Non-linear perception models in Support Vector Machines image coding | |||
G. Camps-Valls, J.
Gutiérrez, G. Gómez, and J. Malo Submited (june 2005) |
|||
|
|||
Abstract The ability of Support Vector Machines (SVMs) to select relevant features of the signal has been recently combined with perceptually motivated image representations to obtain promising transform coding schemes [Robinson03, Gomez05]. However, the proposed SVM-based coding algorithms used too simple (linear) perception models. In this work, we show that the use of non-linear perception models [Malo05] simplifies the training of SVMs and improves the quality of the reconstructed images at the same compression ratio.
|
|||
|
|||
Keywords: Support Vector Machine, Non-linear Perception Models, Image Coding, Perceptual Metric, Maximum Perceptual Error. References: 9 |
|||
|
|||
|
|