移动端身份证号码识别算法实现

Implementation of ID Card Number Recognition Algorithms Based on Mobile Terminal

为实现多场景下快速便捷的身份证号码提取,提出一种基于移动端的身份证号码识别算法。该算法利用搭载Android操作系统的移动设备,进行身份证号码区域定位和提取;首先利用身份证特殊的颜色分布,选取合适的通道分量;再通过图像阈值分割、噪声处理以及形态学处理,将身份证号码图像二值化;同时针对拍摄过程中可能出现的图像倾斜情况,通过图像旋转和投影法进行水平矫正;然后通过投影法将单个字符提取出来;最后使用卷积神经网络进行图像识别。通过MIX手机测试表明,单次身份证号码图像识别时间约为156ms,准确率约为99.1%,能够满足鲁棒性和实时性要求。

In order to realize fast and convenient ID number extraction in multi-scene,this paper proposes an ID number recognition algorithm based on mobile terminal.The algorithm uses mobile devices with Android operating system to locate and extract the ID num?ber area.Firstly,it uses the special color distribution of the ID card to select the appropriate channel components.Then,it binarizes the ID number image through image threshold segmentation,noise processing and morphological processing.At the same time,it aims at the possible image tilt during the shooting process.Horizontal rectification is carried out by image rotation and projection method;then single character is extracted by projection method;finally,image recognition is carried out by convolution neural network.The MIX mobile phone test shows that the recognition time of single ID card number image is about 156 ms,and the accuracy is about 99.1%,it can meet the requirements of robustness and real-time.