Some induction motors present saliencies that periodically modify the air-gap flux following a sinusoidal shape. Consequently, these modulations are reflected in the transient leakage inductance. Therefore, transient leakage inductance can be expressed as a constant term or offset, which comes from the symmetrical machine, plus all motor saliency sinusoids, modulated by the mentioned saliencies. Since saliencies are linked to the rotor position or flux, calculation of transient leakage inductance enables encoderless vector control. A way to calculate transient leakage inductance is to excite the machine with voltage steps and estimate the resulting phase current slopes since they are inversely proportional to phase transient inductance. For creating a vector, only the three current slopes resulting from a single active state are sufficient. The vector sum of the three current slopes leads to the so-called saliency-offset vector, which is a vector composed of an offset, which is created due to the contribution of the transient inductance offset, plus several saliency phasors, superposed to the offset. To access saliencies, it is critical to accurately identify the offset of the saliency-offset vector, especially in induction motors since offset is much bigger than saliencies. This article presents an advanced signal processing scheme that identifies the offset in the saliency-offset vector, assuming offset amplitude differs depending on the voltage step applied. Using a feedforward scheme, the offset is eliminated, leading to a saliency vector where rotor slotting can be extracted from. A novel excitation is proposed to reduce additional current distortions coming from the one-active excitation.