Accurate Image Search using Local Descriptors into a Compact Image Representation
Progress in image retrieval by using low-level features, such as
colors, textures and shapes, the performance is still unsatisfied as
there are existing gaps between low-level features and high-level
semantic concepts.
In this work, we present an improved implementation for the bag
of visual words approach. We propose a image retrieval system
based on bag-of-features (BoF) model by using scale invariant
feature transform (SIFT) and speeded up robust features (SURF).
In literature SIFT and SURF give of good results. Based on this
observation, we decide to use a bag-of-features approach over
quaternion zernike moments (QZM). We compare the results of
SIFT and SURF with those of QZM.
We propose an indexing method for content based search task
that aims to retrieve collection of images and returns a ranked list
of objects in response to a query image. Experimental results
with the Coil-100 and corel-1000 image database, demonstrate
that QZM produces a better performance than known
representations (SIFT and SURF).
Keywords: Content-Based Image Retrieval Systems, feature detection, Bag of visual words.
Download Full-Text
ABOUT THE AUTHORS
Soumia Benkrama
University of Science and Techenology Mohamed Boudiaf, Department of Computer Science, Laboratory Systems Signals Data
Lynda Zaoui
University of Science and Techenology Mohamed Boudiaf, Department of Computer Science, Laboratory Systems Signals Data
Christophe Charrier
University of Caen-Basse Normandie, GREYC, UMR CNRS 6072
Soumia Benkrama
University of Science and Techenology Mohamed Boudiaf, Department of Computer Science, Laboratory Systems Signals Data
Lynda Zaoui
University of Science and Techenology Mohamed Boudiaf, Department of Computer Science, Laboratory Systems Signals Data
Christophe Charrier
University of Caen-Basse Normandie, GREYC, UMR CNRS 6072
