In order to realize the vision of intelligent connected vehicles, it is necessary to model the vehicle-to-vehicle (V2V) channels in various realistic environments, especially when the line-of-sight (LOS) between transmitter (Tx) and receiver (Rx) is obstructed. In this paper, we model obstructed vehicle-to-vehicle (V2V) channels for the 5-GHz band through measurement-validated ray-tracing (RT) simulations. To begin, we establish a realistic V2V RT simulator through integrating three key channel features: small-scale structures (e.g. lampposts, traffic signs), handled by their approximate radar cross sections; large-scale structures (such as buildings and ground), calibrating their electromagnetic and scattering parameters; and obstructing vehicle effects via V2V channel measurements. Then, based on extensive RT simulations, the target channels are characterized comprehensively, in terms of path loss, shadow fading, root-mean-square delay spread, Rician $K$-factor, azimuth/elevation angular spread of arrival/departure, cross-polarization ratio, and their cross-correlations. All the parameters are input into and verified by the 3GPP-like quasi-deterministic radio channel generator (QuaDRiGa). By adding the obstructed V2V scenario into standard channel model families, this paper provides a foundation for evaluating intelligent vehicular communications in challenging conditions.