The geomagnetic field is composed of a variety of sources that act on a wide range of timescales and amplitudes. The separation of magnetic storm effects from quiet variations is needed to accurately quantify impacts of space weather events. The extraction of such quiet contributions within geomagnetic measurements is achieved by the determination of baselines, which, ideally, is done by a simple algorithm which captures quiet sources suitably well, while being applicable to an extensive network of magnetic observatories independent of the period of time. In this work, we apply signal filtering techniques on the horizontal components of geomagnetic field measurements from low- and mid-latitude observatories to determine baselines. The variations within the baseline are investigated for magnetically quiet periods between 1991 and 2019, focusing on long-term trends, seasonal and local time dependencies, and day-to-day variability. The analysis confirms that the contributing quiet sources include the secular variation and the solar quiet (Sq) current system. The non-negligible day-to-day variability, that is typical for Sq in low- and mid-latitudes, is embedded within the baseline. Thus, the filter approach extracts quiet magnetic field variations well. Comparisons with other baseline methods show good agreements. We conclude that the filter approach can be used to determine baselines automatically during magnetically quiet periods without the need of further apriori information and is applicable on a wide network of magnetic observatories. It marks the first step for deriving magnetic indices for (near) real-time space weather applications.