摘要
提出了一种微观图像的特征识别方法。该方法基于图像微观特征的随机性和唯一性,分为微观图像的特征提取和图像识别两部分,主要包括对宏观图像的放大、微观图像的数据采集、数字图像二值化、图像特征信息的提取、数据的归一化处理、特征模板的保存和比较识别。文中介绍了各部分的步骤、特征提取的算法以及图像识别的最大误差限准则。提出的算法能有效地提取所处理图像的微观特征并加以识别。图像识别实验表明,该方法的识别率达到100%,可用于文件的真伪性辨别及防伪。
A method was developed for recognizing microscopic image features based on the randomness and uniqueness of the microscopic features of images. The method consists of feature extraction and image recognition, including magnification of the processed image, acquisition of microscopic image data, binary coding of the digitized image, extraction of image feature information, data generalization, feature template generation and storage, and image comparison and recognition. The algorithm is described with emphasis on image feature extraction and the maximumerror criterion for image recognition. Microscopic features of the processed images can be extracted and recognized effectively by the method. Experimental results show that the recognition rate can reach 100%. The method can distinguish between genuine and fake documents for anticounterfeiting operations.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2003年第8期1038-1040,共3页
Journal of Tsinghua University(Science and Technology)