Academic Journals Database
Disseminating quality controlled scientific knowledge

Amelioration of Colour Averaging Based Image Retrieval Techniques using Even and Odd parts of Images

ADD TO MY LIST
 
Author(s): Dr. H.B.Kekre, | Sudeep D. Thepade, | Varun K. Banura

Journal: International Journal of Engineering Science and Technology
ISSN 0975-5462

Volume: 2;
Issue: 9;
Start page: 4238;
Date: 2010;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: CBIR | Colour averaging | Row Mean | Column Mean | Diagonal Mean | Even Image | Odd Image.

ABSTRACT
Amelioration of the colour averaging based image retrieval techniques given in [1], with the help of Even and Odd image parts is the theme of this paper. The reflection of original image is taken across horizontal and vertical directions to get flip image. The even part of image is obtained by adding original and flip images and the odd part of image is obtained by subtracting flip image from original image. Then the colour averaging methods like row mean, column mean, forward diagonal mean, backward diagonal mean, row & column mean and forward & backward diagonal mean are applied on original image, even part of image and odd part of image. The colour averages (feature vectors) of original image are considered in combination with even colour averages and odd colour averages respectively to get three image retrieval methods per colour averaging techniques as original, original with even, original with odd. The proposed content based image retrieval (CBIR) techniques are tested on a generic image database having 1000 images spread across 11 categories. For each proposed CBIR technique 55 queries (5 per category) are fired on the image database. To compare the performance of image retrieval techniques average precision and recall are computed for all queries. The results have shown improved performance (higher precision and recall values) with proposed methods compared to the simple original image feature vectors. In the discussed image retrieval methods original with even proves to be better than original, while original with odd gives the worst performance. Forward diagonal mean (FDM) gives the highest performance in the discussed six methods of featurevector selection for respective CBIR method (original, original+even, original+odd). In all forward diagonal mean (FDM) feature vectors, original with even proves to be the best with highest crossover point value of precision and recall.
Affiliate Program      Why do you need a reservation system?