Improvement of Automatic Annotation for Semantic Retrieval of Images using Artificial Neural Networks

Abstract

With the increasing growth of the Internet and the digital imaging industry, the need to organize and separate images is strongly felt. As a result, very large image databases were created. In such a situation, there is a great need for efficient tools and methods to search and retrieve the image. In this proposed method, an automatic image annotation method using artificial neural network is presented. In order to evaluate this proposed method, invocation criteria, accuracy and F criterion have been used. We have also used the standard database of images Msrc, Vistex, Corel as a benchmark database to evaluate the proposed method. The proposed method has the feature of simpler color extraction, has less computational complexity, is more accurate and has a higher speed. Image clustering with the help of SOM artificial neural networks, this clustering method has been seen in less AIA system and has been used in this method with higher quality. The results show that the proposed method of AIA has improved the performance in comparison with the conventional methods of automatic image annotation.