| Adaboost人脸检测算法在嵌入式平台上的优化与实现 |
| |
| 引用本文： | 徐淑峰,孙炜,赵峰,张忠.Adaboost人脸检测算法在嵌入式平台上的优化与实现[J].信息技术,2008,32(6). |
| |
| 作者姓名： | 徐淑峰 孙炜 赵峰 张忠 |
| |
| 作者单位： | 上海交通大学微电子学院,上海,200240 |
| |
| 摘 要： | 由于嵌入式平台的计算能力弱、内存空间小,为了能够将高复杂度的Adaboost人脸检测算法移植到嵌入式平台并能够高速运行,主要在算法方面进行了优化,其中包括:级联分类器的训练方法优化、浮点定点化、检测窗口SCALE处理等.优化后的级联分类器体积减小为未优化时的1/10,在基于PowerPC处理器的验证平台上,取得了15fps的检测速度.
|
| 关 键 词： | Adaboost 级联分类器 FFC PowerPC405 Adaboost 人脸 检测算法 嵌入式平台 方法优化 platform embedded detection algorithm face implementation 检测速度 验证平台 处理器 PowerPC 体积减小 SCALE 窗口 定点化 浮点 训练 |
Optimization and implementation of Adaboost face detection algorithm on embedded platform |
| |
| XU Shu-feng,SUN Wei,ZHAO Feng,ZHANG Zhong.Optimization and implementation of Adaboost face detection algorithm on embedded platform[J].Information Technology,2008,32(6). |
| |
| Authors: | XU Shu-feng SUN Wei ZHAO Feng ZHANG Zhong |
| |
| Affiliation: | XU Shu-feng,SUN Wei,ZHAO Feng,ZHANG Zhong(School of Micro Electronics,Shanghai Jiaotong University,Shanghai 200240,China) |
| |
| Abstract: | In order to port the high-complexity Adaboost algorithm to embedded platform with small memory capacity and low computation.This paper mainly develops some methods of algorithm optimization for real-time face detection such as optimization of training cascade,FFC and window scaling.The experimental results show that the size of training cascade optimized is 90% smaller than before.And the proposed method can detect faces at a frame rate of 15 fps on PowerPC platform. |
| |
| Keywords: | Adaboost cascade FFC powerPC 405 |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
