教学课件：《数字图象处理》.ppt

Higher-Order Interpolations Higher-order interpolations can retain details Exact fit: number of coefficients in polynomial same as number of points Least-square error: more points than coefficients Usually implemented using convolution, e.g., Cubic convolution (Sin x) / x Typically 16 or more points Computationally intensive Bilinear interpolation reduces detail Solve 4 equations in 4 unknowns Effective compromise Applications of Geometric Operations Image Registration (图像配准） Needed for comparing images, algebraic operations and clutter removal (motion detection, change detection) If fiducia