Abstract: In this study, 1-D real Doppler signal of each of six number targets was done stationary for better analysis it by using the Mexican Hat Wavelet Scalogram. Firstly, these 2-D T-F distributions were accepted as 2-D radar target images. Secondly, Geometrical moments, central moments, and invariant moments of those 2-D images were obtained respectively. For each of these 2-D images, seven invariant moments were calculated. Method of the Logarithmic Invariant (MLI) was proposed to increase effectively of these invariant moments and classification. Afterwards, obtained feature vector by using proposed method was given to input of Levenberg-Marquardt (LM) neural network classifier. The successful classification performance of the LM was evaluated.