摘要
以TI公司的定点型DSP芯片TMS320VC5509A为核心,通过CMOS摄像头OV7620采集图像,设计了一种基于统计学方法与信息融合的手势图像识别系统.通过分析待识别手势的特征信息,多次提取手势特征的灰度值、手势的面积大小,来计算特征值的期望、方差,并得到广义期望、面积的均值,从而用广义期望与图像的面积信息融合来判断输入手势是否为所要识别的手势.
The paper focuses on the design of a hand gesture image recognition system ,based on the fixed-point DSP TMS320VC5509A and with images captured by digital CMOS camera OV7620 .In the design process ,besides analyses of characteristics of the objects to be recog-nized ,the gray value of those characteristics and the area size of the hand gesture image are extracted multi times .Then the generalized expectation is obtained via calculation of the mathematic expectation and variance of the values related to characteristics according to sta-tistical approach .Fusion of the generalized expectation and the mean of the hand image area provides a foundation to hand gesture recognization .
出处
《陕西科技大学学报（自然科学版）》
2014年第2期146-151,共6页
Journal of Shaanxi University of Science & Technology
关键词
图像识别
统计学方法
定点DSP
信息融合
image recognition
statistical approach
fixed-point DSP
information fusion
