducted further experiments on the Firemaker and IAM databases of offline handwritten English text. Using only one page per writer to train, our method achieved over 91.81% accuracy to classify 900 writers. Overall, we achieved a better performance than the previously published best result based on handcrafted features and clustering algorithms, which demonstrates the effectiveness of our method for handwritten English text also.

A multi-language writer identification method based on image mining and genetic algorithm techniques

Writing identification based on handwriting has many applications in the real world. Due to various forms of written letters in different languages, one of the major challenges in this context is offering an efficient method not being dependent on any specific language. In this paper, we have proposed a new approach based on image mining techniques for offline and text independent writer identification. In this method, each writers’ prominent features are found from training samples, and then identification is done according to them. In the image mining part of the proposed approach, certain techniques including SVM classifier and genetic algorithms are employed. To evaluate this method and show its performance in different languages, CASIA for Chinese, IAM dataset for English and two datasets for Kannada and Persian language handwriting were examined. The experiment results demonstrate that the presented method has over 99% accuracy for these languages. Regarding the results in tested languages and the method details, it is highly likely that our method would have good results in other languages also.

Writer identification using intra-stroke and inter-stroke information for security enhancements in P2P systems

Chinese language has enormous number of characters and complicated stroke structures. So it is very difficult to efficiently and accurately identify a Chinese writer from his/her handwritings. This paper proposes a novel writer identification method for Chinese characters commonly used in Japan which can be used in peer-to-peer (P2P) systems. As a preliminary task, we have analyzed the shapes of strokes and the types of block division structures in Chinese characters and selected some characters for writer identification. The method consists of two efficient algorithms, i.e. the Hidden-feature analysis and the Block-type model, which respectively utilize intra-stroke and inter-stroke features of handwritings to enhance the writer identification accuracy. The Hidden-feature analysis makes template classes of reference characters with online features of training samples such as pen-pressure, pen-speed, pen-altitude, and pen-azimuth of each stroke. The Block-type model also creates such classes for writer identification based on offline features, i.e. the positional information about blocks of sample characters. The experimental results show that the Hidden-feature analysis requires eight Chinese characters while the Block-type model requires only four characters and four ones to achieve writer identification accuracy over 98%. Additionally, the results also demonstrate that any eight Chinese characters are enough to achieve an identification accuracy over 99.9% when the combination of the two algorithms is applied.

Signature Verification for Banking Sector Based on SIFT

A person can be uniquely identified by evaluating one or more distinguishing biological traits. One such trait is scanned images of hand writing. It is useful biometric modality with application in forensic analysis, banking sector, document verification etc. This study focus on text-independent writer identification method based on scale invariant feature transform (SIFT). It consists of three phase: training, enrolment and identification phase. In all of these three phases an isotropic LoG filter is applied first. It is used to segment the handwriting image into word regions (WRs) and then, the SIFT descriptors (SDs) of word region and the corresponding scales and orientations (SOs) are extracted. In the first phase, an SD codebook is constructed by gathering the SDs training samples. In the second phase, SD signatures (SDS) are generated by looking up the SD codebook and SOs is used to generate a scale and orientation histogram (SOH). And finally in last phase, the SDS and SOH of the input handwriting are extracted and matched with the enrolled ones for identification. Here instead of analysing handwritten document we experiment on signature of an individual. Using SIFT algorithm we demonstrate that he/she is genuine or not.

A Novel Approach to Text Dependent Writer Identification of Kannada Handwriting

In this paper a text dependent writer identification method based on Kannada handwriting is proposed. The database of Kannada handwritten words collected from 25 writers is intended to provide training and testing sets for Kannada writer identification research, because there is no well-known database containing Kannada handwritten words available for writer identification problem. A feature vector consisting of directional multi-resolution spatial features based on Radon Transform and Discrete Cosine Transform and structural features such as aspect ratio and on-pixel ratio are extracted from word images. The novel approach of combining the features of two or more words is proposed and performance analysis is done through nearest neighbor classifier with modified 5-fold cross-validation. An average identification accuracy of 93.2582% is achieved by using only the single words, and accuracy of 100% is achieved using the combination of features of three or more words. The writer identification results show that feature vectors extracted from longer words, words having more structural variation and combination of features of two or more words have higher impact on writer identification.

A Method for Calligraphy Writer Identification by Integrating Gabor Filter and Gaussian Markov Random Field

Extracting efiective features to describe texture is always a key problem in writer identiflcation. This paper proposes a novel method for texture feature extraction by integrating Gabor fllter and Gauss Markov Random Field (GMRF). That is to say, the handwriting images are flrstly flltered by a bank of Gabor fllters in which global features such as directional information can be detected; then GMRF models are developed for every flltered image that is ∞exible enough to capture the local spatial structure and the model parameters of all GMRFs are concatenated as texture features for writer identiflcation, and flnally, Support Vector Machine (SVM) is applied to evaluate the performance of calligraphic artist identiflcation. The experimental results show that this method can achieve a classiflcation rate of 98%, which has outperformed the traditional Gabor fllter method.

Offline Text-Independent Writer Identification Based on Scale Invariant Feature Transform

This paper proposes a novel offline text-independent writer identification method based on scale invariant feature transform (SIFT), composed of training, enrollment, and identification stages. In all stages, an isotropic LoG filter is first used to segment the handwriting image into word regions (WRs). Then, the SIFT descriptors (SDs) of WRs and the corresponding scales and orientations (SOs) are extracted. In the training stage, an SD codebook is constructed by clustering the SDs of training samples. In the enrollment stage, the SDs of the input handwriting are adopted to form an SD signature (SDS) by looking up the SD codebook and the SOs are utilized to generate a scale and orientation histogram (SOH). In the identification stage, the SDS and SOH of the input handwriting are extracted and matched with the enrolled ones for identification. Experimental results on six public data sets (including three English data sets, one Chinese data set, and two hybrid-language data sets) demonstrate that the proposed method outperforms the state-of-the-art algorithms.

A Persian writer identification method using swarm-based feature selection approach

Handwriting is one of the most famous biometrics which is processed based on image processing and pattern recognition techniques. However, there are a lot of reports that have already been published on handwritten text identification methods and researchers try to improve the accuracy and speed of such methods. This paper presents an offline Persian handwriting identification method in which some new text features are extracted and best ones are selected using a swarm-based approach. The essence of this feature selection method is bees algorithm, which is a modern swarm-based meta-heuristic approach. In the proposed technique, the adaptive neuro-fuzzy inference system ANFIS is employed as classifier and trained by the input feature vectors. It is also compared with a multi-layer perceptron MLP and fuzzy K-nearest neighbour classifiers. To test the proposed method, we have collected a handwritten Persian text dataset from 125 people who have written six sheets with five lines in each of optional Persian texts. Experimental results showed that the prediction accuracy was about 98% in average while the method training time is less than most related works. It seems this method can be extended for other languages by adjusting its parameters.

A novel sparse model based forensic writer identification

The paper presents a novel method for writer identification based on sparse representation of handwritten structural primitives, called graphemes or fraglets. The proposed method is different from the existing grapheme based methods as the earlier methods use vector quantization based coding (clustering method) to get a document descriptor, while the proposed method uses sparse coding for the same. Literature shows that the sparse coding outperforms vector quantization in many real life applications including face recognition. Sparse coding can achieve comparatively much lower reconstruction error. Secondly, the sparsity allows representation to be specialized and can capture a writer specific features more accurately. Graphemes (fraglets) extracted from a document are represented in terms of Fourier and wavelet descriptors because the fraglet contour may be well described by its global as well as local characteristics. Wavelet descriptors also give a multi-resolution representation of the shape. Results have shown that even with a smaller codebook (than the earlier reported systems), the proposed method achieves better performance.

Offline text-independent writer identification using codebook and efficient code extraction methods

In this paper, an efficient method for text-independent writer identification using a codebook method is proposed. The method uses the occurrence histogram of the shapes in a codebook to create a feature vector for each specific manuscript. For cursive handwritings, a wide variety of different shapes exist in the connected components obtained from the handwriting. Small fragments of connected components are used to avoid complex patterns. Two efficient methods for extracting codes from contours are introduced. One method uses the actual pixel coordinates of contour fragments while the other one uses a linear piece-wise approximation using segment angles and lengths. To evaluate the methods, writer identification is conducted on two English and three Farsi handwriting databases. Both methods show promising performances with the performance of second method being better than the first one.

WRITER IDENTIFICATION BY TEXTURE ANALYSIS BASED ON KANNADA HANDWRITING

Writer identification problem is one of the important area of research due to its various applications and is a challenging task. The major research on writer identification is based on handwritten English documents with text independent and dependent. However, there is no significant work on identification of writers based on Kannada document. Hence, in this paper, we propose a text-independent method for off-line writer identification based on Kannada handwritten scripts. By observing each individual’s handwriting as a different texture image, a set of features based on Discrete Cosine Transform, Gabor filtering and gray level co-occurrence matrix, are extracted from preprocessed document image blocks. Experimental results demonstrate that the Gabor energy features are more potential than the DCTs and GLCMs based features for writer identification from 20 people.

A Survey on Writer Identification Schemes

paper presents a survey of the literature on writer identification schemes and techniques up till date. The paper outlines an overview of the writer identification schemes mainly in Chinese, English, Arabic and Persian languages. Taxonomy of different features adopted for online and offline writer identification schemes is also drawn at. The feature extraction methods adopted for the schemes are discussed in length outlining the merits and demerits of the same. In automated writer identification, text independent and text dependent methods are available which is also discussed in this paper. An evaluation of writer identification schemes under multiple languages is also analyzed by comparing the recognition rate.

AN OFF-LINE TEXT-INDEPENDENT PERSIAN WRITER IDENTIFICATION METHOD

The behavioral-biometrics methods of writer identification and verification have been considered as a research topic for many years. However, many writer identification and verification methods have been designed based on English handwriting properties, but because of many differences between English and Persian handwriting and the challenges facing Persian handwriting analysis, designing such methods has many interests in Persian yet. In this paper, we have presented a fully text-independent and texture based method for identifying writers of Persian handwritten documents. As a result of special properties of Persian handwriting, a modified version of Gabor filter that is called Extended Gabor (XGabor) filter has been used to extract the features. An MLP (Multi Layer Perceptron (Node)) neural network and a K-NN classifier have been employed to classify the extracted features. In the evaluation phase, an exhaustive database of Persian handwritten documents was prepared and the method applied on. The experimental results showed that the accuracy of proposed method is about 97% and it is competitive with others. We believe that the proposed method may be extended to identify writers in other languages by adjusting some parameters.

A text-independent Persian writer identification based on feature relation graph (FRG)

The style of people's handwriting is a biometric feature that is used in person authentication. In this paper, we have proposed a text independent method for Persian writer identification. In the proposed method, pattern based features are extracted from data using Gabor and XGabor filter. The extracted features are represented for each person by using a graph that is called FRG (feature relation graph). This graph is constructed using relations between extracted features by employing a fuzzy method. The fuzzy method determines the similarity between features extracted from different handwritten instances of each person. In the identification phase, a graph similarity approach is employed to determine the similarity of the FRG generated from the test data and the FRGs generated by training data. The experimental results were satisfactory and the proposed method got about 100% accuracy on a dataset with 100 writers when enough training data was used. However, this method has been applied on Persian handwritings but we believe it can be extended on other languages especially in data representation and classification parts.