摘要:针对无定位信息的光学标记识别(OMR)图像填涂区的精确定位问题，提出了一种基于小波变换突变信号检测的图像分割方法。该算法首先计算图像的水平和垂直投影函数，然后投影函数经过迭代小波变换后检测其突变点，突变点能够精确地反映OMR信息的边界位置。检测算法的适应性基于有限次数的小波变换和突变信号检测过程。实验结果表明该算法具有较高的分割精度和稳定性，分割精度均方差可以达到0.416�7个像素。而且由于算法只使用图像的水平和垂直投影信息，因此具有较高的执行效率；投影函数的统计特性和小波变换的多分辨特性则使得该分割算法对噪声不敏感。
关键词:光学标记识别；小波变换；图像分割；突变点检测；多分辨分析�
中图分类号: TP391.4文献标志码:A
�
OMR image segmentation based on mutation signal detection
�
MA Lei�1,2, LIU Jiang�1,2�*, LI Xiao.peng�1,2, CHEN Xia�1,2
1. Shandong Engineering Research Institute for Image Acquisition and Processing, Jinan Shandong 250101,China�；�
2. Shandong Shanda Oumasoft Company Limited, Jinan Shandong 250101,China
Abstract:
Aiming at the problem of accurate positioning of Optical Mark Recognition (OMR) images without any position information, an image segmentation approach based on wavelet transformation mutations signal detection is proposed. Firstly the horizontal and vertical projective operation are processed, then these functions are transformed by wavelet and detected point mutation, these points are a better description of the boundary of OMR information. This algorithm adaptability based on limited times of wavelet transform and mutations signal detection. The experimental results demonstrate that the method possesses high accuracy of segmentation and stability, the mean square error of segmentation accuracy can be 0.4167 pixels. The processing of this method was efficient because the segmentation only based the horizontal and vertical information. This algorithm was not sensitive to noise because of projection functions statistic characteristic and multi-resolution analysis of wavelet.
Concerning the accurate positioning of Optical Mark Recognition (OMR) images without any position information, an image segmentation approach of mutation signal detection based on wavelet transformation was proposed. Firstly, the horizontal and vertical projective operations were processed, and then these functions were transformed by wavelet to detect mutation points, which can better reflect the boundary of OMR information. This algorithm�s adaptability is based on limited times of wavelet transform and mutation signal detection. The experimental results demonstrate that the method possesses high accuracy of segmentation and stability, and the mean square error of segmentation accuracy can be 0.416�7 pixels. The processing of this method is efficient because the segmentation only used the horizontal and vertical information. This algorithm is not sensitive to noise because of the statistic characteristic of projection functions and multi.resolution characteristic of wavelet tranformation.
�Key words:
Optical Mark Recognition （OMR）; wavelet transformation; image segmentation; singularity detection; multiresolutlon analysis
�
0 引言�
光学标记识别(Optical Mark Recognition, OMR)技术具有快速简单、识别率高且成本低的特点，被大量应用于考试、票据、报表、普查等领域，具有很高的应用价值。OMR 一般由定位标记和选项识别区两部分构成�［1］，早期的OMR采集技术使用光学感应设备直接对信息卡上的涂点进行对应采集识别，随着高速图像采集设备的发展，出现了基于图像的OMR采集方式，其方法是首先对信息卡进行信息采集，得到数字图像，使用图像处理对涂点进行识别处理，转化为数字信息。传统的OMR信息卡都含有对填涂区域分割起到关键作用的定位信息�［2］，能够方便地对图像进行分割。然而，随着2004年全国高考领域数字化网上阅卷的兴起，基于图像的OMR识别关键技术也成为相关人员的研究重点，这些技术不断向中学领域扩展并研发出针对普通学校的OMR识别技术，大部分中学领域采用了无标记定位信息的OMR识别技术，针对无标记定位信息的自动分割方法的研究却一直处于空白状态，因此信息卡填涂区的分割定位问题具有重要研究意义。
如图1所示，OMR图像中没有定位信息，图像需要借助于信号检测实现填涂区域的精确分割，这种类型的图像被大量用于中学考试领域。�
小波变换可用于奇异值的检测和处理�［3］，并可消除图像中的噪声从而达到增强图像的目的，也可通过小波变换系数模极大值实现图像的边缘检测和恢复�［4］。小波变换的一个显著特点是能够确定被分析信号的局部奇异性�［5］，奇异位置与不同尺度下小波系数模极大值相关，这些模极大值可以汇聚成一条模极大值曲线。奇异值一般是指是各种瞬态信号的主要特征�［6］，可以使用李氏指标描述奇异值，Mallat等已证明通过不同尺度下的小波局部模极大值的衰减特征可求得李氏指标�［7］。奇异值的分析主要是确定奇异值的位置和检测信号奇异程度�［8］。基于小波变换的突变信号分析方法在实际
应用领域非常广泛，例如输油管线的泄露检测�［9］、生物工程学中的尖刺信号检测�［10］及人体心电信号（Electro Cardio Graph，ECG）处理�［11］。�
本文研究基于小波变换奇异值检测方法分割OMR图像，确定OMR填涂区的边界位置，核心问题是小波母函数的选取和奇异值位置的确定。�
1 小波变换与突变信号检测�
Mallat将函数(信号)的局部奇异性与小波变换后的模局部极大值联系起来，通过小波变换后的模极大值在不同的尺度上的衰减速度来衡量信号的局部奇异性。