随着我国交通服务行业的发展, 火车票票面信息的自动化识别已经成为提升铁路服务效率的重要手段, 针对票面信息识别系统中获取火车票方向不一致的问题, 本文提出了一种基于特征区域聚类下Radon变换的方向检测算法。首先, 算法通过k均值聚类对火车票图像进行聚类分块, 消除复杂背景的干扰, 提取到火车票信息区域; 然后, 通过数字形态学闭运算结合图像块操作保留能够反映火车票位置信息的图像方向特征区域; 最后, 利用改进的Radon变换检测出火车票的倾斜角度。实验结果表明: 该算法的矫正正确率为97.8%, 矫正的时间为16.79 s; 该算法能够消除图像复杂背景、方向特征区域领域像素点对方向检测的干扰, 能够对全角度任意方向的火车票图像进行方向检测, 具有较高的实用价值。

With the development of transportation service industry in China, automatic identification of railway ticket information has become more important to improve the efficiency of railway service. In order to solve the problem of inconsistency in train ticket direction acquisition in ticket surface information recognition system, a direction detection algorithm is proposed in this paper which is based on Radon transform with feature area clustering. Firstly, the algorithm classifies train ticket images by K-means clustering, eliminates the interference of complex background and extracts the railway ticket information area. Then, the image direction feature area which can reflect the train ticket position information is preserved through digital morphology closed operation combined with image block operation. Finally, the tilted angle of train ticket is detected by the improved Radon transform. The experimental results show that the correction accuracy of the algorithm is 97.8%, and the correction time is 16.79 s. By the algorithm, it can eliminate the interference of the image complex background and neighborhood pixels of direction feature area to the direction detection, and detect the direction of train ticket image in any direction with full angles, which are of high practical value.