基于小波分析和SVM的印鉴图像识别

Automatic Seal Identification Based on Multiscale Wavelet Analysis and SVM

针对印鉴图像自动识别,提出了一种基于小波大尺度分解特征提取和支撑向量机的印鉴鉴别方法。印鉴图像预处理之后,用一维环投影向量的循环移位来代替二维图像的旋转,实现印鉴的配准。同时,利用小波分解结果,采用小波分解的大尺度向量作为特征向量,排除高频噪音的干扰,并以此进行SVM训练及测试,实现图像鉴别。应用实例表明,该算法具有很强的抗干扰性、识别率高。

An automatic Chinese seal identification method which is based on Support Vector Machine （SVM） and large-scale wavelet decomposition is proposed.After preprocessing the seal image,the projecting vectors are extracted by dividing the seal imprint into rings.The algorithm uses the cyclic shift of one-dimensional ring projecting vectors instead of rotation of two-dimensional images to register the seals efficiently and accurately.The large-scale wavelet analysis results are used as the intrinsic features,then training and testing of SVM are processed.The experimental recognition results show that the proposed approach is excellent in performance,reliability and adaptability,and it is feasible for practical applications.