自适应的OCT内外部指纹提取算法

针对不同个体手指表皮层与真皮层之间的深度差异大,现有算法从光学相干断层扫描(optical coherence tomography,OCT)图像中提取内外部指纹不具备普适性的问题,提出一种基于结构点聚类的自适应OCT内外部指纹提取算法.首先改进聚类的相似度判断与终止条件的计算方法,自适应分离角质层和乳头层的结构点;然后定义了位于汗腺以及真皮层内部结构上其他组织的伪结构点,提出基于距离度量的由远到近两步法去除伪结构点,并经过插值拟合得到精确的乳头层轮廓;最后通过序列拼接与图像增强获得OCT内外部指纹.与不同内部指纹提取算法进行对比实验的结果表明,该算法能够自适应不同个体的乳头层深度,以更高的普适性和鲁棒性实现从OCT三维数据中同时提取具有可识别性的、大面积和高分辨率的内外部指纹.

Adaptive OCT Internal and External Fingerprint Extraction

The dermal depth of different individual figures varies, which reduces the application universality of many existing methods that extract the internal and external fingerprints from the optical coherence tomography (OCT) images based on dermal depth. This paper proposes an adaptive OCT internal and external fingerprint extraction method based on structure point clustering. Firstly, an adaptive clustering method was proposed with improved similarity calculation and clustering termination condition, to adaptively separate the structural points of the stratum corneum and the papillary layer. Secondly, a two-step far to near pseudo- structure point removal algorithm was proposed to remove the pseudo-structure points caused by the sweat glands and other internal tissues, and thereafter the precise contours of the stratum corneum and papillary layer were obtained by interpolation fitting. Finally, OCT internal and external fingerprints were obtained after fingerprint stitching and enhancement. The experimental results show that the proposed method can adapt to the dermal depth of different individuals, and achieve simultaneous extraction of recognizable, large-area and high-resolution internal and external fingerprints from OCT three-dimensional data with high universality and robustness.