Anguelos Nicolaou

ICDAR 2015 competition on Robust Reading

Results of the ICDAR 2015 Robust Reading Competition are presented. A new Challenge 4 on Incidental Scene Text has been added to the Challenges on Born-Digital Images, Focused Scene Images and Video Text. Challenge 4 is run on a newly acquired dataset of 1,670 images evaluating Text Localisation, Word Recognition and End-to-End pipelines. In addition, the dataset for Challenge 3 on Video Text has been substantially updated with more video sequences and more accurate ground truth data. Finally, tasks assessing End-to-End system performance have been introduced to all Challenges. The competition took place in the first quarter of 2015, and received a total of 44 submissions. Only the tasks newly introduced in 2015 are reported on. The datasets, the ground truth specification and the evaluation protocols are presented together with the results and a brief summary of the participating methods.

Handwritten Text Line Segmentation by Shredding Text into its Lines

In this paper, we propose a novel technique to segment handwritten document images into text lines by shredding their surface with local minima tracers. Our approach is based on the topological assumption that for each text line, there exists a path from one side of the image to the other that traverses only one text line. We first blur the image and then use tracers to follow the white-most and black-most paths from left to right as well as from right to left in order to shred the image into text line areas. We experimentally tested the proposed methodology and got promising results comparable to state of the art text line segmentation techniques.

Sparse radial sampling LBP for writer identification

Sampling Local Binary Patterns, a variant of Local Binary Patterns (LBP) for text-as-texture classification. By adapting and extending the standard LBP operator to the particularities of text we get a generic text-as-texture classification scheme and apply it to writer identification. In experiments on CVL and ICDAR 2013 datasets, the proposed feature-set and a simple end-to-end pipeline demonstrate State-Of-the-Art (SOA) performance. Among the SOA, the proposed method is the only one that is based on dense extraction of a single local feature descriptor. This makes it fast and applicable at the earliest stages in a DIA pipeline without the need for segmentation, binarization, or extraction of multiple features.

Improving patch-based scene text script identification with ensembles of conjoined networks

We present a patch-based classification method for script identificattion in the wild.We describe a novel method based on the use of ensembles of conjoined networks (ECN).The ECN learns discriminative local features and their relative importance in a global classification rule.Our experiments demonstrate state-of-the-art results in three script identification datasets. This paper focuses on the problem of script identification in scene text images. Facing this problem with state of the art CNN classifiers is not straightforward, as they fail to address a key characteristic of scene text instances: their extremely variable aspect ratio. Instead of resizing input images to a fixed aspect ratio as in the typical use of holistic CNN classifiers, we propose here a patch-based classification framework in order to preserve discriminative parts of the image that are characteristic of its class.We describe a novel method based on the use of ensembles of conjoined networks to jointly learn discriminative stroke-parts representations and their relative importance in a patch-based classification scheme. Our experiments with this learning procedure demonstrate state-of-the-art results in two public script identification datasets.In addition, we propose a new public benchmark dataset for the evaluation of multi-lingual scene text end-to-end reading systems. Experiments done in this dataset demonstrate the key role of script identification in a complete end-to-end system that combines our script identification method with a previously published text detector and an off-the-shelf OCR engine.

Visual Script and Language Identification

In this paper we introduce a script identification method based on hand-crafted texture features and an artificial neural network. The proposed pipeline achieves near state-of-the-art performance for script identification of video-text and state-of-the-art performance on visual language identification of handwritten text. More than using the deep network as a classifier, the use of its intermediary activations as a learned metric demonstrates remarkable results and allows the use of discriminative models on unknown classes. Comparative experiments in video-text and text in the wild datasets provide insights on the internals of the proposed deep network.

Local Binary Patterns for Arabic Optical Font Recognition

Optical Font Recognition (OFR) has been proven to increase Optical Character Recognition (OCR) accuracy, but it can also help in harvesting semantic information from documents. It therefore becomes a part of many Document Image Analysis (DIA) pipelines. Our work is based on the hypothesis that Local Binary Patterns (LBP), as a generic texture classification method, can address several distinct DIA problems at the same time such as OFR, script detection, writer identification, etc. In this paper we strip down the Redundant Oriented LBP (RO-LBP) method, previously used in writer identification, and apply it for OFR with the goal of introducing a generic method that classifies text as oriented texture. We focus on Arabic OFR and try to perform a thorough comparison of our method and the leading Gaussian Mixture Model method that is developed specifically for the task. Depending on the nature of proposed OFR method, each methods performance is usually evaluated on different data and with different evaluation protocols. The proposed experimental procedure addresses this problem and allows us to compare OFR methods that are fundamentally different by adapting them to a common measurement protocol. In performed experiments LBP method achieves perfect results on large text blocks generated from the APTI database, while preserving its very broad generic attributes as proven by secondary experiments.

Investigation of feature selection for historical document layout analysis

In this paper we investigate the importance of individual features for the task of document layout analysis, in particular for the classification of the document pixels. The feature set consists of numerous state-of-the-art features, including color, gradient, and local binary patterns (LBP). To deal with the high dimensionality of the feature set, we propose a cascade of an adapted forward selection and a genetic selection. We have evaluated our feature selection method on three historical document datasets. For the classification we used machine learning methods which classify each pixel into either periphery, background, text block, or decoration. The proposed cascading feature selection method reduced the number of features significantly while preserving the cross-validation performance. Furthermore, it selected less features with comparable performance, compared with the conventional feature selection methods. In our analysis we found that LBP features are consistently selected by all feature selection methods on all three datasets. This indicates that LBP correlate highly with the pixel classes much more than any other type of features does. These findings suggest a clue in paradigm for document layout analysis in general.

Binarization with the Local Otsu Filter

In this paper we introduce the use of integral histograms (IH) for document analysis. IH take advantage of the great increase of the memory size available on computers over time. By storing selected histogram features into each pixel position, several image filters can be calculated within constant complexity. In other words, time complexity is remarkably reduced by using more memory. While IH received much attention in the computer vision field, they have not been intensively investigated for document analysis so far. As a first step into this direction, we analyze IH for the toy problem of image binarization which is a prerequisite for many graphics and text recognition systems. The results of our participation in the HDIBCO2010 competition as well as our experiments with all DIBCO datasets show the capabilities of this novel method for Document Image analysis.