Sequential Word Spotting in Historical Handwritten Documents

Abstract

In this work we present a handwritten word spotting approach that takes advantage of the a priori known order of appearance of the query words. Given an ordered sequence of query word instances, the proposed approach performs a sequence alignment with the words in the target collection. Although the alignment is quite sparse, i.e. the number of words in the database is higher than the query set, the improvement in the overall performance is sensitively higher than isolated word spotting. As application dataset, we use a collection of handwritten marriage licenses taking advantage of the ordered index pages of family names.

A brief review of document image retrieval methods: Recent advances

“…Histogram of oriented gradient (HOG) is a technique which counts occurrences of the gradient orientation in the local part of an image. In [34], an extension of the HOG descriptor for a specific case of handwriting has been described. The combination of gradient features and a flexible plus adaptable grid has been used to extract features.…”

Matching Handwritten Document Images

“…Matching two document images has several applications related to information retrieval like spotting keywords in historical documents [1], accessing personal notes [2], camera based interface for querying [3], retrieving from video databases [4], automatic scoring of answer sheets [5], and mining and recommending in health care documents [6]. Since ocrs do not reliably work for all types of documents, one resorts to image based methods for comparing textual content.…”

“…We validate the effectiveness of our method on an application, named as measure of document similarity (mods). 1 mods compares two handwritten 1 In parallel to measure of software similarity (moss) [7], which has emerged as the de facto standard across the universities to compare two software solutions from students. Figure 1: (a) Given two document images D i and D j , we are interested in computing a similarity score S(D i , D j ) which is invariant to (i) writers, (ii) word flow across lines, (iii) spatial shifts, and (iv) paraphrasing.…”

“…We validate the effectiveness of our method on an application, named as measure of document similarity (mods). 1 mods compares two handwritten Figure 1: (a) Given two document images D i and D j , we are interested in computing a similarity score S(D i , D j ) which is invariant to (i) writers, (ii) word flow across lines, (iii) spatial shifts, and (iv) paraphrasing. In this example, the highlighted lines from D i and D j have almost the same content but they widely differ in terms of spatial arrangement of words.…”

An automatic word-spotting framework for medieval manuscripts

“…RELATED WORK The word-spotting problem has always attracted the interest of the pattern recognition community; the seminal works in the field were in speech recognition [27], while the first application of word-spotting to handwritten text was presented a little later by Manmatha et al [26]. Since then, the importance of indexing and browsing old handwritten books leads to a numerous interesting contributions [11], [37], making it an active area of research. Space does not allow an exhaustive review of the literature.…”