Support vector machines and quad-trees applied to image compression
Suaste Gomez Ernesto, G. Adolf, Muhtajuddin Danny · Nordic Signal Processing Symposium · 2004
We present an approach based on Support Vector Machines (SVM) and quad-tree decomposition for compressing still images. Unlike JPEG ,t he method applies the discrete cosine transform (DCT) to regions of variable size, re-scale them to a unique block size, and before the coefficients are entropy encoded, they are approximated by a SVM model. The method was applied to grey scale and colour images and the obtained results improved JPEG’s performance at medium and low bit rates.