Reconfigurable Intelligent Surface(RIS)-assisted communication systems can employ phase shifters to focus radiation energy at the receiver, thereby enhancing link quality and extending coverage. However, designing phase shifters requires cascaded channel state information (CSI), which poses a challenge due to the absence of A/D and D/A converters at RIS nodes. In RIS-assisted orthogonal frequency division multiplexing (OFDM) systems, the design becomes particularly intricate since all subcarriers of an RIS element can only share the same phase shifter. Consequently, the estimation process relies on numerous pilots to provide sufficient degrees of freedom for CSI estimation, significantly increasing the pilot overhead. In this paper, we investigate several low-complexity and low pilot-overhead channel estimation methods for RIS-assisted single-input multiple-output (SIMO) OFDM systems based on compressive sensing techniques. Our study commences by developing a low-complexity orthogonal matching pursuit (OMP) estimator, which jointly estimates sparse time-domain channel impulse responses (CIRs) and their associated sparse angular domains. The OMP-based estimator exhibits very low complexity but is sensitive to the structure of the sensing matrix. Therefore, we introduce two low-complexity sparse Bayesian learning (SBL) channel estimators to enhance robustness to the sensing matrix structure. The first estimator employs approximate message passing (AMP) with unitary transformation (SBL-UTAMP), while the second estimator uses orthogonal AMP (SBL-OAMP). Simulation results indicate that our proposed methods achieve a promising balance between CSI estimation performance and complexity in RIS-assisted SIMO-OFDM systems.