The conventional channel selection approach in cognitive radio network is random selection of the transmission channel based on the channel characteristics while neglecting the channel quality requirement of the video streaming at the application layer. Aiming at solving this problem, a cross-layer optimized channel selection method targeting on the video streaming quality optimization was presented. Via minimizing the end-to-end video distortion, the video encoding quantization parameter at the application layer and the adaptive modulation and channel coding as well as the specific transmission channel at the physical layer were jointly selected. A large number of video transmission simulation experiments were made for the proposed algorithm over the multi-channel cognitive radio networks. The experimental results show that this kind of cross-layer optimized channel selection approach can efficiently improve the objective quality of second user video streaming more than 1.5 dB over the cross-layer optimization without channel-selection method over cognitive radio networks.