Two-dimensional cumulant-based adaptive enhancer

This paper describes a novel two-dimensional (2-D) adaptive filtering algorithm to enhance 2-D signals of small spatial extent embedded in white or colored Gaussian noise. The adaptive filter coefficients converge to a special 2-D slice of the fourth-order cumulant of the input signal. The proposed algorithm will be referred to as the 2-D cumulant-based adaptive enhancer (2DCBAE). Results of the 2DCBAE are compared with images processed using the 2-D adaptive correlation enhancer (2DACE) and the 2-D least mean square (2DLMS) enhancer.