【作者】 荣海泓；

【导师】 席志红；

【作者基本信息】 哈尔滨工程大学 ， 信号与信息处理， 2004， 硕士

【摘要】 本文以掌纹为研究对象，通过对掌纹的线特征提取与选择方法的研究，对复杂背景、低对比度、高噪声环境下的特征提取及目标分割等图象识别技术进行了深入的探讨。 掌纹与指纹一样也具有稳定性和唯一性，利用掌纹的线特征、点特征、纹理特征、几何特征等完全可以确定一个人的身份。其中掌纹的线特征可以在较低空间分辩率的图象中获取，能克服指纹的一些不足，是掌纹自动识别的最主要的特征。由于在线掌纹图象的对比度较低，掌纹中的细小皱纹和乳突纹会形成一定的噪声，掌纹中的线特征的模式十分复杂，因此对掌纹进行线特征提取是一项具有挑战性的研究课题。 本文在形态金字塔与形态小被的理论基础上，提出了切实可行的、适用于掌纹分析的方法与思路。主要研究内容如下： [1]、对掌纹定位与归一化算法进行了改进。对掌纹进行定位的目的就是从掌纹中提取出合适的基准点，建立参照坐标系，以减少采样过程中引进的旋转、平移、扭曲等非线性因素的影响，提高匹配识别算法的鲁捧性；本章采用灰度形态梯度提取掌纹的轮廓线、用角点检测方法提取图象的关键特征点，根据关键点建立参照坐标系，实现对掌纹的归一化分割。 [2]、脱机掌纹中的线特征提取实质上是一个复杂噪声背景下的边缘检测问题。本文提出一种基于形态中值金字塔和LAW模板的脱机掌纹线特征提取方法。该方法借鉴非线性多分辨率分析的思想，能在去除噪声的同时很好地保留掌纹的线特征。 [3]、在线掌纹的线特征稳定、表示方法简单，特征空间小，可实现快速检索与匹配。对在线掌纹进行实时分类与识别是决定掌纹鉴别技术能否实用化的关键。本文提出的基于形态小波与空间方向选择模板的在线掌纹的线特征提取方法，可有效地提取出在线掌纹中的线特征。 有关掌纹线特征提取、选择与基于线特征的匹配方法研究对图象处理和模式识别等都具有重要意义，对掌纹图象的深入研究将推动掌纹识别技术走向实用化产业化。总之，对掌纹图象的分析研究是一项具有重要理论价值和应用背景的课题。更多还原

【Abstract】 The main purpose of this dissertation is to discuss palmprint. By studying the line extraction and selection method from palmprint, this dissertation analyzes the techniques of feature extraction , image segmentation with low contrast and heavy noisy background .Since palmprint has the stablity and uniqueness as fingerprint, the personal authentication can be determined completely by the line, geometry, minute.and texture features. Line features obtained from the lower resolution images has manv advantages compared with fingerprint and have become the most important character for palmprint identification. Line patterns in palmprint are very complicated because of the low contrast and the heavy noise caused by fine wrinkles and ridges, so it is a challenging task to complete the feature extraction. On the basis of morphological pyramind and morphological wavelet, a series of feasible and effective approaches to palmprint analysis have been developed .this dissertation mainly studies the approaches for palmprint identification in the following aspects: [1] Alignment and normalization online palmprint images.The contour lines and the key points for the alignment are extracted out by gray morphological grads and by coner detection method respectively. By these keypoints, we can establish a coordinate for the normalization. [2] Extraction line features from offline palmprintA line extraction algorithm based on morphological median pyramid and Law’s template is proposed for offline palmprint. By this nonlinear multi-resolution analysis, line features can be effectively extracted out and noise can be effectively restrained. [3] Extraction line features from online palmprint.Stable, simple line features of online palmprint are the key issues for popularization the palmprint identification technology. A line extraction method based on morphological wavelet and spatial directionalselection template is proposed for online palmprint, by which lines can beeffectively extracted out.Feature extraction is a important part of Palmprint identification. The methods used for palmprint analysis can be easily extended to other recognition fields. Palmprint analysis is an interesting and important research field with theoretic and practical meaning. More attention has been paid on the key techniques such as extraction line features , Selection line features and classiflcationl ,which are very important to image analysis and pattern recognttion and helpful to generalize this technology into practice.更多还原