Jairo Rocha

A shape analysis model with applications to a character recognition system

A method for the recognition of multifont printed characters is proposed, giving emphasis to the identification of structural descriptions of character shapes using prototypes. Noise and shape variations are modeled as series of transformations from groups of features in the data to features in each prototype. Thus, the method manages systematically the relative distortion between a candidate shape and its prototype, accomplishing robustness to noise with less than two prototypes per class, on average. The method uses a flexible matching between components and a flexible grouping of the individual components to be matched. A number of shape transformations are defined, including filling of gaps, so that the method handles broken characters. Also, a measure of the amount of distortion that these transformations cause is given. Classification of character shapes is defined as a minimization problem among the possible transformations that map an input shape into prototypical shapes. Some tests with hand-printed numerals confirmed the methods high robustness level. >

Character recognition without segmentation

A segmentation-free approach to OCR is presented as part of a knowledge-based word interpretation model. It is based on the recognition of subgraphs homeomorphic to previously defined prototypes of characters. Gaps are identified as potential parts of characters by implementing a variant of the notion of relative neighborhood used in computational perception. Each subgraph of strokes that matches a previously defined character prototype is recognized anywhere in the word even if it corresponds to a broken character or to a character touching another one. The characters are detected in the order defined by the matching quality. Each subgraph that is recognized is introduced as a node in a directed net that compiles different alternatives of interpretation of the features in the feature graph. A path in the net represents a consistent succession of characters. A final search for the optimal path under certain criteria gives the best interpretation of the word features. Broken characters are recognized by looking for gaps between features that may be interpreted as part of a character. Touching characters are recognized because the matching allows nonmatched adjacent strokes. The recognition results for over 24,000 printed numeral characters belonging to a USPS database and on some hand-printed words confirmed the methods high robustness level. >

Perceptually stable regions for arbitrary polygons

Zou and Yan have recently developed a skeletonization algorithm of digital shapes based on a regularity/singularity analysis; they use the polygon whose vertices are the boundary pixels of the image to compute a constrained Delaunay triangulation (CDT) in order to find local symmetries and stable regions. Their method has produced good results but it is slow since its complexity depends on the number of contour pixels. This paper presents an extension of their technique to handle arbitrary polygons, not only polygons of short edges. Consequently, not only can we achieve results as good as theirs for digital images, but we can also compute skeletons of polygons of any number of edges. Since we can handle polygonal approximations of figures, the skeletons are more resilient to noise and faster to process.

3D Part Recognition Method for Human Motion Analysis

A method for matching sequences from two perspective views of a moving person silhouette is presented. Regular (approximate uniform thickness) parts are detected on an image and a skeleton is generated. A 3D regular region graph is defined to gather possible poses based on the two 2D-regular regions, one for each view, at a given frame. The matching process of 3D graphs with a model graph results in interpretations of the human motion in the scene. The objective of this system is to reconstruct human motion parameters and use the analytical information for synthesis. Experimental results and error analysis are explained when the system is used to drive an avatar.

Efficient polygonal decomposition into singular and regular regions via Voronoi diagrams

Abstract.A new polygon decomposition into regular and singular regions is defined; it is a concept that is useful for skeleton extraction and part analysis of elongated shapes. Polygon regions that are narrow according to the Voronoi diagram of the polygon are extended through the boundary that is adjacent and quasiparallel. Regular regions are the narrow ones surrounded by smooth quasiparallel contour segments, while singular regions are the polygon regions that are not regular. We present an efficient algorithm to calculate the decomposition and make a comparative study with previous algorithms.

Singularities and regularities on line pictures via symmetrical trapezoids

An algorithm that decomposes a line image into singular and regular regions is presented. We define a contour trapezoid as the building block for the regular regions, and postulate a maximal trapezoid set as the core of the regular regions. Then, we describe an algorithm that calculates a maximal trapezoid set of a polygon and show how to use it to find a skeleton of a polygonal approximation of a contour. Several experiments are explained to show the behavior of the new concept on real images as compared to previous algorithms and criteria.

The Significance of the ProtDeform Score for Structure Prediction and Alignment

Background When a researcher uses a program to align two proteins and gets a score, one of her main concerns is how often the program gives a similar score to pairs that are or are not in the same fold. This issue was analysed in detail recently for the program TM-align with its associated TM-score. It was shown that because the TM-score is length independent, it allows a P-value and a hit probability to be defined depending only on the score. Also, it was found that the TM-scores of gapless alignments closely follow an Extreme Value Distribution (EVD). The program ProtDeform for structural protein alignment was developed recently and is characterised by the ability to propose different transformations of different protein regions. Our goal is to analyse its associated score to allow a researcher to have objective reasons to prefer one aligner over another, and carry out a better interpretation of the output. Results The study on the ProtDeform score reveals that it is length independent in a wider score range than TM-scores and that PD-scores of gapless (random) alignments also approximately follow an EVD. On the CASP8 predictions, PD-scores and TM-scores, with respect to native structures, are highly correlated (0.95), and show that around a fifth of the predictions have a quality as low as 99.5% of the random scores. Using the Gold Standard benchmark, ProtDeform has lower probabilities of error than TM-align both at a similar speed. The analysis is extended to homology discrimination showing that, again, ProtDeform offers higher hit probabilities than TM-align. Finally, we suggest using three different P-values according to the three different contexts: Gapless alignments, optimised alignments for fold discrimination and that for superfamily discrimination. In conclusion, PD-scores are at the very least as valuable for prediction scoring as TM-scores, and on the protein classification problem, even more reliable.

Positively selected amino acid replacements within the RuBisCO enzyme of oak trees are associated with ecological adaptations

[This corrects the article DOI: 10.1371/journal.pone.0183970.].

Graph Comparison by Log-Odds Score Matrices with Application to Protein Topology Analysis

A TOPS diagram is a simplified description of the topology of a protein using a graph where nodes are α-helices and β-strands, and edges correspond to chirality relations and parallel or antiparallel bonds between strands. We present a matching algorithm between two TOPS diagrams where the likelihood of a match is measured according to previously known matches between complete 3D structures. This totally new 3D training is recorded on transition matrices that count the likelihood that a given TOPS feature, or combination thereof, is replaced by another feature on homologs. The new algorithm outperforms existing ones on a benchmark database. Some biologically significant examples are discussed as well. The method can be used whenever frequencies of edge relationship matches are known, as it is the case for several biopolymer structures.

Polygon Partition into Stable Regions

Zou and Yan have recently developed a skeletonization algorithm of digital shapes based on a regularity/singularity analysis; they use the polygon whose vertices are the boundary pixels of the image to compute a constrained Delaunay triangulation in order to find local symmetries and stable regions. Their method has produced good results but it is slow since its complexity depends on the number of contour pixels. This paper presents an extension of their technique to handle arbitrary polygons, not only polygons of short edges. Consequently, not only can we achieve results as good as theirs for digital images but we can also compute skeletons of polygons of any number of edges. Since we can handle polygonal approximations of figures, the skeletons are more resilient to noise and faster to process.