A programmable image signal processing architecture for embedded vision systems

S. McBader1, P. Lee2
1NeuriCam S.p.A, Trento, Italy
2University of Kent at Canterbury, Canterbury, Kent, UK

Tóm tắt

This paper presents a programmable multiprocessor architecture suitable for image preprocessing in embedded vision systems. The architecture is made up of sixteen 16-bit input/32-bit output parallel processing elements, connected to an intelligent DMA channel and frame buffers. Each element of the SIMD processing array can be programmed using a RISC-like instruction set, and is capable of performing DSP operations commonly used in image preprocessing algorithms. The architecture has been prototyped on an FPGA connected to a 256/spl times/256-pixel CMOS sensor, achieving 3.23 GOPS and up to 667 FPS of throughput at a clock frequency of 50 MHz. It is estimated to achieve a minimum of 6.46 GOPS and 1334 FPS at a clock frequency of 100 MHz when implemented in VLSI.

Từ khóa

#Signal processing #Machine vision #Intelligent sensors #Clocks #Frequency estimation #Parallel processing #Digital signal processing #Signal processing algorithms #Prototypes #Field programmable gate arrays

Tài liệu tham khảo

10.1109/4.364438 0, NeuriCarn, EtherCam Intelligent Camera Briefing 10.1109/ISSCC.1994.344649 10.1109/4.881219 1997, Xilinx, Gate count capacity metrics for FPGAs jung, 1998, a low-power and high-performance cmos fingerprint sensing and encoding architecture, the 24th European Solid-State Circuits Conference, 324 ramacher, 2001, A 53-GOPS Programmable Vision Processor For Processing, Coding-Decoding And Synthesizing of Images, Proc European Solid-State Circ Conf, 160 athanas, 0, Addressing the Computational Requirement of Image Processing with a Custom Computing Machine 0, Mattox Imaging, Genesisplus Product Briefing 10.1109/ISSCC.1991.689148 paillet, 2000, Design Solutions and Techniques for Vision System on a Chip and Fine-grain Parallelism Circuit Integration, Workshop at the IEEE ASIC / System On Chip Conf