龙希田, 石匆, 杨杰, 吴南健. 一种基于多级并行处理器的高速实时手势识别及指尖轨迹追踪系统[J]. 微电子学与计算机, 2013, 30(12): 90-96.
引用本文: 龙希田, 石匆, 杨杰, 吴南健. 一种基于多级并行处理器的高速实时手势识别及指尖轨迹追踪系统[J]. 微电子学与计算机, 2013, 30(12): 90-96.
LONG Xi-tian, SHI Cong, YANG Jie, WU Nan-jian. A Hand Gesture Recognition and Fingertip Tracking System Based on Multiple Levels of Parallel Processors[J]. Microelectronics & Computer, 2013, 30(12): 90-96.
Citation: LONG Xi-tian, SHI Cong, YANG Jie, WU Nan-jian. A Hand Gesture Recognition and Fingertip Tracking System Based on Multiple Levels of Parallel Processors[J]. Microelectronics & Computer, 2013, 30(12): 90-96.

一种基于多级并行处理器的高速实时手势识别及指尖轨迹追踪系统

A Hand Gesture Recognition and Fingertip Tracking System Based on Multiple Levels of Parallel Processors

  • 摘要: 提出一种基于多级并行处理器的实时手势识别及指尖轨迹追踪系统.该系统包含64×64处理单元(PE)阵列,64个简化精简指令集(RISC)构架的协处理器组,以及一个RISC微处理器.PE阵列实现对原始图像数据的噪声消除、平滑滤波、形态学处理等预处理,协处理器组用于完成对预处理后图像的特征统计、提取,RISC中央处理器则用于对整个系统进行控制,完成复杂的数学运算,以及与系统外界进行交互的工作.该系统使用 FPGA 实现.由于采用多级并行处理器构架,使系统的运算处理能力大大提高.试验结果表明,本系统具有良好的鲁棒性,并且满足实时手势识别、指尖轨迹追踪的要求.

     

    Abstract: This paper proposes hand gesture recognition and fingertip tracking system based on multiple levels of processors.The system contains a 64×64 Processing Elements (PE) array,a coprocessor group consists of 64 coprocessors and a RISC CPU. The PE array can carry out noise cancelation, smooth filtering, and image morphology.The coprocessor group is responsible for feature statistics and feature extraction.The RISC CPU controls the whole system and is capable of conducting complex arithmetic operations.It can also perform data exchange with other systems.The proposed system is implemented in FPGA.Due to its multiple level processors architecture the processing rate is significantly improved.Experiments demonstrated that our proposed system has good robustness and fully satisfy the real-time hand gesture recognition and finger-tip tracking requirement.

     

/

返回文章
返回