摘要
重点研究实现掌纹方位无关性匹配的掌纹图像预处理算法.由掌纹采集设备获取掌纹图像作为输入数据.首先,对掌纹图像应用中值滤波器去除系统噪声,并应用阈值分割法分割背景区和目标区.然后,利用内部边界跟踪算法找到手掌边界,计算边界像素与手掌的底端中部Pm之间的欧氏距离.根据欧式距离曲线图中的两个局部最小值,自动确定两个指根部位置.最后,基于第1和第3个指根部建立参照坐标系,自动定位掌纹感兴趣(ROI)区域.对两个公用的掌纹图像库进行实验,实验结果验证了该方法的有效性.
This paper emphasis on the research of palmprint image preprocessing to match without the influence of orientation. Palmprint images captured by palmprint collection device are used as input data. First, traditional median filter is utilized to reduce system noises effectively, and threshold segment method is applied to palmprint images for image segment. Then, the inner border tracing algorithm is employed to find the palm border, and the distance between border pixels and the middle point Pm of the intersection line that is formed by the wrist and bottom margin of a palmprint image. Two local minimums are found in the distance distribution diagram. We can find two finger-web locations by these local minimums. Finally, the coordinate system can be founded based on the first and the third finger-web location. Palmprint center parts can be orientated automatically. Experiments are carried out on two common image databases. The results verify the effectiveness of the method.
出处
《沈阳化工学院学报》
2007年第4期300-302,309,共4页
Journal of Shenyang Institute of Chemical Technolgy
关键词
掌纹图像
二值化
ROI区域的定位
图像预处理
palmprint images
binarization
region of interest location
image preprocessing
