Abstract: Aiming at the second generation identification card’s character identification，this paper puts forward a method that based on character string’s profile detection.This method firstly adopts texture segmentation of the area of image character，after gaining the area of character，then adopts fluctuation profile concavo-convex character to approximately detect single character’s width，and accordingly choose the steady local character，and use the method about frame sentence identification to identify the character.Through the experimentation of ten pieces of the second generation identification card’s character identification which is screened in the nature scene，this method is proved validity.

Key words: digital binary, threshold, gray scale, texture, profile detection, character identification

摘要： 针对第二代身份证图像字符识别，提出了一种基于字符串轮廓检测的方法。该方法采用对图像字符区域先进行纹理分割，获取字符区域后对其上下轮廓凹凸特征近似检测单个字符的宽度，从中选出稳定的局部特征，利用结构语句识别的方法进行字符识别。通过对自然场景下拍摄的10幅第二代身份证图像的字符识别实验，证明了该方法的有效性。

关键词: 二值化, 阈值, 灰度值, 纹理特征, 轮廓检测, 字符识别