This paper discusses an innovative real-time oriented image compression system, based on a simple algorithm designed explicitly to be implemented on a low-cost SIMD computer architecture featuring a much lower power consumption than traditional DSPs or dedicated hardware. For this reason the considered approach is suitable to be integrated on portable systems, where power consumption is a critical design issue. The algorithm, based on a hierarchical decomposition of the input image, has been tested on a special purpose SIMD system, PAPRICA, exploiting its features such as its massive parallelism and its capability to operate on local data as well as to handle hierarchical data structures. According to the proposed approach, the quality of the decompressed image can be traded for a lower power consumption as well as a higher processing speed. A comparison between the discussed algorithm and the standard JPEG compression technique is also presented.