Open AccessPerformant Retrieval Image using Rectangular Mask and Combination of Color, Texture and Shape Descriptors
Author(s) -
Nawal Chifa*,
Abdelmajid Badri,
Yasine Ruichek
Publication year - 2019
Publication title -
international journal of recent technology and engineering
Language(s) - English
Resource type - Journals
ISSN - 2277-3878
DOI - 10.35940/ijrte.c4143.098319
Subject(s) - computer science , artificial intelligence , image retrieval , content based image retrieval , pattern recognition (psychology) , computer vision , local binary patterns , metric (unit) , rotation (mathematics) , texture (cosmology) , digital image , image (mathematics) , image processing , histogram , operations management , economics
The evolution of computer technologies has led to the growth of digital images, which has made the search for similar images in this volume of data a very important research component. Since several works have proposed image search systems entitled CBIR (Content-Based Image Retrieval). This paper presents a new and powerful method for creating CBIR in order to improve the accuracy of search through visual content. The originality of our method lies in its invariance to the rotation of images queries. She consists of applying rectangular masks of different size on the image, and extracting the color descriptor from the visible region on the mask, and then combining the result descriptor to the Uniform Local Binary Pattern (ULBP) texture features and add canny edge features. We compare the query features to the extract ones, using metric distance. We evaluate our techniques using Corel1K and Ukbench dataset. The average precision measured gives good results comparing to the others existing retrieval systems