Image Retrieval Using Normalized Histogram Distance in HSV Color Model

M.Kom, Arif Rahman Image Retrieval Using Normalized Histogram Distance in HSV Color Model. [Artikel Dosen]

[thumbnail of arif-fullpaper-IIS092 (1).pdf] Text
arif-fullpaper-IIS092 (1).pdf

Download (378kB)

Abstract

Color is one of features that used in image
retrieval systems. Pixel color distribution in an image
can be represented by a color histogram. Similarity
degree between images determined by calculate the
distance of histogram. Images with smaller value of
histogram distance should be considered more similar
than the images that have greater distance value. For
color images, histogram distance is calculated for each
color component. In this research HSV color model is
used. Each color component is quantized into 128, 64
and 64 levels respectively. H color component has more
number of levels than the other, because it has biggest
influence for human eye’s perception to colors. A
number of pixels in images are very diverse, so the
histogram should be normalized in order to be invariant
to the image size. Normalization is done by divide the
number of pixels for each level with total number of
pixels in the image. Image retrieval results are ranked
based on histogram distance value.

Item Type: Artikel Dosen
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Depositing User: Arif Rahman
Date Deposited: 09 Apr 2018 09:45
Last Modified: 09 Apr 2018 09:45
URI: http://eprints.uad.ac.id/id/eprint/10122

Actions (login required)

View Item View Item