Interpolating Leaf Quad Tree image compression

In the past, wavelet and transform based methods have dominated the field of image compression, however they bring computational complexity to the compression process and are typically more suited to high bit rate compression. The Quad Tree (QT) is a simple image compression method but is not as effective as transform based codecs, such as JPEG. This paper presents a novel variation of the QT called the Interpolating Leaf Quad Tree (ILQT). The ILQT uses QT decomposition, combined with bi-linear interpolation to better approximate image blocks. The ILQT is shown to outperform JPEG and other QT variations in Rate-Distortion (RD) tests at low bit rates, it also outperforms existing methods in terms of perceptual quality.