Abstract
The second fundamental form (SFF) characterizes surface bending as value and direction of normal vector to surface. The value of SFF can be used for blur elimination by simple subtractions of the SFF from image signal. This operation narrows amplitude fronts saving contours as inflection lines. However, it sharpens all small fluctuations and introduces image distortion like noise. Therefore blur recognition and elimination using SFF has to be accompanied by procedure of image estimate optimization in accordance with regularization functional which acts as nonlinear filter. Two iterative methods of original image estimate optimization are suggested. The first method uses dynamic regularization basing on condition of iteration process convergence. The second method implements the regularization in curved space with metric defined on image estimate surface. The given iterative schemes have faster convergence in comparison with known ones.