欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对印刷套准检测存在的精度低、速度慢的问题，提取了印刷标志图像的Tamura纹理特征，包括粗糙度、对比度和方向度，以描述其印刷标志套准或套不准特征。设计了支持向量机的分类器对印刷标志图像进行套准识别，并采用高斯径向基核函数用于非线性数据的分类。实验结果证明，采用建议的印刷标志图像特征提取和分类方法，识别准确率达到90％，识别时间为0．032751s，方法在识别；住确率和识别速度上都优于人工检测和文献[8]的方法。
英文摘要：
To solve the ac：euraey and speed of printing registration detection, Tamura textures of the printing mark images are ex- tracted, including coarseness, contrast and directionality, which describe the registration features of printing marks. The classifier of Support Vector Machine （SVM） is proposed to identify the registration of printing mark images, and Gaussian radial basis time-tinn is used as kernel function in SVM for nonlinear classification. Experimental results show that recognition accuracy achieves 90% , and occupied time of image recognition is 0. 032 751 s using the proposed method of feature extraction and classifieation. The proposed method is superior to manual detection and the method in the reference [ 8 ] in recognition accuracy and time.
同期刊论文项目
同项目期刊论文
Type Synthesis, Kinematic Analysis and Optimal Design of a Novel Class of Sch？nflies-Motion Parallel
Active vibration isolation of macro–micro motion stage disturbances using a floating stator platform
Study of deposition characteristics of multi-nozzle near-field electrospinning in electric field cro
期刊信息
位置：