Alejandro Frery

Hyperspectral images clustering on reconfigurable hardware using the k-means algorithm

Unsupervised clustering is a powerful technique for understanding multispectral and hyperspectral images, k-means being one of the most used iterative approaches. It is a simple though computationally expensive algorithm, particularly for clustering large hyperspectral images into many categories. Software implementation presents advantages such as flexibility and low cost for implementation of complex functions. However, it presents limitations, such as difficulties in exploiting parallelism for high performance applications. In order to accelerate the k-means clustering, a hardware implementation could be used. The disadvantage in this approach is that any change in the project requires previous knowledge of the hardware design process and can take several weeks to be implemented. In order to improve the design methodology, an automatic and parameterized implementation for hyperspectral images has been developed in a hardware/software codesign approach. An unsupervised clustering technique k-means that uses the Euclidian distance to calculate the pixel to centers distance was used as a case study to validate the methodology. Two implementations, a software and a hardware/software codesign one, have been implemented. Although the hardware component operates at 40 MHz, being 12.5 times less than the software operating frequency (PC), the codesign implementation was approximately 2 times faster than software one.

Image Processing for Computer Graphics and Vision

Image processing is concerned with the analysis and manipulation of images by computer. Providing a thorough treatment of image processing, with an emphasis on those aspects most used in computer graphics and vision, this fully revised second edition concentrates on describing and analyzing the underlying concepts of this subject. As befits a modern introduction to this topic, a good balance is struck between discussing the underlying mathematics and the main topics of signal processing, data discretization, the theory of color and different color systems, operations in images, dithering and half-toning, warping and morphing, and image processing. Significantly expanded and revised, this easy-to-follow text/reference reflects recent trends in science and technology that exploit image processing in computer graphics and vision applications. Stochastic image models and statistical methods for image processing are covered, as is probability theory for image processing, and a focus on applications in image analysis and computer vision. Features: Includes 5 new chapters and major changes throughout Adopts a conceptual approach with emphasis on the mathematical concepts and their applications Introduces an abstraction paradigm that relates mathematical models with image processing techniques and implementation methods - used throughout to help understanding of the mathematical theory and its practical use Motivates through an elementary presentation, opting for an intuitive description where needed Contains adopted innovative formulations whenever necessary for clarity of exposition Provides numerous examples and illustrations, as an aid to understanding Focuses on the aspects of image processing that have importance in computer graphics and vision applications Offers a comprehensive introductory chapter for instructors This comprehensive text imparts a good conceptual understanding of the topic, as a basis for further study, and is suitable both as a textbook and a professional reference. The current extended edition is a must-have resource and guide for all studying or interested in this field.

Image formation in vibro-acoustography with depth-of-field effects.

We study the image formation of vibro-acoustography systems based on a concave sector array transducer taking into account depth-of-field effects. The system point-spread function (PSF) is defined in terms of the acoustic emission of a point-target in response to the dynamic radiation stress of ultrasound. The PSF on the focal plane and the axis of the transducer are presented. To extend the obtained PSF to the 3D-space, we assume it is a separable function in the axial direction and the focal plane of the transducer. In this model, an image is formed through the 3D convolution of the PSF with an object function. Experimental vibro-acoustography images of a breast phantom with lesion-like inclusions were compared with simulated images. Results show that the experimental images are in good agreement with the proposed model.

Sampling and Reconstruction

Analog images must be sampled before being represented on the computer. In order to be visualized they must be displayed on a device that is able to reconstruct color, such as a CRT monitor. The sampling process is called rasterization; it is carried out by some sampling device, such as a scanner or TV camera, or by discretizing a continuous mathematical description of a scene, as in the case of the rendering process of image synthesis systems. The display device reconstructs the discrete image, creating an optical-electronic version that is perceived by the eye. Thus, an understanding of sampling and reconstruction is a good foundation for producing good-quality images.

Clustering of fully polarimetric SAR data using finite G p 0 mixture model and SEM Algorithm

This paper presents a novel method for clustering multilook polarimetric SAR images by combining the stochastic expectation-maximization (SEM) algorithm with the mixture of Gp 0 distributions, using the method of moments for parameter estimation. The pixel values of multilook SAR data are complex covariance matrices, and they are described by mixtures of gp 0 laws. This distribution can describe different type of targets; like urban areas, forest and pasture. The proposed clustering technique can be applied to unsupervised classification and segmentation process. Experiments with real image data provide good results.

Warping and Morphing

This chapter studies topological filters designed to change the shape of the objects of an image. This process is called deformation or warping, and therefore we talk about warping filters. Together with amplitude filters, which change the image’s color information, warping filters can be used to create a transition between images of different objects, in a technique known as morphing. Warping and morphing filters are important in many applications, from the correction of preexisting image distortions to the creation of special effects in the entertainment industry.

JDIPF: Java digital image processing framework

A tool created with the aim of contributing to the teaching of image processing techniques is presented. It allows the specification of filters in a simple and intuitive manner. Being a Java application, the system is portable and runs in many different environments.

Identifying Communities in Social Media with Deep Learning

This work aims at analyzing twitter data to identify communities of Brazilian Senators. To do so, we collected data from 76 Brazilian Senators and used autoencoder and bi-gram to the content of tweets to find similar subjects and hence cluster the senators into groups. Thereafter, we applied an unsupervised sentiment analysis to identify the communities of senators that share similar sentiments about a selected number of relevant topics. We find that is able to create meaningful clusters of tweets of similar contents. We found 13 topics all of them relevant to the current Brazilian political scenario. The unsupervised sentiment analysis shows that, as a result of the complex political system (with multiple parties), many senators were identified as independent (19) and only one (out of 11) community can be classified as a community of senators that support the current government. All other detected communities are not relevant.

Modeling Dengue vector population using remotely sensed data and machine learning

Mosquitoes are vectors of many human diseases. In particular, Aedes ægypti (Linnaeus) is the main vector for Chikungunya, Dengue, and Zika viruses in Latin America and it represents a global threat. Public health policies that aim at combating this vector require dependable and timely information, which is usually expensive to obtain with field campaigns. For this reason, several efforts have been done to use remote sensing due to its reduced cost. The present work includes the temporal modeling of the oviposition activity (measured weekly on 50 ovitraps in a north Argentinean city) of Aedes ægypti (Linnaeus), based on time series of data extracted from operational earth observation satellite images. We use are NDVI, NDWI, LST night, LST day and TRMM-GPM rain from 2012 to 2016 as predictive variables. In contrast to previous works which use linear models, we employ Machine Learning techniques using completely accessible open source toolkits. These models have the advantages of being non-parametric and capable of describing nonlinear relationships between variables. Specifically, in addition to two linear approaches, we assess a support vector machine, an artificial neural networks, a K-nearest neighbors and a decision tree regressor. Considerations are made on parameter tuning and the validation and training approach. The results are compared to linear models used in previous works with similar data sets for generating temporal predictive models. These new tools perform better than linear approaches, in particular nearest neighbor regression (KNNR) performs the best. These results provide better alternatives to be implemented operatively on the Argentine geospatial risk system that is running since 2012.

Spatio-temporal changes in water quality in an eutrophic lake with artificial aeration

Abstract In this work we present novel results concerning water quality changes in an eutrophic water body connected with an artificial aeration system installed in it. Sixty one in-situ and laboratory measurements of biogeochemical variables were recorded monthly between October 2008 and June 2011 to evaluate temporal and spatial changes in San Roque reservoir (Argentina). t-Student mean difference tests, carried out over the whole period, showed with 95% confidence that a monitoring point located at the centre of the water body is representative of the chemical behaviour of the reservoir. Thermal stratification was observed in all sampling sites in the summer, but the frequency of these episodes was markedly lower in bubbling zones. Mean chlorophyll-a concentrations were 58.9 μg·dm−3 and 117.0 μg·dm−3 in the absence and in the presence of thermocline respectively. According to the t-Student test, this difference was significant, with p < 0.001. Phosphate release from sediments was corroborated under hypoxia conditions. ANOVA one way analysis did not show significant spatial differences for any variable. Mean normalize spatial index (MENSI) was developed to compare data from different regions affected by high temporal variability. It proved to be useful to quantify spatial differences. Structure analysis of temporal series was used to scrutinize both chemical and spatial association successfully. Three chemically different zones were determined in the reservoir. This study demonstrated that spatial comparisons by means of marginal statistics may not be an adequate method when high temporal variation is present. In such a case, temporal structure analysis has to be considered.