Abstract: This paper focuses on the region of interest (ROI) extraction of online palmprint images which involve transition and rotation. Firstly, a threshold is used to make the palmprint image become a BW image. Then the contour line of the BW image is smoothed by morphology algorithmic rule. We extract Freeman chain code from the BW image and translate it into a function of angle. The key points are extracted by comparing the angle changes of the contour line of every neighborhood point. Finally the palmprint images are aligned according to the key points, and thus the region of interest of the palmprint image can be extracted easily. This method provides a necessary preprocessing for further feature extraction and matching. The experimental results on two palmprint image databases demonstrate the effectiveness of the proposed method.

Key words: palmprint image, key point, contour line, morphology, region of interest (ROI)

摘要： 论文重点研究具有一定自由度在线掌纹图象的感兴趣区域提取算法。首先结合掌纹图象的特点采用全局阈值二值化掌纹图象，然后利用形态学算子平滑掌纹轮廓，提取轮廓线Freeman链码并对链码进行角度变换，最后通过考察轮廓线上各点附近轮廓线的角度变化来提取掌纹图象感兴趣所需要的定位点，从而提取感兴趣区域。感兴趣区域的提取为特征提取和特征匹配打下了基础。最后通过实验在两个公开的掌纹数据库证明了这种算法的有效性。

关键词: 掌纹图象, 定位点, 轮廓线, 形态学, 感兴趣区域