In the visual sensor network(WVSN), reducing communication energy consumption is a crucial problem due to the high data volume to transmit. In this paper, we present intelligent communication methods which are optimal buffer sizing and multi-radio approach supporting IEEE 802.11g(WiFi) and IEEE 802.15.4 (ZigBee) so as to maximize the lifetime of the battery-constrained WVSN nodes while satisfying the given user specific delay tolerance. To optimize the energy consumption we find the optimal buffer size vs. data volumes and suggest a run-time radio-selection algorithm. Simulation results show that the lifetime of the proposed WVSN node is extended by 207.16% ~ 252.52 compared to the WVSN node with single radio in ideal data rate, and extended by 212.36% ~ 290.01% compared to the WVSN node with single radio in effective data rate. Compared to method [13], our proposed method prolongs the system lifetime from 111.16% to 138.17% in ideal data rate, and from 56.15% to 65.99% in effective data rate.