摘要
掌纹图像的分割是针对一幅掌纹,找出感兴趣的目标区域(ROI),使之从背景中分离出来,它是掌纹特征提取和进一步的匹配的关键步骤。传统的Otsu阈值化算法能有效地将掌纹从背景中分离,通过旋转与平移,使掌纹图像进一步精确定位与归一化,并选择纹线集中的部分实现了在线掌纹图像的分割。实验结果验证了此法的有效性。
Palmprint image segmentation is usually to find a region-of-interesting （ROI） and segment it from the background, which is the key step in pre-processing, feature extraction and automated recognition, The traditional Otsu algorithm is effective in dividing each palm from the background. In order to normalize the ROI, we rotate and translate palmprint images to reduce the nonlinear interference and distortion. The experimental results demonstrate the effectiveness and accuracy of the proposed method.
出处
《电脑知识与技术》
2006年第5期154-155,共2页
Computer Knowledge and Technology
基金
江苏省教育厅自然科学基金项目(05KJD140030)。
关键词
掌纹
OSTU算法
分割
定位
Pahnprint
Otsu algorithm
Segmentation
Orientation
