The critical distinction between the emerging intelligent transparent surface (ITS) and intelligent reflection surface (IRS) is that the incident signals can penetrate the ITS instead of being reflected, which enables the ITS to combat the severe signal penetration loss for high-speed railway (HSR) wireless communications. This paper thus investigates the channel estimation problem for an ITS-assisted HSR network where the ITS is embedded into the carriage window. We first formulate the channels as functions of physical parameters, and thus transform the channel estimation into a parameter recovery problem. Next, we design the first two pilot blocks within each frame and develop a serial low-complexity channel estimation algorithm. Specifically, the channel estimates are initially obtained, and each estimate is further expressed as the sum of its perfectly known value and the estimation error. By leveraging the relationship between channels for the two pilot blocks, we recover the Doppler shifts from the channel estimates, based on which we can further acquire other channel parameters. Moreover, the Cramer-Rao lower bound (CRLB) for each parameter is derived as a performance benchmark. Finally, we provide numerical results to establish the effectiveness of our proposed estimators.