Banca de QUALIFICAÇÃO: ERICK FERNANDES DA CRUZ

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : ERICK FERNANDES DA CRUZ
DATA : 05/08/2019
HORA: 14:00
LOCAL: sala 407, 4º andar, Bloco B, Campus SA da Fundação Universidade Federal do ABC, localizada na Avenida dos Estados, 5001, Santa Terezinha, Santo André, SP
TÍTULO:

Application for Interactive Assisted Image Segmentation

PÁGINAS: 73
GRANDE ÁREA: Ciências Exatas e da Terra
ÁREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
ESPECIALIDADE: Processamento Gráfico (Graphics)
RESUMO:

There are several image segmentation techniques in the literature, and the importance of them can be seen in many applications in areas such as computer vision and pattern recognition. Among these areas we have the interactive image segmentation which uses supervised segmentation algorithms to separate an image in two regions with the use of annotations made by the user. A method that shows interesting results in the area of image segmentation is the Superpixel Laplacian Coordinates, which solved this problem with good results based on the quality of the segmentation and with respect to the execution time even in high resolution images. Although its performance is superior to other algorithms used in interactive segmentation, it presents some costly steps with respect to its execution time, particularly the execution of the SLIC algorithm to generate superpixels. This work aims to create an interactive interactive image segmentation application that is capable of segmenting high resolution images using the Superpixel Laplacian Coordinates algorithm. We reduce the total execution time of the method using state of the art superpixel algorithms found in literature, while maintaining the quality of segmentation. For this we compare 4 superpixel algorithms, including the SLIC itself, using different input parameters to achieve a better comparison of its advantages and disadvantages. In addition we implement the parallelization of some steps of the Superpixel Laplacian Coordinates to further reduce its processing time. Our work aims to generating an interactive segmentation method that is fast enough to be used in high resolution images, as well as providing a free and open source application to be reused in future works.

MEMBROS DA BANCA:
Presidente - Interno ao Programa - 1851117 - HARLEN COSTA BATAGELO
Membro Titular - Examinador(a) Interno ao Programa - 1672977 - JOAO PAULO GOIS
Membro Titular - Examinador(a) Externo à Instituição - WALLACE CORREA DE OLIVEIRA CASACA - UNESP
Membro Suplente - Examinador(a) Interno ao Programa - 1061139 - MARIO ALEXANDRE GAZZIRO
Membro Suplente - Examinador(a) Externo ao Programa - 2127212 - FERNANDO TEUBL FERREIRA