Maritime security is related to national economic and political interests and is strategically important. One efficient way to accomplish maritime border protection is to use the netted forward scatter radar (FSR). FSR is a special type of bistatic radar that operates in a relatively narrow scattering area along the transmitter-receiver baseline, where the effect of the electromagnetic waves forward scattering on targets is dominant above other scattering mechanisms, and in this case, a forward scatter (FS) cross section may increase by orders of magnitude in comparison with the monostatic radar cross section (RCS). Considered in this study are the major problems of marine forward scattering radar detection and estimation of length of low-profile (small and slow) marine targets using a pre-processing approach. It is based on the assumption that the variation of the phase and amplitude in the Doppler signal signature is stronger inside the FS zone than in an outside region. Two variants of preprocessing algorithms are presented in the study, one for the envelope and the other for the phase. Both variants are based on the use the local variance filtering. The results obtained prove the sufficient improvement in a signal-to-clutter ratio (SCR). Estimation of the marine target length under the low SCR is designed using the assumption of known or previously estimated velocity. Presented results demonstrate high accuracy of length estimation. Considered steps of targets detection and target attributes evaluations are necessary for maritime targets classification. The designed algorithms are verified using a set of experimental records of signals from different marine targets obtained using marine FSR developed by the teams from