Query Word Image Research Articles

Hough Transform-Based Angular Features for Learning-Free Handwritten Keyword Spotting.

Handwritten keyword spotting (KWS) is of great interest to the document image research community. In this work, we propose a learning-free keyword spotting method following query by example (QBE) setting for handwritten documents. It consists of four key processes: pre-processing, vertical zone division, feature extraction, and feature matching. The pre-processing step deals with the noise found in the word images, and the skewness of the handwritings caused by the varied writing styles of the individuals. Next, the vertical zone division splits the word image into several zones. The number of vertical zones is guided by the number of letters in the query word image. To obtain this information (i.e., number of letters in a query word image) during experimentation, we use the text encoding of the query word image. The user provides the information to the system. The feature extraction process involves the use of the Hough transform. The last step is feature matching, which first compares the features extracted from the word images and then generates a similarity score. The performance of this algorithm has been tested on three publicly available datasets: IAM, QUWI, and ICDAR KWS 2015. It is noticed that the proposed method outperforms state-of-the-art learning-free KWS methods considered here for comparison while evaluated on the present datasets. We also evaluate the performance of the present KWS model using state-of-the-art deep features and it is found that the features used in the present work perform better than the deep features extracted using InceptionV3, VGG19, and DenseNet121 models.

Telugu Word Image Retrieval using Deep learning Convolutional Neural Networks

Telugu word image retrieval (TWIR) is a still challenging task due to the structure complexity of Telugu word image. An efficient TWIR system can be implemented by a holistic representation of word image that comprises of every possible extracted feature. Further, it is also required to retrieve more relevant word images even there is a noisy query word image. Here, it is proposed an efficient TWIR system that utilizes deep learning convolutional neural networks (DL-CNN) to extract the feature map from the query and database word images. In addition, principal component analysis (PCA) is employed to compute the principal features form the feature map and pairwise hamming distance is considered as a similarity metric to retrieve most relevant Telugu word images from the database. Extensive simulation analysis disclosed that proposed TIWR system obtained a superior performance over conventional TIWR systems in terms of mean average precision (mAP) and mean average recall (mAR).

Dependence Models for Searching Text in Document Images.

The main goal of existing word spotting approaches for searching document images has been the identification of visually similar word images in the absence of high quality text recognition output. Searching for a piece of arbitrary text is not possible unless the user identifies a sample word image from the document collection or generates the query word image synthetically. To address this problem, a Markov Random Field (MRF) framework is proposed for searching document images and shown to be effective for searching arbitrary text in real time for books printed in English (Latin script), Telugu and Ottoman scripts. The English experiments demonstrate that the dependencies between the visual terms and letter bigrams can be automatically learned using noisy OCR output. It is also shown that OCR text search accuracy can be significantly improved if it is combined with the proposed approach. No commercial OCR engine is available for Telugu or Ottoman script. In these cases the dependencies are trained using manually annotated document images. It is demonstrated that the trained model can be directly used to resolve arbitrary text queries across books despite font type and size differences. The proposed approach outperforms a state-of-the-art BLSTM baseline in these contexts.

Word spotting in historical documents using primitive codebook and dynamic programming

Word searching and indexing in historical document collections are a challenging problem because text characters are often touching or broken due to degradation or aging effects. In this paper, we present a novel approach towards word spotting using text line decomposition into character primitives and string matching. The text lines are initially separated by a segmentation process. Then each text line is described as sequences of primitive labels which correspond to single characters or parts of characters. These representative primitives are considered from a codebook of shapes generated from training pages taken from the collection. During indexation, the text lines are transcribed into strings of primitives in off-line stage and stored in files. For this purpose, an efficient indexation strategy using multi-label approach is used by a combination of two-level analysis of the primitives: coarse and fine levels. During retrieval, the query word image is encoded into strings of coarse and fine primitives chosen according to the codebook. Finally, a dynamic programming method based on approximate string matching is used to find similar primitive sequences in the text lines from the collection in runtime. We present the experimental evaluation on datasets of real life document images, gathered from historical books of different scripts. Experimental results show that the method is robust in searching text in noisy documents.

Document Image Retrieval Based on Keyword Spotting Using Relevance Feedback

Keyword Spotting is a well-known method in document image retrieval which is based on query word image. In this paper, a document image retrieval system based on keyword spotting and relevance feedback is presented. Relevance feedback as an interactive method is used in this paper to improve the performance of Document Image Retrieval System (DIRS). In the proposed method, we compare several strategies of positive and negative feedbacks which include Only Feedback, Only Negative Feedback and Positive and Negative Feedback. Experiments show that using relevance feedback in DIR outperforms common DIR.

A keyword retrieval system for historical Mongolian document images

In this paper, we propose a keyword retrieval system for locating words in historical Mongolian document images. Based on the word spotting technology, a collection of historical Mongolian document images is converted into a collection of word images by word segmentation, and a number of profile-based features are extracted to represent word images. For each word image, a fixed-length feature vector is formulated by obtaining the appropriate number of the complex coefficients of discrete Fourier transform on each profile feature. The system supports online image-to-image matching by calculating similarities between a query word image and each word image in the collection, and consequently, a ranked result is returned in descending order of the similarities. Therein, the query word image can be generated by synthesizing a sequence of glyphs when being retrieved. By experimental evaluations, the performance of the system is confirmed.

A word spotting method for Farsi machine-printed document images

In this paper, a word spotting approach for Farsi printed document images has been presented. The main idea of the paper is the font recognition of Farsi document images and query word modification according to the document image's font before searching. This operation increases the similarity between the query word image and its instances in the document image; therefore, the performance of the word spotting system increases. In the proposed word spotting approach, after the query word modification, the query word image rectangle is searched in the text lines of the document image using XNOR similarity measurement. In order to increase the recall rate, we considered an almost low value as an acceptance/rejection threshold (\delta) and in order to increase precision rate, we used some other features, e.g., number of holes, ascenders, descenders, and dots. With multilevel matching and considering the mentioned features, the problem of justifying the operation (aligning the text to both the left and right) that occurs during the writing of Farsi documents has been solved. This approach was applied on a computer-made dataset consisting of 440 Farsi printed document images, and a precision rate of 97.5% at a recall rate of 92.1% was obtained. Moreover, when applying this approach on a dataset consisting of 224 Farsi scanned document images, a precision rate of 87.6% at recall rate of 79.3% was obtained.