基于图像熵和傅里叶变换的复杂背景分割  

Complex Background Segmentation based on Image Entropy and Fourier Transform

作　　者：李镇锋 陈晓荣[1] 陈梦华 卢德运 陈运豪 LI Zhenfeng;CHEN Xiaorong;CHEN Menghua;LU Deyun;CHEN Yunhao(School of Optical-Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093

出　　处：《软件工程》2021年第11期19-23,共5页Software Engineering

摘　　要：在机动车登记证识别过程中,由于机动车登记证图像是无规则的彩色背景,传统的图像分割算法难以得到较好的分割效果,为了提高识别的准确度,提出了基于图像熵和傅里叶变换的复杂背景分割方法。首先,使用形态学腐蚀运算分割出图像中的直线,再利用直线从整体图像中提取出局部图像。其次,计算每幅局部图像的图像熵,若该局部图像的熵小于阈值T,使用Otsu(最大类间方差)分割算法对其进行分割;若该局部图像的熵大于阈值T,则先通过傅里叶变换在频率域滤波后再使用Otsu算法对其进行分割。实验结果表明,该算法对机动车登记证图像能够有效进行分割,进而从复杂背景中提取出不含噪声的字符信息。In the process of vehicle registration certificate recognition,it is difficult for traditional image segmentation algorithms to obtain better segmentation results,due to the irregular color background of the vehicle registration certificate image.In order to improve recognition accuracy,this paper proposes a complex background segmentation method based on image entropy and Fourier transform.First,morphological erosion operation is used to segment the straight lines,which are used to extract partial images from the overall image;secondly,the image entropy of each partial image is calculated.If the entropy of the local image is less than the threshold T,Otsu(maximum between-class variance)segmentation algorithm is used to segment it.If the entropy of the partial image is greater than the threshold T,it will be filtered in the frequency domain through Fourier transform and then segmented by using Otsu algorithm.The experimental results show that the algorithm can effectively segment the motor vehicle registration certificate image,and then extract the noise-free character information from the complex background.

关 键 词：复杂背景分割 图像熵 傅里叶变换 OTSU分割 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]