The Study On Methods Of Image Rotation, Interpolation And Computation For Geometry Moment Based On Chain Code

The method of rotation is the basic method of image. Based on the author's deep study on actual methods, the course of deriving of image rotation on the basis of shear principle is provided and it is carried on the optimization in the realization, the problems of precision and bottleneck in original method based on shear principle are revolved. Speed and the quality of rotary image can be achieved.In order to resolve the problem of interpolation, A kind of interpolation algorithm based on quadratic surface fitting is addressed after the author study all sorts of interpolation methods. It has very good smoothing function to ladder straight edge which can not be avoided in nearest-neighborhood interpolation. It can solve brightness loss problem in linear interpolation. It also can avoid great mathematical operations, serious time loss and so on the limitation. This algorithm has the high application value.Moment is an important shape descriptor in computer vision. Based on the author's deep study on all methods for geometry moment, two algorithms based on chain code are addressed: one is that the freeman chain codes and coordinates of image region boundary are generated by automation, then the geometry moment can be evaluated by the use of discrete version of Green's theorem ; the other is that the geometry moment is computed based on vertex chain code. Compared with the existing algorithms, the first one has the following merits: accurate result, lower expenses of the memory and higher operating speed; the second one simplify the formulas for the salient image and the result is accurate.