Abstract.A method is presented for radar target rescognition, using the moments of the parameters of the scattered or emitted signals for classification. Large size samples of signal parameters, formed by signals emitted by each of the resolution cells of the searching region, are used to obtain a determined number of normal distributed moments which represent statistical features of the cellular emissions. Taking these moments as components a classification vector is obtained in the moment space. Classification process is carried out assigning every cell of the searching region to one of the previously determined classes during an adaptive process, where the border limiting each class is determined accordingly Neyman -Pearson criterion. The larger size samples, the lower the classification error, which makes possible the distinction of processes very similarly to each other.

No abstract