Image Retrieval based on Wavelet Computation and Neural Network Classification

We describe a simple way to retrieve images from a database. During training a wavelet-based description of the histogram of circular window of each image is first obtained using Daubechies 4- wavelet transformation. Resulting coefficients are used to train a neural network (NN). For image retrieval an image is presented to the system. The system responds with the most similar images. Results are given two different databases. With first database a 98.44% of efficiency with training images was obtained; a 50 % was obtained with images different from those used for training. With the second database corresponding efficiencies were of 88% and 64%.