New Index and Search Descriptor Combined Image of Text and Color Applied to Two Databases (Coil-100 and Corel-DB)
El Mehdi, EL Houssif Noureddine, Hassan Silkan, Rmil Mohammed · International Journal of Applied Mathematics & Statistics/International journal of applied mathematics and statistics · 2017
We propose a fast and efficient image retrieval system based on color and texture features. The color features are represented by color histograms and texture features are represented 2-D complex Gabor filtering. It is observed that color features in combination with the texture features derived from the brightness component provide approximately similar results when color features are combined with the texture features using all three components of color, but with much less processing time. To assess and validate our results, many experiments were held using the curve precision/Recall in four color spaces HSV, RGB, L*a*b* and YCbCr. The descriptor was implemented to two different databases Cool-100 and Corel-DB. An analysis of time was held to compare the speed of the treatment pertaining to the proposed approach with other available best approaches