一种改进的Mean Shift指纹图像分割算法

摘要

After analysis and research on the traditional fingerprint image segmentation algorithm, an improved Mean Shift algorithm for fingerprint image segmentation was introduced. The algorithm makes fully use of the inherent directional characteristics of fingerprint image to abstract each area of fingerprint image after segmenting into a sample point and to consider the grayscale mean value of the pixels within the fingerprint region as the mean vector, thus achieving the fingerprint image segmentation effectively. Experimental results show that the background area and fuzzy algorithm can accurately be separated from fingerprint image with the algorithm. The algorithm improves the accuracy of segmentation and has good accuracy for most of fingerprint images.%基于传统指纹图像分割算法,提出一种改进的Mean Shift指纹图像分割算法.该算法利用指纹图像固有的方向性特性,把经过分割后的每个指纹图像区域抽象为一个样本点,将区域内像素点的灰度均值作为均值向量,从而有效地实现了指纹图像分割.实验结果表明,该算法能准确地将指纹图像中的模糊区域和背景区域分离,提高了指纹图像分割的精确度,并且对于多数指纹图像准确性较好.