字符识别耦合的ROI视觉引导应用

Visual Guide Technology Based on Character Recognition and ROI

字符是标识产品的重要信息,由于产品表面成像画质多样性,当字符目标不清晰或者背景干扰大,往往会影响识别算法的精准度.鉴于此,提出了一个基于Emgucv与Tesseract的字符识别机制,用来识别平板电脑表面薄膜字符.首先利用网络摄像头,对平板电脑表面薄膜字符区域取像;再对获取的灰度图进行阈值分割得到包含目标的二值图;然后利用形态学处理去除杂质干扰、提取目标特征,得出感兴趣区域(ROI);最后基于Tesseract开源库,实现对ROI区域的字符识别.整个系统图像处理部分由C#和Emgucv实现,根据字符识别结果,用运动控制卡传递命令给机构,对薄膜进行分流,完成视觉引导.通过实验测试本文字符识别系统性能,结果表明本机制与当前字符识别技术相比具有更好的识别效果.

Character is important information identifying a product. Owing to the product surface imaging quality diversity, with the character not clear or background interference, the accuracy of recognition algorithm would be greatly influenced. Therefore, this paper proposes a character recognition mechanism based on Emgucv and Tesseract to identify the tablet PC＇s surface film character. First, to capture the tablet PC＇s surface film character area via a webcam. Then, to harvest a binary image containing the character object via making threshold segmentation in the grey-scale map obtained. Third, to get ROI region via wiping off impurity. Finally, based on the open source Tesseract library, to reallzc the character recognition in ROI region. The image processing in the whole system is achieved by C# and Emgucv, according to the character recognition results, the films were shunt based on the motion control card to send orders to complete visual guide. It comes to a conclusion that this mechanism has better recognition effect compared with the current character recognition technology.