Segmentação de Imagens com Fundo Azul utilizando a multiplicação dos Canais HSV

Lucas Ferrari de OliveiraDeives Mesquita KistGerson Geraldo H. CavalheiroGéri Eduardo MeneghelloMaria Angela Andre Tillmann

In this article we had analyzed the possible use of HSV color space to segment color images of seeds with a bluebackground and with a ruler. This is the first step to extract features and classify seeds for future works. The principal idea was detect on the histogram the blue area, because the major area is composed by the background. We transform the images on the HSV space color and calculate the product between the channels. After the multiplication, the blue background was evidenced. We applied a Gaussian blur filter (19x19) to eliminate the noise and smallest structures. In the next step was detected the value of peak on the histogram and divide for 2 to determine the threshold value. We tested the technique with 121 images of different kinds of seeds and we could segment, correctly, the seeds in the 80.16% of images. This result show the potential of the technique encouraging its use and improvements.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: