摘要
将手势识别算法在DSP嵌入式系统上实现,并将算法进行改进使其能适合嵌入式系统的特点,采用改进的边缘像素直方图的方法实现手势识别,最终用Bhattacharyya距离匹配模版,识别手势含义。该方法在嵌入式系统上实现,有实际应用的价值,并对前景的平移和尺寸变化不敏感。算法简单快速,稳健性好,有应用前景。
In this paper, gesture recognition is implemented on embedded system, and the algorithm to fit the embedded system is improved. Orientation histogram of edge pixels is used to feature extraction and Bhattacharyya distance is used for matching. This method has practical application value, and it has robustness to gesture translation. This method is fast and simple, and it has application prospects.
出处
《电视技术》
北大核心
2012年第15期133-137,共5页
Video Engineering
基金
上海大学创新基金项目(A10-0107-09-902)
上海大学大学生创新活动计划(CXXJ-11-027)
关键词
手势识别
嵌入式
方向直方图
gesture recognition
embedded
orientation histogram
