Image classification using Hybrid MLP method

Content based image retrieval system is a research area which is growing very fast, here the query image’s visual contents are used to search similar or exact images from a very large scale database of images. An effective system is proposed in our method in which both the semantically and visually relevant features are used for retrieval of the related images. For an efficient retrieval of image, it is necessary for a classification algorithm to be much efficient that it can find and locate the image at a different like in a compressed level. Our system proposes a novel training algorithm for MultiLayer Perceptron Network to classify compressed images more efficiently. Here a hybrid MLP method is proposed for retrieval several features and shortens the semantic gap between low-level visual feature and highlevel perception.