The high pilot overhead of channel estimation (CE) poses a significant challenge that hinders the wider application of reconfigurable intelligent surfaces (RIS) in wireless systems. Two timescale CE strategy can reduce the pilot overhead effectively by leveraging the semi-stationary characteristic of the base station (BS)-RIS channels. However, this strategy is unsuitable for real-time CE due to its reliance on an iterative optimization algorithm for BS-RIS CE, which entails high computational complexity. This paper reconsiders the optimization method within the framework of two timescale CE strategy for BS-RIS channels. Firstly, after completing the received pilot data matrix, the CE equations are simplified to a second-order nonlinear rank-one optimization problem. Subsequently, the complex-valued matrix of received pilots in the gradient equations is decomposed in blocks and represented in real terms, and an optimization method based on principal eigenvalue approximation is proposed. The proposed method establishes a semi-closed-form relation between the received pilots and the channel parameters. For the scenario of Rician channel and typical antenna configurations, the simulation results show that the proposed method has lower computational complexity compared to the referenced iterative method. And it can reduce more than 85% pilot overhead when the channel coherent time of BS-RIS is 4 times of RIS-User channels. When the signal-to-noise ratio (SNR) of received pilots is less than 16 dB, the estimation accuracy surpasses that of the iterative algorithm. Consequently, the proposed CE method is more competitive in scenarios requiring high real-time CE or the scenarios that RIS is located far away from BS and close to users.