In the context of a stereo-camera, measuring the image quality allows to define the accuracy of the 3D reconstruction. In fact, depending on the precision of the camera position data, on the kind of reconstruction algorithm, and on the adopted camera model, it determines the vertical accuracy of the reconstructed terrain model. Aim of this work is to describe the results and the method implemented to determine the Line Spread Function (LSF) of the Stereoscopic Channel (STC) of the SIMBIOSYS imaging system for the BepiColombo mission. BepiColombo is the cornerstone mission n.5 of the European Space Agency dedicated to the exploration of the innermost planet of the Solar System, Mercury, and it is expected to be launched in 2016. STC is a double push-frame single-detector camera composed by two identical sub-channels looking at ±21° wrt the nadir direction. STC has been designed so to have many optical elements common to both sub-channels. Also the image focal plane is common to the sub-channels and this permits the use of a single detector for the acquisition of the two images, i.e. one for each viewing direction. Considering the novelty of the design, conceived to sustain a harsh environment and to be as compact as possible, the STC unit is very complex. To obtain the most accurate 3D reconstruction of the Mercury surface, a camera model as precise as possible is needed, and an ad-hoc calibration set-up has been designed to calibrate the instrument both from the usual geometrical and radiometrical points of view and more specifically for the instrument stereo capability. In this context LSF estimation was performed with a new method applying a particular oversampling approach for the curve fitting to determine at first the entire calibration system transfer function and at the end the optical properties of the single instrument.