Abstract:

This paper presents a new interpolation algorithm, which is based on the content division of images. In particular, the algorithm first utilizes the D CT coefficient in the HSV space to divide a given image into the target and the background, and divides the target into two different grades in importance. It then adopts different interpolation methods depending on the different requirements of the target, background and different grades. Extensive experiments show that the new algorithm improves the time complexity of scaling while delivering results comoarable to classic algorithms.

Key words: DCT coefficient, content division, cube-convolution interpolation, bilinear interpolation, neighbor interpolation