Improved Resolution Synthesis for Image Interpolation
Buyue Zhang, Jay S. Gondek, Morgan T. Schramm, Jan P. Allebach · Technical programs and proceedings/Technical program and proceedings · 2006
The Resolution Synthesis algorithm (ResSynth or RS) is an image interpolation algorithm which computes the minimum mean-squared error estimate of the high-resolution image, given the low-resolution image. ResSynth produces sharper images than the commonly used Bilinear and Bicubic interpolation; and it is computationally efficient. However, the original ResSynth algorithm has some deficiencies, which are: 1) it tends to aggravate noise and JPEG artifacts in the image; 2) it produces halos around strong edges in the image; 3) it occasionally generates “color sprinkles” - pixel errors around the edges. To overcome these problems, we developed the New ResSynth algorithm based on the original algorithm. New ResSynth significantly improves the quality of the interpolated images.