An Effective Content Based Image Retrieval System Based on Global Representation and Multi-Level Searching

To retrieve important images from a dissimilar collection by using visual queries as search arguments are the arduous and substantial open problems. In this paper the writers have mentioned the designs and implementations of a simple yet very effective Content-Based Image Retrieval (CBIR) system. The colors, textures and the shapes features are the vital parts of this system. With the three main consequent searching steps the searching becomes multilevel. Such propounded systems are very unique as they consider one feature at each step and use the results of the previous step as the input for the next coming step in multilevel pattern whereas in the earlier methods all the features are combined at once for the single-level search of an average CBIR system. The propounded method is very simple and comfortable to adopt. The retrieval grade of the propounded method is valuated using bi-benchmark datasets for an image classification. The above system of methods shows very good results in terms of amelioration in retrieval qualities, when compared with the literature. In proposed work we get accuracy like between 68.15 % to 94.86% in used different features.