Digital Signal Processing Reference
In-Depth Information
Image Retrieval Based on Color-Spatial Distributing
Feature
Qian Zhao
1,2
, Jialin
Cao
1,2
, and Yueli Hu
2
1
Dept. of Electronic Science and Technology, Shanghai University of Electric Power,
Shanghai, 200090, China
2
School of Mechatronics Engineering and Automation, Shanghai University,
Shanghai 200072, China
Abstract.
In allusion to the problems that the traditional image retrieval method
is prone to lose the spatial information of colors, an image retrieval method
based on color-spatial distributing feature is proposed. According to visual at-
tention computational modethe weighted histogram which reflects the pixel
position importance are constructed after all pixels are weighed by the pixel
color contrast in multi-scale neighborhoods. In the meantime, the spatial rela-
tionship feature of same or similar colors is considered by colors distributing
cohesion. The experiments show that the method mentioned above has high
accuracy and its retrieval results match human visual percept well.
Keywords:
Image Retrieval, Color Histogram, spatial cohesion, image
similatity.
1
Introduction
With the development of multimedia network technology and rapid growth of image
application, large numbers of people are using the internet for searching and browsing
through different multimedia databases. Content-based image retrieval (CBIR) has
become increasingly important in recent decades, which early usually uses color,
texture, shape, spatial relationship etc. low-level features to describe image content,
and take them as the retrieval index. Color is the most intuitive and distinct visual
feature, For every object all has their own color features and the same kind of object
often has similar color information, color features have been applied widely in image
retrieval. Since color histogram is usually easy to extract , fairly convenient to design
similarity measures and fairly insensitive to variations originated by camera rotation
or zooming (Del Bimbo, 1999;Smeulders, et al., 2000), color histogram is the most
traditional and the most widely used way to represent color patterns in an image. But
the conventional color histogram retrieval method is prone to lose the spatial informa-
tion of colors and have the higher feature dimensions[1], which results in poor
retrieval accuracy. Stricker M et al. [2] propose a method based on color block histo-
gram and color block moment, which firstly divides the image into some block
and then calculates color histogram of each block. LI[3] think that human eyes are
