An optimizing design idea for optical remote sensing imaging systems based on restoration processing was proposed. The main imaging-chain components effecting restoration processing performance were explored by theoretical analyzed, semi-physical simulation experiments and orthogonal experiments. According to deducing a mathematical model of restoration problems, it points out in theory that the main factors impacting the restoration performance are the imaging system Modulation Transfer Function space-variant (MTFSV), data compression ratio (CR) and nonuniformity correction error (NCR). Then, the restoration distortions that effect the interpretation capability of images owing to the above factors were analyzed by using a semi-physical imaging platform and simulation experiments. Finally, the relationship between restoration performance and the above metrics was established by using combined mathematics and an orthogonal experiment for the restored images with multilevel MTFSV, CR and NCR. Experimental results indicate that the restoration performs well in a quite low MTF value at Nyquist frequency (only 0.07), the MTFSV is less than 10%, CR less than 4∶1 and the NCR is no more than 3%, which not only implements the MTF area more than 70% but also keeps Signal to Noise Ratio(SNR) invariant and achieves available image quality for the interpretation as well.