Wireless video sensor networks (WVSNs) feature the characteristics of easy deployment and low maintenance cost and therefore are suitable for a wide range of important applications, including remote surveillance, health care delivery and traffic control systems. Compared to traditional communication systems, Sensor nodes in WVSN operate under resource constrains such as power supply and transmission bandwidth. The main object of this paper is to study the resource allocation of a wireless sensor node and improve the system performance, especial the video distortion. More specifically, by taking into account the relations between separated modules on a sensor node, we build a comprehensive power-rate-distortion (PRD) framework for WVSN optimization. Based on this framework, we address the problem of power allocation between video encoder module and wireless transmission module on the node to achieve the optimal reconstructed video quality under power and rate constraints.