激光扫描翻译笔应用模式识别方法

Application pattern recognition method of laser scanning translation pen

当前模式识别方法在进行图像文字特征识别时,对文本图像中的文字位置把握不够准确,造成识别文本图像点云曲率过低。因此,研究激光扫描翻译笔应用模式识别方法。对激光扫描翻译笔的模式识别方法进行设计,对有噪声的目标图像进行预处理,矫正畸变的激光扫描文本图像。对机关扫描文本图像中文字的位置进行定位,从而获得文本文字轮廓位置信息。通过文本模式识别方法,对文本图像中的文字特征进行分析。整合文字特征图像,对特征文字进行翻译,并通过模式识别输出翻译文本信息。至此,完成对激光扫描翻译笔的模式识别方法设计。通过实验分析得到:传统方法1识别特征子集的识别图像点云曲率最高为0.11 m^(-1),传统方法2识别特征子集的识别图像点云曲率最高为0.09 m^(-1),本文方法识别特征子集的识别图像点云曲率最高为0.12 m^(-1)。因此激光扫描翻译笔应用模式识别方法更好。

The current pattern recognition method does not grasp the position of the text in the text image accurately enough when identifying the image text features,resulting in the low curvature of the point cloud of the recognition text image.Therefore,the application pattern recognition method of laser scanning translation pen is studied.The pattern recognition method of the laser scanning pen is designed,the noisy target image is preprocessed,and the laser scanning text image with distortion is corrected.Position the position of the text image Chinese the organ scanned text to obtain the position information of the text outline.The text features in the text image are analyzed by the text pattern recognition method.Integrate text feature images,translate feature text,and output translated text information through pattern recognition.At this point,the pattern recognition method design of the laser scanning translation pen is completed.Experimental analysis shows that the point cloud curvature of the identified image of the traditional method 1 recognition feature subset is up to 0.11 m^(-1),the traditional method 2 identifies the feature subset of the identifying image point cloud curvature is up to 0.09 m^(-1),and the point cloud curvature of the identified image of the feature subset identified by the proposed method is up to 0.12 m^(-1).Therefore,the laser scanning pen applies a better pattern recognition method.