Cookies Policy
We use cookies to improve our site and your experience. By continuing to browse our site you accept our cookie policy. Find out More
Close
  • Menu
Interest
Topics
Details

Details

001
Publications

2018

Automatic identification of pollen in microscopic images

Authors
Santos, EMDS; Marcal, ARS;

Publication
Lecture Notes in Computational Vision and Biomechanics

Abstract
A system for the identification of pollen grains in bright-field microscopic images is presented in this work. The system is based on segmentation of raw images and binary classification for 3 types of pollen grain. The segmentation method developed tackles a major difficulty of the problem: the existence of clustered pollen grains in the initial binary images. Two different SVM classification kernels are compared to identify the 3 pollen types. The method presented in this paper is able to provide a good estimate of the number of pollen grains of Olea Europea (relative error of 1.3%) in microscopic images. For the two others pollen types tested (Corylus and Quercus), the results were not as good (relative errors of 14.5% and 20.3%, respectively). © 2018, Springer International Publishing AG.

2018

Towards Automatic Calibration of Dotblot Images

Authors
Marcal, ARS; Martins, J; Selaru, E; Tavares, F;

Publication
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Abstract
This paper addresses the issue of calibration (or normalization) of macroarray (dotblot) images. It proposes 3 parameters for the evaluation of the impact on the recorded markers of under- and over-exposure during the experimental acquisition of dotblot images – volume (V), saturation (S) and apparent radius (R). These parameters were evaluated using 101 dotblot images obtained from 16 different experiments, with 404 control markers in total. A procedure to simulate the changes on markers by increasing and decreasing exposure times is also presented. This can be the basis of a normalization procedure for dot blot images, which would be an important improvement in the current laboratory image acquisition protocol, reducing the subjectivity both at the acquisition level and at the subsequent image analysis stage. © 2018, Springer International Publishing AG, part of Springer Nature.

2018

Robust Detection of Water Sensitive Papers

Authors
Marcal, ARS;

Publication
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Abstract
The automatic analysis of water-sensitive papers (WSP) is of great relevance in agriculture. SprayImageMobile is a software tool developed for mobile devices (iOS) that provides full processing of WSP, from image acquisition to the final reporting. One of the initial processing tasks on SprayImageMobile is the detection (or segmentation) of the WSP on the image acquired by the device. This paper presents the method developed for the detection of the WSP that was implemented in SprayImageMobile. The method is based on the identification of reference points along the WSP margins, and the modeling of a quadrilateral that takes into account possible false positive and negative identifications. The method was tested on a set of 360 images, failing to detect the WSP in only 1 case (detection accuracy of 99.7%). The segmentation accuracy was evaluated using references obtained by a semi-automatic method. The average values obtained for the 359 images tested were: 0.9980 (precision), 0.9940 (recall) and 0.9921 (Hammoude metric). © 2018, Springer International Publishing AG, part of Springer Nature.

2018

Evaluation of Chaos Game Representation for Comparison of DNA Sequences

Authors
Marcal, ARS;

Publication
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Abstract
Chaos Game Representation (CGR) of DNA sequences has been used for visual representation as well as alignment-free comparisons. CGR is considered to be of great value as the images obtained from parts of a genome present the same structure as those obtained for the whole genome. However, the robustness of the CGR method to compare DNA sequences obtained in a variety of scenarios is not yet fully demonstrated. This paper addresses this issue by presenting a method to evaluate the potential of CGR to distinguish various classes in a DNA dataset. Two indices are proposed for this purpose - a rejection rate and an overlapping rate. The method was applied to 4 datasets, with between 31 to 400 classes each. Nearly 430 million pairs of DNA sequences were compared using the CGR. © 2018, Springer Nature Switzerland AG.

2016

PhenoSat – a tool for remote sensing based analysis of vegetation dynamics

Authors
Rodrigues, A; Marcal, ARS; Cunha, M;

Publication
Remote Sensing and Digital Image Processing

Abstract
PhenoSat is a software tool that extracts phenological information from satellite based vegetation index time-series. This chapter presents PhenoSat and tests its main characteristics and functionalities using a multi-year experiment and different vegetation types – vineyard and semi-natural meadows. Three important features were analyzed: (1) the extraction of phenological information for the main growing season, (2) detection and estimation of double growth season parameters, and (3) the advantages of selecting a sub-temporal region of interest. Temporal NDVI satellite data from SPOT VEGETATION and NOAA AVHRR were used. Six fitting methods were applied to filter the satellite noise data: cubic splines, piecewise-logistic, Gaussian models, Fourier series, polynomial curve-fitting and Savitzky-Golay. PhenoSat showed to be capable to extract phenological information consistent with reference measurements, presenting in some cases correlations above 70% (n=10; p=0.012). The start of in-season regrowth in semi-natural meadows was detected with a precision lower than 10-days. The selection of a temporal region of interest, improve the fitting process (R-square increased from 0.596 to 0.997). This improvement detected more accurately the maximum vegetation development and provided more reliable results. PhenoSat showed to be capable to adapt to different vegetation types, and different satellite data sources, proving to be a useful tool to extract metrics related with vegetation dynamics. © Springer International Publishing AG 2016.

Supervised
thesis

2017

Reconhecimento Automático de Sinais de Trânsito em Imagens Digitais

Author
Moisés Vungo

Institution
UP-FCUP

2017

Análise Automática de Imagens Pulmonares de Tomografia Computorizada

Author
Susana Augusta Cardoso Leal Lopes

Institution
UP-FCUP

2015

Elaboração de um método semiautomático para detecção do tecido adiposo epicárdico em Imagens de Ressonância Magnética

Author
Cristiana Sofia dos Santos Machado de Araújo

Institution
UP-FCUP

2015

Visão Computacional para veículos aéreos não tripulados (UAV)

Author
Mónica Cristina Gandra da Rocha Salgado

Institution
UP-FCUP