In addition, an optimisation procedure called kinematic fitting will be detailed in order to improve the measured resolutons on the observables. In particular, the emphasis will be put on the calibrations of the time of flight detector and the electromagnetic calorimeters. After the introduction of GPDs and their link with the DVCS process, the experimental situation will be depicted. This thesis shows the most recent results on the DVCS analysis in the COMPASS experiment, based on the 20 data taking. The recoiled proton is detected in a time of flight detector placed around the target. All the involved particles are detected in this process: the magnetic guidance system before the target allows to measure the incident muon, while the diffracted muon and the photon are detected in the forward spectrometer and the three electromagnetic calorimeters. GPDs can be studied through the Deeply Virtual Compton Scattering process (DVCS), where the proton is probed by a virtual photon, in order to produce a real photon in the final state and a recoil of the proton remaining intact.In particular, this process is studied in the COMPASS experiment at CERN, where a polarised muon beam of 160 GeV interacts on a liquid hydrogen target. These functions provide information on the longitudinal momentum and transverse position of quarks and gluons, including their correlations. Nevertheless at energies close to the proton mass, the usual perturbative methods cannot be used, and the partons dynamics is therefore orchestrated by structure functions called Generalised Partons Distributions (GPDs). These constituents of matter are actually made out of quarks and gluons (gathered under the denomination of partons), and are governed by the laws of quantum chromodynamics (QCD). We demonstrate the feasibility using harmonic fitting to both cross sections and beam spin asymmetries with both real and pseudo-data.Īlthough protons and neutrons are known to be the main constituents of the visible matter in the universe, they still remain nowadays a conundrum in modern physics. As an example, for typical JLab lab-frame kinematics, we simplify the reduced DVCS observables to their dominant terms, providing a sufficient number of equations for local determination of the twist-2 CFFs. We investigate the exercise of locally extracting the real and imaginary parts of the four twist-2 Compton form factors (CFFs) $\$.