Restoration of Images of Linear Ultrasonic Array by Neural Network

Image restoration techniques are used to improve the image quality. Inverse filtering and Wiener filters are classical image restoration techniques. In these techniques some simplifying assumptions are necessary (e.g. position invariancy and linearity). In 1988 Zhou1 proposed using Hopfield neural network for image restoration. Advantage of this technique compared to previous techniques is that no assumption about the statistics of the image is necessary e.g. stationarity and position invariancey of imaging system. The main idea of this technique is to define a criterion function and equating it with the energy function of the Hopfield network, then determining weights and initial conditions of the network1. This technique has several shortages: