A novel Kohonen SOM-based image compression architecture suitable for moderate density FPGAs

摘要

Kohonen self-organizing map (K-SOM) has proved to be suitable for lossy compression of digital images. The main drawback of this technique is that it is a very computationally intensive task if software implementation is performed alone. Fortunately, the structure is fairly easy to convert into parallel processing units. However, it consumes much of a microchip's internal resources and results in utilizing more than a single microchip to realize the structure in pure hardware. Previously proposed K-SOM realizations were mainly targeted on implementing on an ASIC (Application Specific Integrated Circuit) with low restriction on resource utilization. In this paper, we, in contrast, propose a novel architecture of K-SOM which compromises between image quality, frame rate throughput and FPGA's resource utilization. The architecture is based on unsigned integer arithmetic in all operations and has been successfully synthesized on a single moderate resource FPGA with acceptable image quality and frame rate.