A design problem for image processing (Q4511212)

The paper studies two-dimensional images being described by a real-valued function \(u_0\), defined on a rectangular domain \({\mathcal D}: 0\leq x_1\leq L_1\), \(0\leq x_2\leq L_2\). The values \(u_0= (x_1, y_1)\) may represent the gray levels ranging from \(0\) to \(255\). The processed values of the initial image \(u_0\) will be denoted by \(u(x_1, x_2, t)\) for \(t> 0\), \((x_1,x_2)\in {\mathcal D}\), where \(t\) will be the parameter time. This then becomes the setting to smooth noise and perturbations in \(u_0\) by applying diffusion. Thus \({\partial u\over\partial t}= \Delta^2u\) where \(n^v\Delta u=0\). The patterns in the image can then be stabilized depending on the parameters of the differential equation. A method is clearly shown for designing parameters of dynamical systems of reaction-diffusion type such that specific nodes of the images become stable. A very complete set of graphs surrounding the description of the parameters are included within the paper. For the entire collection see [Zbl 0940.00036].