Artificial neural network color-based positioning system for multiple objects underwater

Tracking objects underwater is a very hard task because of the hostile environment that the water presents. Many parameters should be considered in order to lessen the effect of the hostility. Such parameters underwater are not considered in the vision tracking and/or positioning of objects underwater as long as the image taken is not that distorted. This study proposes an image-based positioning system using neural network for colored objects submerged underwater. The sample data for the Artificial Neural Network model is gathered by empirical methods using actual experimental set-up. The neural network is represented by the following variables: HSI components and panning values as inputs and the coordinates of each colored objects as outputs.