【作者】 张淑雅； 赵晓宇； 赵一鸣； 李均利；

【Author】 ZHANG Shu-ya ZHAO Xiao-yu ZHAO Yi-ming LI Jun-li (Institute of DSP and Software Techniques,Ningbo University,Ningbo 315211,China)

【机构】 宁波大学数字技术与应用软件研究所；

【摘要】 本文主要研究如何区分人类视觉中的黑白老照片和彩色照片,并提出了一种对两者分类的方法。首先根据它们各自的特点,定义了一系列指标,对图像进行预分类,找出其中特征明显的图像,其次将这些指标作为 SVM 学习参数来区分剩余图像。SVM 是近年来在统计学习理论的基础上发展起来的一种新的模式识别方法,能够解决线性及非线性分类问题,对样本数量和维数不敏感。实验结果表明指标定义是合理的,效果较满意。更多还原

【Abstract】 This paper mainly studies how to distinguish old black and white photographs and chromophotographs that in human vision,and presents an algorithm to classify them.First,we define some indexes according to the different characteristics between them,and fred out images with obvious characteristics by advance classification,then,we classify other images by SVM,and the indexes above are used as learning parameters.SVM is a new method of pattern recognition developed on the base of statistical theory in recent years.It can solve the problems of linear and nonlinear classification,and is insensitive to dimensions and numbers of training set.The experimental results show that the definition of indexes is reasonable and the effect is satisfying.更多还原