摘要
印鉴识别过程由图像预处理、印鉴的提取与修复、印鉴的配准和识别几部分组成。针对印鉴图像的特点,通过在HSI颜色空间中提取红色分量的方法,将印鉴提取出来,进行二值化处理,采用了中值滤波和形态学滤波相结合的预处理方法,通过计算印鉴图像的SIFT特征点进行图像的配准。在印鉴识别部分,通过比较印鉴图像的不变矩相关性、环投影相关性、差图像中点状、线状成分的多少以及特征参数来判断印鉴图像的真伪。在仍然不能确定印鉴真伪的情况下,采用模拟人工识别的手段进行最终的识别。实验表明,该印鉴识别流程能够很好地识别印鉴的真伪。
The seal identification process contains the image preprocessing, extraction and repair of seal, the registration and identification of seal. In view of the characteristics of seal image, we extract the seal by dividing the red component in the HSI color space, preprocess the seal image by using a combination of median filtering and morphological filtering methods, and then proceed to the binarization processing. The seal image registration is accomplished by calculating the SIFT features of seal image. In the stage of seal image recognition, the similarity between the model seal and the sample seal is calculated by comparing their invariant moment rel- evance, ring projection relevance, difference image composition and feature parameters. If the similarity is not credible, artificial simu- lation recognition is proposed. The result of the experiments shows that the methods we proposed can recognize the authenticity of seal very well.
出处
《无线电工程》
2012年第10期51-54,共4页
Radio Engineering
关键词
图像预处理
印鉴提取
印鉴配准
印鉴自动识别
人工模拟识别
image preprocessing
seal extraction
seal registration
automatic ~aal identification
artificial simulation recognition
