Scientific Journal Impact Factor: 3.449 (ISRA), Impact Factor: 1.852

A neural filtering technique is proposed to enhance the digital images when images are contaminated by impulse noise. This filter is obtained by combining nonlinear filter and feed forward neural network with back propagation algorithm. Nonlinear filtering output is converted into one dimensional sequence in four different ways for neural network training. The neural network is trained using three well known images and the network architecture is tuned to optimization. Using this optimized structure, unknown images are tested. Extensive simulation results show that the proposed intelligent filter is superior performance than the other existing neural filters and nonlinear filters in terms of eliminating impulse noise and preserving edges and fine details