Image Classification Using Artificial Neural Networks

An artificial neural network (ANN) is a computing algorithm based on a simplistic model of the brain, or perhaps more accurately, a ganglion. ANN’s are composed of large numbers of simple computing elements (analogous to neurons) that recieve input signals and transmit a signal that is some function of the inputs.
For species indentification, the ANN is trained to recongnize species by being fed inputs specific to one species as well as inputs relating to all other species in the group (i.e., Species A and NOT Species A). The ANN itself then creates the “species classifiers” by selectively weighting the input characters and adjusting its own internal configuration to maximize identification accuracy.

Image encoding using wavelets

Information is efficiently extracted from the images and encoded for input to the ANN using wavelet transformation. This process reduces the total amount of information presented to the system by discarding much of the high-frequency data.