Banca de QUALIFICAÇÃO: TAIANE VIANA DE CARVALHO

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : TAIANE VIANA DE CARVALHO
DATA : 29/06/2017
HORA: 16:00
LOCAL: LACI
TÍTULO:

Image Segmentation Using data Clustering

PALAVRAS-CHAVES:

Image processing, image segmentation, textures, Gaussian blending, clustering.

PÁGINAS: 35
GRANDE ÁREA: Engenharias
ÁREA: Engenharia Elétrica
RESUMO:

Segmentation is one of the most complex steps in image processing and the focus on research in this area has been increasing. Segmenting means subdividing an image into its distinct regions from pixel properties, such as grayscale and texture. This work presents a new segmentation method based on the extraction of texture characteristics in images. The method consists of generating a set of data based on the values of the pixels present in the image and then grouping them into classes that represent the textures that make up the image. The grouping of the pixels occurs through the application of a clustering technique that initially groups the points of the data set into auxiliary classes using a vector quantization technique. These auxiliary classes will be linked into a single class depending on the degree of sharing of texture characteristics. Traditional methods that calculate similarity or dissimilarity between groups have some limitations in increasing the distance between the means of these groups. An alternative used in this work is a new method that employs the concept of Gaussian mixture calculates the divergence between a data set modeled by Gaussians with the model of a single Gaussian that describes the whole set. This approach aims to propose an efficient method of image segmentation based on textures through the application of a clustering method.

MEMBROS DA BANCA:
Presidente - 2579664 - ALLAN DE MEDEIROS MARTINS
Interno - 1242315 - PABLO JAVIER ALSINA
Externo ao Programa - 1279464 - AGOSTINHO DE MEDEIROS BRITO JUNIOR