Recursive filtering algorithm to estimate images observed by signal-dependent colored noise

We propose a least-squares linear Altering algorithm for image recovery when the measurement equation responds to general signal-dependent noise model. The noisy components that corrupt the image, multiplicative and additive, are colored and white sequences, respectively. Under the assumption of that the state-space model is unknown and using that the second and fourth-order moments of the signal and the covariance functions of the noises are known and expressed in semi-degenerate kernel form, the recursive filtering algorithm is obtained by an innovation approach. The proposed algorithm is applied to restore an image which is affected by signal-dependent colored noise.