首页    期刊浏览 2024年12月04日 星期三
登录注册

文章基本信息

  • 标题:Low-Power VLSI Implementation in Image Processing using Programmable CNN
  • 本地全文:下载
  • 作者:Sarita Chauhan ; Janardan Sharma ; Pratibha Soni
  • 期刊名称:International Journal of Advanced Research in Computer Engineering & Technology (IJARCET)
  • 印刷版ISSN:2278-1323
  • 出版年度:2014
  • 卷号:3
  • 期号:5
  • 页码:1818-1824
  • 出版社:Shri Pannalal Research Institute of Technolgy
  • 摘要:The low power CMOS implementation is based on a combination of MOS transistors operating in di erent modes: weak and strong- inversion. We propose, MOS transistors operating in the lateral bipolar mode.This combination has enabled a VLSI implementation of a simpli ed version of the original CNN model with the main characteristics of low-power consumption, program-mability, and embedded photo sensors to process images directly projected on the chip. For VLSI implementations of CNN's .It is usual to consider simpli ed versions of the Chua- Yang model in order to reduce circuit com-plexity.Several applications of CNN's in solving image processing tasks such as noise removal, edge and corner detection, hole lling, connected component detection, etc. To obtain low-power consumption and to directly process the small currents from the sensors we use MOS transistors operating in weak inversion for all the di eren- tial pairs .in such a way that only local matching between transistors belonging to each di erential pair is required. Due to their better matching properties, MOS transistors operated in the lateral bipolar mode are used for the distribution of the values de ning the programmable parameters (templates A and B, and the o set I), as well as the values that limit the range of the cell output and input. As in other CNN implementations integrated photo sensors are used to allow a parallel and direct input of the images to be processed by the network. The approach considered in and uses a Darlington phototran-sistor in order to provide the appropriate current levels required by the MOS circuitry operating in strong inver-sion. Software proposed to be used is MATlab and VHDL.
  • 关键词:VHDL; MATLAB; Verilog
国家哲学社会科学文献中心版权所有