Project ID:2015w14
Students:Niv Zehngut
Supervisor:Nemirovsky-Rotman Shira

Link to project documentation folder

A new method for texture interpolation


Image interpolation is a very important image processing tool, as it occurs in almost all digital photos at some stage, whether this be in Bayer demosaicing for colored images, or in photo enlargement. Specifically, the problem of texture image interpolation is important and challenging, as textures comprise many every-day, natural images. Simple methods such as nearest neighbor and bilinear interpolation usually produce smoother images, which may distort the texture pattern and thus yield undesirable effects in the regions of the textures.

In this project we will implement and test a novel method for texture image interpolation. The proposed algorithm is based on a technique which was previously suggested in a Master’s research at the faculty, and later registered as a patent in the US patent office. The existing method was developed and tested for black and white textures; we will expand the algorithm for colored textures and test its performance for this case.