【摘要】 将机器视觉技术融入装车机，实现对火车棚车特征字符区域进行定位、分割与识别。首先对采集到图像进行预处理，定位出字符区域位置，然后进行字符矫正和剔除不属于字符的区域，接着基于连通域分析法分割字符，最后论述并使用模板匹配、OCR识别以及卷积神经网络3种识别方法对同一分割的字符进行识别，得出每种方法识别的准确率。实验结果证明，在同样的图像预处理及字符分割情况下，卷积神经网络识别结果最好，识别准确率达到96%。该研究能很好识别火车特征字符，同时也为其他类型特征字符识别提供研究思路。更多还原

【Abstract】 The machine vision technology is integrated into the loading machine to realize the location, segmentation,and recognition of the characteristic character area of the train box car. Firstly, the acquired image is preprocessed to locate the character region, then the character correction is carried out and the region that does not belong to the character is removed, and then the character is segmented based on the connected domain analysis method. At last,the recognition accuracy rate of each method is obtained by discussing and using three recognition methods, namely template matching, OCR recognition, and convolutional neural network. The experimental results showed that under the same condition of image preprocessing, and character segmentation, the recognition result of convolutional neural network was the best, and the recognition accuracy reached 96%. This research can identify train characteristics characters well, and also provide research ideas for other types of character recognition.更多还原