Fast iterative reverse filters using fixed-point acceleration

Abstract Iterative reverse filters have been recently developed to address the problem of removing effects a black box image filter. Because numerous iterations are usually required achieve desired result, processing speed is slow. In this paper, we propose use fixed-point acceleration techniques tackle problem. We present an interpretation existing as and discuss their relationship with gradient descent. then extensive experimental results demonstrate performance named after: Anderson, Chebyshev, Irons, Wynn. also compare these that descent acceleration. Key findings work include: (1) Anderson can make non-convergent filter convergent, (2) T-method technique highly efficient effective, (3) in terms speed, all benefit from one techniques.