This paper proposes a novel text localization method in natural images based on the connected components (CC) approach. First, CC are isolated by convolving a multi-scale pyramid with a specifically designed linear spatial filter followed by hysteresis thresholding. Next, non-textual CC are pruned employing a local classifier consisting of a cascade of multilayer perceptron (MLP) fed with increasingly extended feature vectors. The stroke width feature is estimated in linear time complexity by computing the maximal inscribed squares in the CC. Candidate CC and their neighbors are then checked using a more global MLP classifier that takes into account the target CC and their vicinity. Finally, text sequences are extracted in all pyramid levels and fused using dynamic programming. The main contribution of the proposed method is its execution speed, being capable of processing 1080p HD video at nearly 30 frames per second on a standard laptop. In addition, it delivers competitive results in terms of precision and recall on the ICDAR 2013 Robust Reading dataset.

This article presents two novel full quadrant approximations for the arctangent function that are specially suitable for real-time applications. The key point of the proposed approximations is that they are valid in a full quadrant. As a result, they can be easily extended to two and four quadrants. The approximations we define are rational functions of second and third order, respectively. This article provides a comparison of the precision and performance of the proposed functions with the best state-of-the-art approximations. Results show that the third-order proposed function outperforms the existing ones in terms of both precision and performance. The second-order proposed function, on the other hand, is the most suitable one for real-time applications, since it has the highest performance. Furthermore, it attains an adequate precision for most applications in the computer vision field.

This paper presents a new, efficient technique for supervised texture segmentation based on a set of specifically designed filters and a multi-level pixel-based classifier. Filter design is carried out by means of a neural network, which is trained to maximize the filters' discrimination power among the texture classes under consideration. Texture features obtained with these filters are then processed by a classification scheme that utilizes multiple evaluation window sizes following a top-down approach, which iteratively refines the resulting segmentation. The proposed technique is compared to previous supervised texture segmenters by using both synthetic compositions and real outdoor textured images.

In this work, a general procedure for the application of Molecular Quantum Similarity Measures (MQSM) into the Quantitative Structure-Toxicity Relationships (QSTR) field is presented. The methodology is presented from its theoretical background, beginning with the formulation of MQSM and the role of density functions. Other topics, like molecular modelling, electronic density fitting, similarity scaling and molecular superposition are also analysed. The theoretical part is complemented with three different application examples dealing with acute aquatic toxicity related to different fish species. The molecular sets are constituted by different kinds of benzene derivatives. Acceptable correlations are obtained for the aquatic toxicity using quantum similarity derived descriptors, allowing a fairly confident classification of the studied compounds according to their toxicity level.

A new approach, based on the use of fragment Quantum Self-Similarity Measures (MQS-SM) as descriptors of electronic substituent effect in aromatic series, was proposed. The novelty of this approach consists of the fact that the corresponding MQS-SM are not derived, as usual, from ordinary density functions (DF) but from the so-called domain averaged Fermi holes. This approach was applied to the study of substituent effects on the acidobasic dissociation constants in 6 series of para-substituted aromatic carboxylic acids. It has been shown that MQS-SM calculated for each particular set of acids correlate with the Hammett substituent constants. As a consequence, the corresponding similarity measures can be used as new efficient descriptors of the substituent effect, which hopefully could replace empirical sigma constants in QSAR models.

This work presents a schematic description of the theoretical foundations of quantum similarity measures and the varied usefulness of the enveloping mathematical structure. The study starts with the definition of tagged sets, continuing with inward matrix products, matrix signatures, and vector semispaces. From there, the construction and structure of quantum density functions become clear and facilitate entry into the description of quantum object sets, as well as into the construction of atomic shell approximations (ASA). An application of the ASA is presented, consisting of the density surfaces of a protein structure. Based on this previous background, quantum similarity measures are naturally constructed, and similarity matrices, composed of all the quantum similarity measures on a quantum object set, along with the quantum mechanical concept of expectation value of an operator, allow the setup of a fundamental quantitative structure- activity relationship (QSPR) equation based on quantum descriptors. An application example is presented based on the inhibition of photosynthesis produced by some naphthyridinone derivatives, which makes them good herbicide candidates. © 2004 Wiley

A novel methodology to derive adjacency matrices for 3-dimensional molecules is presented in this work. These matrices, which are derived from atomic quantum similarity calculations, allow redefining several adjacency matrix-based topological indices, such as Randić, Zagreb or Chi's. The present derivation is built upon a previous work, where a simpler class of density function was used to describe the atoms composing the molecule, and the present proposal suggests using fitted atomic densities from the Atomic Shell Approximation procedure, which has been proved to appropriately reproduce, in its application space, atomic ab initio densities. The construction of such matrices as well as the derived indices are presented, along with some QSPR, where these topological indices are used as molecular descriptors for heat of formation, inhibition, and toxicity with promising results.

A novel method for computing new descriptors to construct Quantitative Structure–Toxicity Relationships is presented. First, a brief review on the classical graph theory is presented and, then, the link with molecular similarity is drawn. In the applications section, molecular topological indices are calculated using the interatomic Molecular Quantum Similarity Measure with a Coulomb weight operator. The use of similarity matrices instead of classical topological ones has been adopted according to the connection between molecular topology and the general theory of Quantum Similarity. Afterwards, the molecular descriptors, which include the structural information necessary to properly describe the system, are employed to derive numerical correlation with toxicities. The QSAR model is built using a multilineal regression technique.Finally, some application examples are presented, including polycyclic aromatic hydrocarbons and aquatic toxicants, demonstrating the applicability of the exposed methodology.

ChemInformVolume 36, Issue 22 Other Subjects Topological Quantum Similarity Indices Based on Fitted Densities: Theoretical Background and QSPR Application. Anna Gallegos, Anna Gallegos Dep. Med. Chem., Astra Zeneca R&D, S-431 83 Moelndal, Swed.Search for more papers by this authorXavier Girones, Xavier Girones Dep. Med. Chem., Astra Zeneca R&D, S-431 83 Moelndal, Swed.Search for more papers by this author Anna Gallegos, Anna Gallegos Dep. Med. Chem., Astra Zeneca R&D, S-431 83 Moelndal, Swed.Search for more papers by this authorXavier Girones, Xavier Girones Dep. Med. Chem., Astra Zeneca R&D, S-431 83 Moelndal, Swed.Search for more papers by this author First published: 09 May 2005 https://doi.org/10.1002/chin.200522220AboutPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat No abstract is available for this article. Volume36, Issue22May 31, 2005 RelatedInformation

The nature of the Ga-Ga bond in the naked dianion PhGaGaPh2- and its Na+-coordinated counterpart is discussed using the recently proposed formalism based on the analysis of domain-averaged Fermi holes. The analysis shows clearly that three electron pairs, i.e., two sigma and one pi, contribute to the Ga-Ga bonding interactions in the free dianion PhGaGaPh2- and in the sodium-complexed species (PhGaGaPh)Na-2. The eigenvalues and the eigenvectors of the Fermi hole indicate, however, that the Ga-Ga bonding situation does not correspond to classical triple bonds and should not be interpreted using classical bonding models. The reason is that only one of the three electron pairs involved in Ga-Ga bonding, namely that contributing to the Ga-Ga pi bond, corresponds to an "ordinary" shared electron pair bond in the sense of Lewis. The bonding interactions of the other two pairs are much more complex and have no classical counterparts. The eigenvalues and the eigenvectors of the latter Fermi holes cannot be considered as fully bonding, but they have partial lone-pair character. This is why the calculated bond orders yield values which are close to a weak double bond. Essentially the same bonding picture holds also for the Na+-coordinated species, but the coordination of sodium causes further weakening of the contributions of all three electron pairs to the Ga-Ga bonding. This weakening yields an effective reduction of the multiplicity of the Ga-Ga bond. The calculated bond order therefore has a value which is otherwise typical for single bonds.

In this work, an extension of the already studied Topo‐Geometrical Superposition Approach (TGSA) is presented. TGSA, a general‐purpose, fast, automatic, and user‐intuitive three‐dimensional molecular alignment procedure, was originally designed to superpose rigid molecules simply based on atomic numbers, molecular coordinates, and connectivity. The algorithm is further developed to enable handling rotations around single bonds; in this way, common structural features, which were not properly aligned due to conformational causes, can be brought together, thus improving the molecular similarity picture of the final alignment. The present procedure, implemented in Fortran 90 and named TGSA‐Flex, is deeply detailed and tested over four molecular sets: amino acids, nordihydroguaiaretic acid (NDGA) derivatives, HIV‐1 protease inhibitors, and 1‐[2‐hydroxyethoxy)methyl]‐6‐(phenylthio)thymine (HEPT) derivatives. TGSA‐Flex performance is evaluated by means of computational time, number of superposed atoms (also comparing it with respect to the rigid approach), and index of fit between the compared structures. © 2003 Wiley Periodicals, Inc. J Comput Chem 25: 153–159, 2004

A new approach is presented for the development of quantitative structure–property relations (QSPR) based on the extraction of relevant molecular features with self-organizing maps and the use of a modified fuzzy-ARTMAP classifier for variable prediction. The present methodology is demonstrated for the development of a QSPR for the aqueous-phase infinite dilution activity coefficient γ∞, based on a data set of 325 diverse organic compounds. The QSPR was developed using a set of 11 molecular descriptors (four connectivities vχ1–4, Coulomb self-similarity measure, electron–nuclear attraction, dipole moment, sum of atomic numbers, number of filled levels, average polarizability, and nuclear–nuclear repulsion). The final set of molecular descriptors was selected from an initial pool of 23 topological and quantum chemical descriptors, including six molecular quantum similarity measures, by means of a topological analysis of self-organization of the data set. Additional interpolated information to enhance the training of the neural system was obtained from the self-organization analysis. The resulting fuzzy-ARTMAP–based QSPRs performed with errors that were on the average seven times smaller compared to previous published models. The use of only four molecular quantum similarity measures proved to be sufficient for building a lnγ∞ fuzzy-ARTMAP–based QSPR with reasonable accuracy. © 2004 American Institute of Chemical Engineers AIChE J, 50:1315–1343, 2004

A functional model for the in vitro inactivation of voltage-dependent K channels is developed.The model q expresses the activity as a function of the aminopyridine pK , the interaction energy with the receptor, and a quotient a of partition functions.Molecular quantum similarity theory is introduced in the model to express the activity as a function of the principal components of the similarity matrix for a series of agonists.To validate the model, a set of five active (protonated) aminopyridines is considered: 2-aminopyridine, 3-aminopyridine, 4-aminoquinoleine, 4- aminopyridine, and 3,4-diaminopyridine.A regression analysis of the model gives good results for the variation of the observed activity with the overlap similarity index when pyridinic rings are superposed.The results support the validity of the model, and the hypothesis of a ligand-receptor entropy variation depending mainly on the nature of the ligand.In addition, the results suggest that the pyridinic ring must play an active role in the interaction with the receptor site.This interaction with the protonated pyridinic nitrogen can involve a cation- p interaction or a donor hydrogen bond.The amine groups, at different relative positions of the pyridinic nitrogen, can form one or more hydrogen bonds due to the C symmetry of the inner part of the pore in the K channel. q 4 2003 Elsevier Science B.V. All rights reserved. q

A functional model for the in vitro inactivation of voltage-dependent K(+) channels is developed. The model expresses the activity as a function of the aminopyridine pK(a), the interaction energy with the receptor, and a quotient of partition functions. Molecular quantum similarity theory is introduced in the model to express the activity as a function of the principal components of the similarity matrix for a series of agonists. To validate the model, a set of five active (protonated) aminopyridines is considered: 2-aminopyridine, 3-aminopyridine, 4-aminoquinoleine, 4-aminopyridine, and 3,4-diaminopyridine. A regression analysis of the model gives good results for the variation of the observed activity with the overlap similarity index when pyridinic rings are superposed. The results support the validity of the model, and the hypothesis of a ligand-receptor entropy variation depending mainly on the nature of the ligand. In addition, the results suggest that the pyridinic ring must play an active role in the interaction with the receptor site. This interaction with the protonated pyridinic nitrogen can involve a cation-pi interaction or a donor hydrogen bond. The amine groups, at different relative positions of the pyridinic nitrogen, can form one or more hydrogen bonds due to the C(4) symmetry of the inner part of the pore in the K(+) channel.

A new approach allowing the theoretical modeling of the electronic substituent effect is proposed. The approach is based on the use of fragment Quantum Self-Similarity Measures (MQS-SM) calculated from domain averaged Fermi Holes as new theoretical descriptors allowing for the replacement of Hammett sigma constants in QSAR models. To demonstrate the applicability of this new approach its formalism was applied to the description of the substituent effect on the dissociation of a broad series of meta and para substituted benzoic acids. The accuracy and the predicting power of this new approach was tested on the comparison with a recent exhaustive study by Sullivan et al. It has been shown that the accuracy and the predicting power of both procedures is comparable, but, in contrast to a five-parameter correlation equation necessary to describe the data in the study, our approach is more simple and, in fact, only a simple one-parameter correlation equation is required.

The application of Molecular Quantum Similarity Measures (MQSM) to correlate biological activities for three different sets of steroids is reported. A general protocol for the generation of descriptors is detailed, thus covering molecular superposition, electronic density fitting, and quantum similarity calculation issues. Satisfactory Quantitative Structure-Activity Relationship (QSAR) models (r(2) in [0.69,0.94] and q(2) in [0.59,0.73]), comparable to previous studies, are obtained in all cases, where steroid binding affinities to different enzymes are studied. In this work, MQSM, properly scaled using Carbó Index, are related to activity using a Partial Least Squares routine.

The nature of bonding in isocoordinated molecules of SF6 and CLi6 was analyzed using the recently proposed approach based on the scrutiny of the so-called domain-averaged Fermi holes. It has been shown that although the molecule of SF6 does not satisfy the charge criterion of hypervalence, the actual picture of bonding is consistent with the traditional hypervalent model assuming the existence of six localized albeit very polar SF bonds around the central atom. On the other hand, while the molecule of CLi6 represents the ideal candidate for hypervalence according to charge criterion, the picture of bonding is, in this case, considerably different from what the concept of hypervalence is traditionally associated with and can be better characterized by the term hypercoordination.