【摘要】 以视觉注意机理为依据来选取指纹图像质量评价的特征,使特征集能够全面覆盖人对图像进行评价的各个方面。同时,提出了基于极坐标中心敏感特性的细节点可靠性评价方法和基于Otsu算法的灰度对比度评价方法。根据这些方法提取的特征能够更准确地描述图像的质量状态。试验结果表明:本文的特征集在BP神经网络和SVM两种分类器上都获得了很高的分类准确率。更多还原

【Abstract】 Based on the mechanism of human visual attention,five features are chosen to simulate the process of subjective quality assessment of fingerprint images.Among these features,two of them are extracted by two proposed methods,a minutiae reliability assessment based on Polar Coordinates Centrum Sensitivity(PCCS)and a gray-scale image contrast based on Otsu algorithm.Experimental results show that this feature set on both Error Back Propagation(BP)Neural Network and Support Vector Machine(SVM)gives high classification accuracy.更多还原