Adaptive adjustment of the CNN output function to obtain contrast enhancement
In this paper we propose an adaptive modification of the output function of the CNN (Cellular Neural Network) model to perform contrast enhancement of an image. First, we define the output function to operate in the interval [0,1] with variable saturation limits in order to adapt the behaviour of the network to the grey levels in the neighbourhood of every cell. Then we propose a three-layers CNN where the mean value of the neighbourhood of a pixel is obtained by the first layer and the calculation of the mean deviation of the pixel values from the mean in the same neighbourhood is carried out by the second one. These parameters are control signals that define the saturation limits of the piecewise linear output function of each cell in the third layer, the output of the network, adapting it to the neighbourhood of each cell. Some examples are presented to demonstrate the capabilities of the model.
Unable to display preview. Download preview PDF.
- E. B. Kosmatopoulos, M. M. Polycarpou, M. A. Christodoulou, P. A. loannou. “High-Order Neural Network Structures for Identification of Dynamical Systems”. IEEE Transactions on Neural Networks, Vol. 6, No. 2, pp. 422–431.Google Scholar
- F. J. López Aligué, M. I. Acevedo Sotoca, M. A. Jaramillo Morán. “A High Order Neural Model”. Lecture Notes in Computer Science, No. 686, “New Trends Neural Computation”. pp. 108–113, Springer-Verlag, Berlin. June 1993.Google Scholar