A Platform to Extract Knowledge from Graphic Documents. Application to an Architectural Sketch Understanding Scenario

Abstract

This paper proposes a general architecture to extract knowledge from graphic documents. The architecture consists of three major components. First, a set of modules able to extract descriptors that, combined with domain-dependent knowledge and recognition strategies, allow to interpret a given graphical document. Second, a representation model based on a graph structure that allows to hierarchically represent the information of the document at different abstraction levels. Finally, the third component implements a calligraphic interface that allows the feedback between the user and the system. The second part of the paper describes an application scenario of the above platform. The scenario is a system for the interpretation of sketches of architectural plans. This is a tool to convert sketches to a CAD representation or to edit a given plan by a sketchy interface. The application scenario combines different symbol recognition algorithms stated in terms of document descriptors to extract building elements such as doors, windows, walls and furniture. This work has been partially supported by the spanish project CICYT TIC2003-09291 and catalan project CeRTAP PVPC.

Mining Historical Documents for Near-Duplicate Figures

“…Likewise, we have exploited both the classic ideas of the GHT [2], and the recent extensions by [28]. There is an active community working on computerized historical document analyses [10][12] [18]: however, while great many papers address query-by-content, including [9][15] [20] the task of motif discovery in this domain has not been addressed thus far.…”

“…While others have worked on these datasets, we did not directly compare our results to theirs. The published approaches on these datasets are so slow (an O(n 3 ) warping method for the music symbols [9] [20], and an O(n 3 ) adjacency grammar method for the architectural symbols [15]), that in both cases the authors abandoned any attempt at a full leaving-one-out on the entire dataset, and instead created various smaller subsets (hand crafted and thus difficult to meaningfully compare to). However, our accuracies are so close to perfect in every case that our claim is clearly demonstrated: the GHT on downsampled images is an effective distance measure for these kinds of images.…”

“…We achieve these aims by testing on datasets of hand-drawn figures. Two of the datasets are from a collection of old music scores (17th-19th centuries) [9] [20], and thus are very representative of our domain of interest, and the third one is a modern architectural symbol dataset [15], in which various users hand copied symbols, and is thus also very similar to the task at hand. As Figure 10 shows, these are non-trivial problems.…”

Efficiently Finding Near Duplicate Figures in Archives of Historical Documents

“…We achieve these aims by testing on datasets of handdrawn figures. Two of the datasets are from a collection of old music scores (17 th -19 th centuries) [13] [26], and thus are very representative of our domain of interest, and the third one is a modern architectural symbol dataset [21], in which various users hand copied symbols, and is thus also very similar to the task at hand. As Figure 11 shows, these are non-trivial problems.…”

An efficient and effective similarity measure to enable data mining of petroglyphs

“…The final dataset we tested is an architectural symbol dataset (Sanchez et al 2004) consisting of 7414 samples in 50 classes (the printed version is shown in Fig. 23) by 21 users.…”