Based on robust filtering theory,this paper develops polytopic models of uncertain multisensor system and centralized robust fusion estimation by linear matrix Inequality(LMI) method,and proves the exact transforming condition,by which robust centralized estimation can be transformed to the distributed fusion method with the same fusion estimation performance.

Retinex image enhancement method based on robust estimation;

Adaptive image smoothing based on robust estimation;

Motion segmentation based on multi-model robust estimation;

The limitation of standard optical flow estimation is analyzed and the reasons are pointed out,the practicability of optical estimation algorithm has been improved by using robust estimation technology.

Based on the robust estimation in the state-space and the adaptive modification of prediction covariance matrix, a new algorithm was proposed to track Thevenin equivalent parameters.

For the FDI research of an artillery position servo-system,the robust H2/H∞ estimation was used to generate the residual to restrain the influence of model uncertainty and noise on fault detection and diagnosis.