| 一种新型2-D模板卷积结构的FPGA实现 |
| |
| 引用本文： | 朱磊,张贵清,张天序. 一种新型2-D模板卷积结构的FPGA实现[J]. 微计算机信息, 2007, 23(8): 193-195 |
| |
| 作者姓名： | 朱磊 张贵清 张天序 |
| |
| 作者单位： | 430074,武汉,华中科技大学图像信息处理与智能控制教育部重点实验室,华中科技大学图像识别与人工智能研究所 |
| |
| 摘 要： | 在实时图像处理过程中,2-D模板卷积是一种重要的操作,FPGA以其特有的全并行处理机制使得这种操作在硬件设计中得以较好的实现。本文提出了一种基于FPGA的模板卷积运算的新方案,相比传统方案,这种方案在结构上能以较少的硬件资源达到相同的流水深度。为了满足系统对实时性的要求,针对某些特定类型的模板,给出了一种简化的卷积器结构,同时,介绍了一种模板归一化除法运算的硬件实现方法。
|
| 关 键 词： | 2-D模板卷积 实时图像处理 |
| 文章编号： | 1008-0570（2007）03-2-0193-03 |
| 修稿时间： | 2007-01-23 |
FPGA Implementation of a New Architecture for 2-D Mask Convolution |
| |
| ZHU LEI,ZHANG GUIQING,ZHANG TIANXU. FPGA Implementation of a New Architecture for 2-D Mask Convolution[J]. Control & Automation, 2007, 23(8): 193-195 |
| |
| Authors: | ZHU LEI ZHANG GUIQING ZHANG TIANXU |
| |
| Affiliation: | State Education Commission Key Laboratory for Image Processing and Intelligent Control,Institute of Pattern Recognition and Artificial lntelligence,Huazhong University of Science and Technology, Wuhan 430074, China. |
| |
| Abstract: | 2-D mask convolution is one of the essential operations in real-time image processing, FPGA make this operations achieve high performance with its unique parallel mechanism. A new method for calculating 2-D mask convolution based on FPGA is presented. Compared with traditional methods, it achieves the same depth of pipelining with fewer hardware resources. Also, in order to meet the real-time requirement of image processing system, a simplified structure of convolver for special mask was given. A method to implement the division operation for normalization of masks with hardware was also introduced. |
| |
| Keywords: | FPGA |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
