【摘要】 将图像预处理技术应用于PCB(印制电路板)图件识别研究。为改善PCB图件图像质量,提高后续BP神经网络对图件图像的识别率,采用MATLAB语言对原始图像进行了预处理,主要包括基于直方图均衡化和中值滤波法的二值图像去噪与平滑操作、基于四叉树分解算法的图件图像分割操作,并提出一种行、列扫描算法来提取图件图像的八维特征向量,试验结果表明了这种处理的有效性。更多还原

【Abstract】 Digital image processing has become an essential means in many modern technology fields.This paper applies this technology to the recognition of diagram pieces in PCB.To improve the quality of images and increase the identification rate of BP neural network of them,we preprocess the original images of PCB with MATLAB,which mainly includes the smoothing of digital image basing on the medium value filtering and the image diwision of PCB,and put forward an algorithm extracting the 8-dimension feature vector of diagram pieces′ image by row or column scanning.Experimental results show that this approach is efficient.更多还原