摘要
在恶劣的工业现场环境中,拍摄的工件图像会失真,为此提出了在一定噪声和模糊范围内能够正确识别字符标号的方法。工件在传送过程中可能发生偏移,使得被测件与摄像镜头的相对位置不确定,所获得的图像会发生偏转或平移,为此提出了解决字符的平移,缩放及旋转问题的方法;设计了神经网络识别系统。
Images taken on blocks would distort in inferior industrial spot occasions, the essay proposed approaches to correctly recognize characters and symbols in certain noise and fuzzy scopes. Blocks might offset in transmitting process thus made the related position of tested and camera unsteady and the captured images would cause flexion and advection . Thus the author proposed approaches to solve advection, zoom and rotation with the designing of neural network recognition system.
关键词
二值化
模式识别
图像处理
神经网络
BP算法
Binarization
Mode Recognition
Image Processing
Neural Network
BP Computation
