Very Long Baseline Interferometry (VLBI) experiments are organized within the geodetic, astrometric and astronomical communities for different applications, requiring different observation strategies adopted in scheduling. Currently, the next-generation geodetic and astrometric VLBI Global Observing System (VGOS) is being established. Over the last years, evidence was presented that the delays introduced by the angular structure of the geodetic radio sources contribute significantly to the VGOS observable error budget. Consequently, correcting these structure delays through imaging will play an important part in the future, requiring a different scheduling approach. Within this work, a new source-centric VLBI scheduling approach is presented for improved imaging capabilities of geodetic observations. The algorithm is tested for a seven- and nine-station network and compared with classical geodetic schedules. Monte Carlo simulations are utilized to determine the expected geodetic and astrometric parameter precision, and two independent processing pipelines are used to assess the potential for astronomical source imaging. Based on the simulation results, it is revealed that with the new scheduling approach twice as many sources can be properly imaged. Furthermore, the precision of the Earth orientation parameter estimates is improved on average by 15 % , while the source position coordinate estimates are on average improved by 50 % . Tests with two VGOS networks of twelve and 29 antennas further reveal that the scheduling approach is also applicable to future VGOS networks.