Manuel J. Fonseca

Using fuzzy logic to recognize geometric shapes interactively

Presents a simple method, based on fuzzy logic, to recognize multi-stroke sketches of geometric shapes and uni-stroke gestural commands. It uses temporal adjacency and global geometric properties of figures to recognize a simple vocabulary of geometric shapes drawn in different line styles. The geometric features used (convex hull, largest-area inscribed and smallest-area enclosing polygons, perimeter and area ratios) are invariant with rotation and scale of figures. Through experimental evaluation we have found the method very usable with acceptable recognition rates although the multi-stroke approach poses problems in choosing appropriate values for timeouts. Although we have privileged simplicity over robustness, the method has proved suitable for interactive applications.

Experimental evaluation of an on-line scribble recognizer

Abstract We present a fast, simple and compact approach to recognize scribbles (multi-stroke geometric shapes) drawn with a stylus on a digitizing tablet. Regardless of size, rotation and number of strokes, our method identifies the most common shapes used in drawings, allowing for dashed, continuous or overlapping strokes. Our method combines temporal adjacency, fuzzy logic and geometric features to classify scribbles with measured recognition rates over 97%.

Indexing high-dimensional data for content-based retrieval in large databases

Many indexing approaches for high-dimensional data points have evolved into very complex and hard to code algorithms. Sometimes this complexity is not matched by increase in performance. Motivated by these ideas, we take a step back and look at simpler approaches to indexing multimedia data. In this paper we propose a simple, (not simplistic) yet efficient indexing structure for high-dimensional data Points of variable dimension, using dimension reduction. Our approach maps multidimensional points to a 1D line by computing their Euclidean Norm and use a B/sup +/-Tree to store data points. We exploit B/sup +/-Tree efficient sequential search to develop simple, yet performant methods to implement point, range and nearest-neighbor queries. To evaluate our technique we conducted a set of experiments, using both synthetic and real data. We analyze creation, insertion and query times as a function of data set size and dimension. Results so far show that our simple scheme outperforms current approaches, such as the Pyramid Technique, the A-Tree and the SR-Tree, for many data distributions. Moreover, our approach seems to scale better both with growing dimensionality and data set size, while exhibiting low insertion and search times.

A comparison of methods for sketch-based 3D shape retrieval

Sketch-based 3D shape retrieval has become an important research topic in content-based 3D object retrieval. To foster this research area, two Shape Retrieval Contest (SHREC) tracks on this topic have been organized by us in 2012 and 2013 based on a small-scale and large-scale benchmarks, respectively. Six and five (nine in total) distinct sketch-based 3D shape retrieval methods have competed each other in these two contests, respectively. To measure and compare the performance of the top participating and other existing promising sketch-based 3D shape retrieval methods and solicit the state-of-the-art approaches, we perform a more comprehensive comparison of fifteen best (four top participating algorithms and eleven additional state-of-the-art methods) retrieval methods by completing the evaluation of each method on both benchmarks. The benchmarks, results, and evaluation tools for the two tracks are publicly available on our websites [1,2].

Towards content-based retrieval of technical drawings through high-dimensional indexing

Abstract This paper presents a new approach to classify, index and retrieve technical drawings by content. Our work uses spatial relationships, visual elements and high-dimensional indexing mechanisms to retrieve complex drawings from CAD databases. This contrasts with conventional approaches which use mostly textual metadata for the same purpose. Creative designers and draftspeople often reuse data from previous projects, publications and libraries of ready to use components. Usually, retrieving these drawings is a slow, complex and error-prone endeavor, requiring either exhaustive visual examination, a solid memory, or both. Unfortunately, the widespread use of CAD systems, while making it easier to create and edit drawings, exacerbates this problem, insofar as the number of projects and drawings grows enormously, without providing adequate retrieval mechanisms to support retrieving these documents. In this paper, we describe an approach that supports automatic indexing of technical drawing databases through drawing simplification techniques based on geometric features and efficient algorithms to index large amounts of data. We describe in detail the indexing structure (NB-Tree) we have developed within the context of a more general approach. Experimental evaluation reveals that our approach outperforms some of the best indexing structures published, enabling us to search very large drawing databases.

Thesaurus-based 3D Object Retrieval with Part-in-Whole Matching

Research in content-based 3D retrieval has already started, and several approaches have been proposed which use in different manner a similarity assessment to match the shape of the query against the shape of the objects in the database. However, the success of these solutions are far from the success obtained by their textual counterparts.A major drawback of most existing 3D retrieval solutions is their inability to support partial queries, that is, a query which does not need to be formulated by specifying a whole query shape, but just a part of it, for example a detail of its overall shape, just like documents are retrieved by specifying words and not whole texts. Recently, researchers have focused their investigation on 3D retrieval which is solved by partial shape matching. However, at the extent of our knowledge, there is still no 3D search engine that provides an indexing of the 3D models based on all the interesting subparts of the models.In this paper we present a novel approach to 3D shape retrieval that uses a collection-aware shape decomposition combined with a shape thesaurus and inverted indexes to describe and retrieve 3D models using part-in-whole matching. The proposed method clusters similar segments obtained trough a multilevel decomposition of models, constructing from such partition the shape thesaurus. Then, to retrieve a model containing a sub-part similar to a given query, instead of looking on a large set of subparts or executing partial matching between the query and all models in the collection, we just perform a fast global matching between the query and the few entries in the thesaurus. With this technique we overcame the time complexity problems associated with partial queries in large collections.

Sketch-based retrieval of drawings using spatial proximity

Currently, there are large collections of drawings from which users can select the desired ones to insert in their documents. However, to locate a particular drawing among thousands is not easy. In our prior work we proposed an approach to index and retrieve vector drawings by content, using topological and geometric information automatically extracted from figures. In this paper, we present a new approach to enrich the topological information by integrating spatial proximity in the topology graph, through the use of weights in adjacency links. Additionally, we developed a web search engine for clip art drawings, where we included the new technique. Experimental evaluation reveals that the use of topological proximity results in better retrieval results than topology alone. However, the increase in precision was not as high as we expected. To understand why, we analyzed sketched queries performed by users in previous experimental sessions and we present here the achieved conclusions.

Content-based retrieval of technical drawings

This paper presents a new approach to classify, index and retrieve technical drawings by content. Our work uses spatial relationships, shape geometry and high-dimensional indexing mechanisms to retrieve complex drawings from CAD databases. This contrasts with conventional approaches which use mostly textual metadata. Creative designers and draftspeople often re-use data from previous projects, publications and libraries of ready-to-use components. Usually, retrieving these drawings is a slow, complex and error-prone endeavour. Unfortunately, the widespread use of CAD systems, while making it easier to create drawings, exacerbates this problem, insofar as the number of projects grows enormously, without providing adequate searching mechanisms to support retrieving these documents. We describe an approach that supports automatic indexation of technical drawing databases through drawing simplification, feature extraction and efficient algorithms to index large amounts of data. We describe in detail our classification process and present results from usability tests on our prototype.

Sketch-based retrieval of complex drawings using hierarchical topology and geometry

Due to the proliferation of drafting packages, there are a lot of vector drawings available for people to integrate into documents. Moreover, creative designers and drafts-people often reuse data from previous projects and libraries of ready-to-use components. Usually, retrieving these drawings is a slow, complex and error-prone endeavor. While text-driven attempts at classifying image data have been recently supplemented with query-by-image content, these work only for bitmap-type data and cannot handle vectorial information. Furthermore, they are not very good at retrieving partial information, which makes it difficult to search complex documents. Here we present a new approach to allow efficient retrieval of complex vector drawings by content using simple sketches as queries. Our work uses a new hierarchical topology description mechanism, which allows comparing complex drawings to simple queries. In particular, we produce a multilevel description of drawings, from the topology graph, using different levels of detail and subparts of the information. We also show that the graph spectrum is stable and reliable for use as a topology description. Finally, experimental tests show that our approach is good at retrieving complex drawings from two different application domains, CAD and clip-art figures.

A Simple Approach to Recognise Geometric Shapes Interactively

This paper presents a simple method to recognise multistroke sketches of geometric shapes. It uses temporal adjacency and global geometric properties of figures to recognise a simple vocabulary of geometric shapes including solid and dashed line styles, selection and delete gestures. The geometric features used (convex hull, smallest-area regular polygons, perimeter and area scalar ratios) are invariant with rotation and scale of figures. We have found the method very usable with acceptable recognition rates although the multi-stroke approach poses problems in choosing appropriate values for time-outs. Although we have privileged simplicity over robustness, the method has proved suitable for interactive applications.