NU

Field Programmable Gate A rrays (FPGAs) have been re-cently use d as an ee ctive platform for implementing many im-age/signal pr ocessing applic ations.MATLAB is one of the most popular languages to model image/signal processing applic ations. We present the MATCH compiler that takes MATLAB as input and pr oduces a hardware in RTL VHDL, which can be mapp ed to an FPGA using commercial CAD tools. This dramatic ally reduces the time to implement an application on an FPGA. We present results on some image and signal processing algorithms for which har dwar ewas synthesize d using our c ompiler for the Xilinx XC4028 FPGA with an external memory. We also present comp arisonswith manually designed hardwar es for the applica-tions. Our results indic ate that FPGA hardware can be generated automatically reducing the design time from days to minutes, with the tr adeo that the automatically generated hardware is 5 times slower than the manually designed hardwar e. 1