The invention relates to the field of array
signal processing, and discloses a high-precision
noise adaptive direction finding method, which comprises the following steps of: establishing a
noise adaptive array direction finding model based on a uniform
linear array to obtain an array receiving
data vector; carrying out de-
impact smoothing preprocessing on the array receiving data; carrying out
noise type judgment by utilizing distribution characteristics of
impact noise, and carrying out
Gaussian smoothing preprocessing on the array receiving data of an
impact noise background by adopting partial median filtering; calculating an array receiving
data covariance matrix, and reconstructing smooth noise through the matrix; then, carrying out characteristic
decomposition on the array receiving
data covariance matrix to obtain a
signal subspace and a noise subspace, and further eliminating noise components in the
signal subspace by utilizing a subspace projection technology; and finally, carrying out direction finding by adopting an ESPRIT direction finding
algorithm. The method is suitable for direction finding of
impact noise and
Gaussian noise at the same time, the implementation speed is high, and the method still has high direction finding precision under the strong
impact noise background with the characteristic index smaller than 0.5.