This thesis discusses the performance of a solar-powered wireless visual sensor network and its visual applications. We examine the performance of a layered clustering model in sparing communication energy consumption and prolonging the system lifetime. The experimental result illustrates that the system can transmit the same amount of video packets with less energy consumption when video quality is at achievable minimum distortion rate. Therefore, the visual sensor network may achieve higher performance by applying rechargeable solar cell and layered clustering. After receiving all the video data, the sink may be applied with advanced post-processing techniques. We propose an innovative post-processing algorithm, Parallel Self-Organizing Tree Map (PSOTM) that can be implemented in the sink. By means of processing visual data in parallel, PSOTM may achieve faster image segmentation with insignificant impacts on the visual quality.