Catalin I. Tomai

Transcript mapping for historic handwritten document images

There is a large number of scanned historical documents that need to be indexed for archival and retrieval purposes. A visual word spotting scheme that would serve these purposes is a challenging task even when the transcription of the document image is available. We propose a framework for mapping each word in the transcript to the associated word image in the document. Coarse word mapping based on document constraints is used for lexicon reduction. Then, word mappings are refined using word recognition results by a dynamic programming algorithm that finds the best match while satisfying the constraints.

Discriminatory power of handwritten words for writer recognition

Analysis of allographs (characters) and allograph combinations (words) is the key for the identification/verification of a writers handwriting. While allographs are usually part of words and the segmentation of a word into allographs is a subjective process, analysis of handwritten words is a natural option, complementary to allograph and document-level analysis. We consider four different types of features obtained using both segmentation-based and segmentation-free approaches: (i) GSC (gradient, structural and concavity) features that are extracted from the cells of a grid superimposed on the word image (ii) WMR (word model recognizer) features, extracted from the cells of superimposed grids on the segmented characters (iii) SC (shape curvature) features that describe characters by the distribution of curvature values on their contours and (iv) SCON (shape context) features that measure the similarity between character contour shapes. Their individual and accumulated performance is evaluated for the writer identification and verification tasks on over 75000 words images, written by more than 1000 writers. Experimental results show that handwritten words are very effective in discriminating handwriting and that both segmentation-free and segmentation-based approaches are valid.

Individuality of numerals

The analysis of handwritten documents from the view-pointof determining their writership has great bearing onthe criminal justice system. In many cases, only a limitedamount of handwriting is available and sometimes it consistsof only numerals. Using a large number of handwrittennumeral images extracted from about 3000 samples writtenby 1000 writers, a study of the individuality of numerals foridentification/verification purposes was conducted. The individualityof numerals was studied using cluster analysis.Numerals discriminability was measured for writer verification.The study shows that some numerals present a higherdiscriminatory power and that their performances for theverification/identification tasks are very different.

Information retrieval system for handwritten documents

The design and performance of a content-based information retrieval system for handwritten documents is described. System indexing and retrieval is based on writer characteristics, textual content as well as document meta data such as writer profile. Documents are indexed using global image features, e.g., stroke width, slant, word gaps, as well local features that describe shapes of characters and words. Image indexing is done automatically using page analysis, page segmentation, line separation, word segmentation and recognition of characters and words. Several types of queries are permitted: (i) entire document image; (ii) a region of interest (ROI) of a document; (iii) a word image; and (iv) textual. Retrieval is based on a probabilistic model of information retrieval. The system has been implemented using Microsoft Visual C++ and a relational database system. This paper reports on the performance of the system for retrieving documents based on same and different content.

Group discriminatory power of handwritten characters

Using handwritten characters we address two questions (i) what is the group identification performance of different alphabets (upper and lower case) and (ii) what are the best characters for the verification task (same writer/different writer discrimination) knowing demographic information about the writer such as ethnicity, age or sex. The Bhattacharya distance is used to rank different characters by their group discriminatory power and the k-nn classifier to measure the individual performance of characters for group identification. Given the tasks of identifying the correct gender/age/ethnicity or handedness, the accumulated performance of characters varies between 65% and 85%.

Adaptive texture image retrieval in transform domain

A large number of algorithms have been proposed to retrieve and analyze texture images. While much effort has been made to find algorithms applicable to all textures for superior retrieval performance, less work has been done to adaptively integrate various texture retrieval and analysis algorithms. As no individual texture retrieval algorithm is suited for every texture category, a hybrid scheme would outperform any individual method. In this paper, an adaptive retrieval scheme (ARS) for texture image indexing is proposed to dynamically adapt different transforms to different texture patterns for better retrieval performance. The experiments on the Brodatz texture database show that ARS significantly outperforms any individual transform.

Recognition of handwritten foreign mail

Foreign mail recognition (FMR) is part of the more general problem of recognizing destination addresses in a mail stream. It is defined as the problem of finding the country of destination of a mail piece sent to a foreign address. We discuss some of the differences between FMR and domestic mail recognition (DMR) and present its specific challenges. Two complementary baseline algorithms that use heuristics in combining word, character and digit recognizers are presented. Their performance is improved by reducing the search space for the address elements using address configuration distributions. Preliminary results are presented.