Paper Info
Title
A novel content-based image retrieval approach using fuzzy combination of color and texture
Abstract
A novel content-based image retrieval approach using fuzzy combination of color and texture image features is expressed in this paper. To accomplish this, color histogram and autocorrelogram of the partitioned image as color features and Gabor wavelet as texture feature are used. Color and texture features are separately extracted and kept as feature vectors. In comparing images similarity stage, the difference between feature vectors is computed. Since center of image is more important, higher weight is considered for it in the comparison of autocorrelograms, and due to this fact the retrieval performance is improved; and also finding the most similar regions using autocorrelogram of the other regions, makes the algorithm more invariant to rotation and to somehow to changing the viewing angle. To make the final decision about images similarity ratio, a fuzzy rule-based system is utilized. Experimental results show this method improved the performance of contentbased image retrieval systems.
Year
DOI
Venue
2011
10.1007/978-3-642-23896-3_2
AICI (3)
Keywords
Field
DocType
color feature,fuzzy combination,retrieval performance,novel content-based image retrieval,texture image feature,color histogram,contentbased image retrieval system,feature vector,texture feature,partitioned image
Computer vision,Feature vector,Pattern recognition,Color histogram,Image texture,Feature (computer vision),Computer science,Image retrieval,Artificial intelligence,Content-based image retrieval,Visual Word,Color image
Conference
Volume
ISSN
Citations 
7004
0302-9743
3
PageRank 
References 
Authors
0.38
16
2
Name
Order
Citations
PageRank
Mohsen Fathian141.41
Fardin Akhlaghian Tab2154.45