Exemplary Content Based Image Retrieval using visual contents & genetic approach

Today, as revolutionary internet & multimedia technologies grows the data storage and image acquisition enables the creation of huge multi-content repository. In this case, it´s mandatory to build appropriate system which efficiently manages those repositories and also facilitates the optimal retrieval techniques. So, Content Based Image Retrieval (CBIR) [1]have become origin of rapid, accurate retrieval technique. It is used to retrieve the similar images across innumerable images using visual contents (color, shape, texture) of their input image from the repository or databases. The optimal retrieval results using CBIR is possible by adopting appropriate content feature extraction methods. There are number of methods analyzed for achieving accuracy but still is not fully accomplished. In this paper, we are going to explore the efficacious Content Based Image Retrieval technique by apt feature extraction methods using Color histogram (HSV), Polar raster edge sampling technique, Fast discrete curvlet transformation for color, shape and texture respectively. Using this approach, the feature vectors of input image for color, shape & texture are extracted and it is fused using genetic coding and then it uses nearest neighborhood classifier(Euclidean distance)for retrieving the similar images of query for the better retrieval results.