A non-Newtonian gradient for contour detection in images with multiplicative noise

In this paper, a new operator for contour detection in images with multiplicative noise is presented. Traditional methods of edge detection, as those based in gradient operator or measures of variance, follow a logic and a math formulation in correspondence with the Differential and Integral Calculus of Newton. This work presents a new operator of non-Newtonian type which had shown be more efficient in contour detection than the traditional operators. Like the regular gradient, a non-Newtonian gradient can be used in a number of more complex methods, which shows its potential in the contours detection in images affected by multiplicative noise.