No abstract

The biggest problem in the research of Content Based Image Retrieval (CBIR) is bridge the gap between low-level features and high-level semantics., Still many shortcomings for image retrieval system only with the low level visual features due to the semantic space. It is better for the relevance feedback based on the user involvement in image retrieval system. By using the help of user's feedback, the resultant high-level semantic will be obtained. Relevance feedback is a technique for incorporating semantic information in image retrieval. This paper illustrates a development upon a relevance feedback approach that utilizes semantic grouping and clustering technique to close the distance between low-level features and high-level semantics. Distinctively, the past system is improved by incorporating the images in the same group as the query image in the collection of retrieved images. Shared the retrieval results with relevance feedback technology, image feature dimensional reduction was prepared using the Clustering concepts. The given system reduces semantic gap and the storage of image signatures, and also improves the retrieval efficiency and performance. The result shows the efficiency of our proposed system.

No abstract

This paper deals with the implementation of Neural Network based face recognition system. As we know that face recognition system is one of the biometric information processing which has speed up in the last few decades. The developed algorithm for the face recognition system originates an image based approach, which uses the Two-Dimensional Discrete Cosine Transform (2D-DCT) to compress image, and then Self Organizing Map (SOM) Neural Network to recognize the face and its simulated in MATLAB. With the help of 2D-DCT the image vectors are extracted and these vectors sends to the neural network classifier which is developed using self organizing map, algorithm to recognize trained faces, faces with variations in expressions, changes of illumination, upto certain degrees. The alternate way of the same face recognition system is developed with the help of principle component analysis (PCA) instead of Two Dimensional Cosine Transform and Self-Organizing Map Neural Network to recognize the faces. In this proposed algorithm we use unsupervised single neural network as a classifier for both Two Dimensional Discrete Cosine Transform (2D-DCT) and Principal Component Analysis (PCA).