Abstract:
Keyword-based image retrieval systems have
become popular for many image database
applications. To improve the performance of
keyword-based web image queries, combination of
keyword and visual feature based image retrieval
system is presented in this paper. Firstly, DOM
(Document Object Model) trees are constructed from
collected web pages. And several text blocks are
segmented based on text cohesion. Then, visual
features are extracted from color images in RGB
(Red, Green and Blue) color space by using color
histogram. When user query is entered, text blocks
which contain web images are taken as the
associated texts of corresponding images and
TF*IDF values are used to index web images.
Finally, keyword and visual features are combined
by using Gaussian Mixture Model to produce the
relevance images.
