Financial support

Heterogeneous Pattern Recognition and its Applications in Biology and Nanotechnology

Members

Núbia Rosa da Silva, Odemir M. Bruno


Introduction

This project aims to propose an analysis of heterogeneous patterns, develop methods for this new concept and apply them to real problems of pattern recognition. The concept is inspired by the chemistry with the definitions of mixtures. Some mixtures have homogeneous appearance when viewed with the naked eye, however, to observe them at the nanoscale is possible to understand the heterogeneity of its components, for example colloids such as milk, blood and gelatin. In this work, the same concept is applied in pattern recognition. In certain types of images it is possible to see heterogeneous patterns, that not always, can be viewed on a standard scale. Current methods of texture analysis does not adequately describe such heterogeneous patterns. The approach commonly used by researchers analyzes the image information in a global way, taking all its features in an integrated manner. The multi-scale approach, widespread in the literature, analyzes the patterns at different scales, but still preserving the homogeneous analysis, that is, considering the characteristics in a global manner. Therefore, it is necessary to develop methods to identify the different patterns observed at different scales, where it is perceived heterogeneity, thus defining the concept of heterogeneous pattern related to images.



Approaches


Image analysis of histological sections.

Methodology of analysis inspired by the anatomical study of vegetable histology, in which the samples are composed of different regions: Cuticle, Adaxial epidermis, Palisade parenchyma and Spongy parenchyma (1, 2, 3 e 4). Distinguishing the different patterns in each species (A, B e C), it is possible to distinguish these patterns between different images. The comparison of textures between the histological classes was performed obtaining samples of each region textures and comparing them between different images. The correspondence of the patterns is performed in accordance with each region of the histological section, and the patterns that correspond are differentiable according to each region of the cut.




Applications in Nanotechnology.

In this case, the analysis of different patterns is not trivial. The electron microscope image of Field Emission Gun (FEG), below, shows an image of nanostructures of oxalic acid (saturated organic acid).Note that the image does not show features at separated regions. But when you select a specific scale you can see that the image is formed by three distinct patterns that appear irregularly distributed throughout the image.