印鉴图像预处理方法研究  

Study on image preprocessing method of seal imprint

作　　者：项阳[1] 潘新[1] 

机构地区：[1]内蒙古农业大学计算机与信息工程学院,内蒙古呼和浩特010010

出　　处：《微型机与应用》2017年第15期41-44,共4页Microcomputer & Its Applications

摘　　要：为了提高印鉴识别的准确性,在识别前需要对印鉴图像进行预处理。针对印鉴常与银行票据要素混杂的实际情况,基于RGB彩色模型,将影响印鉴提取的两个干扰源——字体和书写线分别进行处理。首先通过选取合适的阈值二值化图像,利用Canny算子和Hough变换提取字体和书写线边缘,然后与原二值图像做"与"运算,求出需要填充的印鉴缺失信息轮廓,并对该轮廓进行闭运算、中值滤波,最后填充缺失部分,得出完整的印鉴图像。实验结果显示,用此方法处理后的印鉴图像质量较高,能够满足后期识别的需要。In order to improve the accuracy of seal identification,the image preprocessing of seal imprint is needed before identification process. According to realities that the seal imprint often mixes with bank note elements,based on the RGB color mode,two sources of interference which are characters and writing lines are processed separately. First of all,selecting a appropriate threshold to get binary image and take edges of characters and writing lines by using Canny operator and Hough transform,and then among the edges and original binary image do'and operation'to seek outlines of missing information. Secondly,using closed operation and median filtering for the outlines. Finally,filling the missing information of the image,so a complete image of seal imprint is presented. The experimental results show the image which is processed by this method has a character of high quality and can satisfy the requirements for the succeeding verification work.

关 键 词：图像预处理 RGB彩色模型 CANNY算子 HOUGH变换 图像边缘 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]