Fast algorithm of modified cubic convolution interpolation

Cubic convolution interpolation algorithm is one of the most widely used image interpolation methods. Compared with the linear interpolation, it has a more accurate result, but a lower interpolation speed due to its computational complexity. The cubic convolution algorithm needs sixteen points in two-dimensional image interpolation. The new algorithm proposed in this paper reduces the necessary points to nine by using the correlation among neighboring pixels. Furthermore, the times of multiplication and addition involved in interpolation can be reduced to 53% and 48% respectively of those by original cubic convolution. So the new algorithm achieved lower computational complexity and a higher interpolation speed.