Anca Ignat

Analysis and Optimization of Nonsmooth Arches

It is our aim to present a new treatment for some classical models of arches and for their optimization. In particular, our approach allows us to study nonsmooth arches, while the standard assumptions from the literature require

Iris features using Dual Tree Complex Wavelet Transform in texture evaluation for biometrical identification

W^{3, \infty}

Gender recognition with Gabor filters

-regularity for the parametric representation. Moreover, by a duality-type argument, the deformation of the arches may be explicitly expressed by integral formulas. %This also provides a complete solution of the so-called As examples for the shape optimization problems under study, we mention the design of the middle curve of a clamped arch such that, under a prescribed load, the obtained deflection satisfies certain desired properties. In all cases, no smoothness is required for the design parameters.

Pitfalls in using Dual Tree Complex Wavelet Transform for texture featuring: A discussion

Iris characteristics are accurate and reliable for person identification. We used a new method to extract combined features for color and texture characterization using Dual Tree Complex Wavelet Transform. We applied this feature vectors selection on the UPOL iris color image database. The identification of the subsequent images of the same iris was obtained with competitive results.

A study on automatic recognition of positive and negative emotions in speech

The problem of gender identification was approached in this paper starting from images with faces. In order to extract features, Gabor filters were applied using various orientation angles in order to capture significant gender information. Different classifiers were tested (Support Vector Machines, k-NN, discriminant analysis, neural network) on images from the FERET and AR databases. We obtain very good identification results comparable with those obtained by state-of-the-art algorithms.

Iris Recognition Using Color and Texture Features

Dual Tree Complex Wavelet Transforms (DTCWT) have arisen a lot of interest in the last decade. With the possibility of computing characteristics which are rotation invariable and respond well to oscillations around singularities, shift variance, aliasing and the lack of directionality, DTCWT is an interesting tool to be analyzed and employed in a lot of stages of image processing, (among them being texture featuring). Characteristics that are well observable with the human eye, have to be scaled and transformed, in order to be compared and measured under the similarity aspects. A discussion and technical remarks on these aspects are presented, to well understand them, avoiding strange pitfalls.

Combining Features for Texture Analysis

The paper is focused on an experimental study on positive and negative emotion vocal recognition. After some considerations about the positive and negative emotions, the paper gives a short description of the three corpuses used in the work we have accomplished. The paper describes three sets of coefficients used, the statistic features used to generate the three sets of feature vectors and the two classification methods used in this study. The recognition results obtained for every corpus are shown and some conclusions and directions of development are presented.

Iris recognition using statistics on pixel position

We approach the problem of iris recognition by combining both color and texture information. For color features, a well-known global color criterion was extended and for texture we adapted three classical methods: Gray-Level Co-occurrence Matrix (GLCM), Local Binary Patterns (LBP), and Gabor filters. As classifiers we employed k-NN (with Euclidean, Manhattan, Canberra and some variations of these distances), Support Vector Machines (SVM) and a two steps recognition process based on k-NN. We tested the methods using RGB, HSV, and LAB color spaces on two irises datasets (UBIRIS and UPOL). We get better results by combining color and texture features than by only considering color and texture separately.

Automatic Evaluation of Surface Nanostructuring Using Image Processing

In the present paper we consider building feature vectors for texture analysis by combining information provided by two techniques.The first feature extraction method the Discrete Wavelet Transform is applied to the entire image. By computing the Gini index for several subimages of a given texture, we choose one that maximizes this measure. For the selected subimage we apply the second technique a Gabor filter for feature extraction. When we combine the two vectors, the classification results are better than the one obtained using only one set of features. The classification was performed on the Brodatz album, using a naive Bayes classifier.

Tips on Texture Evaluation with Dual Tree Complex Wavelet Transform

In this paper we present experiments made on an extension of a well-known simple global color criterion and we present the results obtained in iris recognition on two known iris databases, UBIRIS and UPOL. All tests were done for three different color spaces, RGB, HSV and LAB, and two discriminative classifiers, k-NN (k - Nearest Neighborhood) and SVM (Support Vector Machine) were used in our experiments. For the k-NN, three classic distances, Canberra, Euclidean and Manhattan and other two new distances we proposed were used in experiments. The results obtained using the extension color criterion are compared with the ones obtained with the original criterion.