Regularization methods for image texture classification

Texture plays an important role for the analysis of multimedia data. In this paper we present several regularization methods for image texture classification. The motivation is that we observe that the commonly used method LSE is unstable in practical applications. This motivates us to develop much more stable method. Regularization is such a technique which successfully suppress the instability due to noise or truncation error when computing. We present several regularization methods include standard regularization (SR), penalized regularization (PR) and total variation based regularization (TVR) for reducing instability in texture analysis, and apply which to image texture classification. Optimization technique is employed for computation. Experiment results demonstrate our new algorithms are superior to LSE and seems promising in practical applications.