摘要
针对人工检测刹车片字符时低效率的现状,提出一种基于机器视觉软件Halcon的字符检测算法。该算法在光照不均匀的条件下,充分利用了刹车片的外形特征,通过数幅图像测试后确定合适的分割方法和最佳阈值,定位出感兴趣区域(字符区域),通过适度膨胀解决字符断裂的难题,并特别处理了易发生混淆的I和1,测试图像达到了96%的识别率。结果表明,该算法可高效准确地识别出刹车片字符,且具有较高的实用价值。
A character detection system based on the machine vision software Halcon is proposed in view of the low efficiency of manual detection of the brake characters. Under the condition of uneven illumination, the algorithm makes full use of the shape characteristics of the brake, determines the appropriate segmentation method and the best threshold value through 30 images testing, and locates the ROI （ character region）. It solves the problem of character breaking through appropriate expansion and especially distinguishes I and 1. The recognition rate of the testing images is 96% . The results show that the algorithm can identify the characters of the brake efficiently and accurately.
出处
《电子科技》
2016年第10期101-103,共3页
Electronic Science and Technology
