最大稳定极值区域与角点检测结合的医疗图像文本区域检测算法
|
摘要
医疗图像的文本区域检测在网络协同诊疗和医疗云的迅速发展中起着至关重要的作用。鉴于医疗图像相对复杂、文本字体太小而难以提取这一特殊问题,有针对性地提出了一种基于最大稳定极值区域算法与改进的角点检测算法相结合的医疗图像文本区域检测算法。该算法首先使用成分特征分析对最大稳定极值区域进行滤除,其次使用改进的最小核值相似区(smallest univalue segment assimilating nucleus,SUSAN)角点检测算子对剩余最大稳定极值区域进行检测并滤除,最后将检测出的所有文本区域使用膨胀处理进行合并即可得到医疗图像的文本区域。实验结果表明,使用该算法提取出的医疗图像文本区域的准确率、召回率和综合性能分别为0. 9、0. 92和0. 91,达到了理想的检测效果。
The text area detection of medical images plays a crucial role in the rapid development of network collaborative diagnosis and medical cloud. In view of the relatively complicated medical image and the small font size,it is difficult to extract this special problem. A medical image text region detection algorithm was proposed based on the combination of the maximally stable extremal regions algorithm and the improved corner detection algorithm. The algorithm first uses the component feature analysis to filter the maximally stable extremal regions,and then uses the improved smallest univalue segment assimilating nucleus( SUSAN) corner detection operator to detect and filter the remaining maximally stable extremal regions,and finally use all the detected text regions to expand.Processing merges to get the text area of the medical image. The experimental results show that the accuracy,recall and overall performance of the medical image text area extracted by the algorithm are 0. 9,0. 92 and 0. 91,respectively,which achieves the ideal detection effect.
The text area detection of medical images plays a crucial role in the rapid development of network collaborative diagnosis and medical cloud. In view of the relatively complicated medical image and the small font size,it is difficult to extract this special problem. A medical image text region detection algorithm was proposed based on the combination of the maximally stable extremal regions algorithm and the improved corner detection algorithm. The algorithm first uses the component feature analysis to filter the maximally stable extremal regions,and then uses the improved smallest univalue segment assimilating nucleus( SUSAN) corner detection operator to detect and filter the remaining maximally stable extremal regions,and finally use all the detected text regions to expand.Processing merges to get the text area of the medical image. The experimental results show that the accuracy,recall and overall performance of the medical image text area extracted by the algorithm are 0. 9,0. 92 and 0. 91,respectively,which achieves the ideal detection effect.
引文
1胡燕平,李乐乐,卢清君,等.基于远程医疗的异地转诊就医一体化实践探索[J].中国医院管理,2017,37(8):76Hu Yanping,Li Lele,Lu Qingjun,et al.Exploration of long-distance medical alliance based on telemedicine[J].Chinese Hospital Management,2017,37(8):76
2 Ye Q,Doermann D.Text detection and recognition in imagery:Asurvey[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2015,37(7):1480-1500
3吕国宁,高敏.视觉感知式场景文字检测定位方法[J].智能系统学报,2017,12(4):563-569Lv Guoning,Gao Min.Scene text detection and localization scheme with visual perception mechanism[J].CAAI Transactions on Intelligent System,2017,12(4):563-569
4刘亚亚,于凤芹,陈莹.基于连通区域和统计特征的图像文本定位[J].计算机工程与应用,2016,52(5):165-168Liu Yaya,Yu Fengqin,Chen Ying.Text location in image based on connected-component and statistical features[J].Computer Engineering and Applications,2016,52(5):165-168
5何立强,刘浩,陈永.边缘与灰度检测相结合的场景图像文本定位[J].中国图象图形学报,2013,18(12):1601-1609He Liqiang,Liu Hao,Chen Yong.Text extraction from scene image based on edge and gray-scale detection collaboration[J].Journal of Image and Graphics,2013,18(12):1601-1609
6 Ahmed K T,Iqbal M A.Region and texture based effective image extraction[EB/OL].https://doi.org/10.1007/s10586-017-0915-3
7李英,田春娜,颜建强,等.一种图像中的文字区域检测新方法[J].西安电子科技大学学报(自然科学版),2013,40(6):187-192Li Ying,Tian Chunna,Yan Jianqiang,et al.New technique for text region location in images[J].Journal of Xidian University(Natural science edition),2013,40(6):187-192
8 Matas J,Chum O,Urban M,et al.Robust wide-baseline stereo from maximally stable extremal regions[J].Image and Vision Computing,2004,22(10):761-767
9张鹏伟,张伟伟.用卷积神经网络分类最大稳定极值区域实现汉字区域定位[J].国防科技大学学报,2017,39(3):91-96Zhang Pengwei,Zhang Weiwei.Scene Chinese text localization by convolutional neural network classifying maximum stable extremal regions[J].Journal of National University of Defense Technology,2017,39(3):91-96
10高晶,吴育峰,吴昆,等.基于角点检测的图像匹配算法[J].仪器仪表学报,2013,34(8):1717-1725Gao Jing,Wu Yufeng,Wu Kun,et al.Image matching method based on corner detection[J].Chinese Journal of Scientific Instrument,2013,34(8):1717-1725
11 Matas J,Chum O,Urban M,et al.Robust wide-baseline stereo from maximally stable extremal regions[J].Image and Vision Computing,2004,22(10):761-767
12 Smith S M,Brady J M.SUSAN-A new approach to low level image processing[J].International Journal of Computer Vision,1997,23(1):45-78
2 Ye Q,Doermann D.Text detection and recognition in imagery:Asurvey[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2015,37(7):1480-1500
3吕国宁,高敏.视觉感知式场景文字检测定位方法[J].智能系统学报,2017,12(4):563-569Lv Guoning,Gao Min.Scene text detection and localization scheme with visual perception mechanism[J].CAAI Transactions on Intelligent System,2017,12(4):563-569
4刘亚亚,于凤芹,陈莹.基于连通区域和统计特征的图像文本定位[J].计算机工程与应用,2016,52(5):165-168Liu Yaya,Yu Fengqin,Chen Ying.Text location in image based on connected-component and statistical features[J].Computer Engineering and Applications,2016,52(5):165-168
5何立强,刘浩,陈永.边缘与灰度检测相结合的场景图像文本定位[J].中国图象图形学报,2013,18(12):1601-1609He Liqiang,Liu Hao,Chen Yong.Text extraction from scene image based on edge and gray-scale detection collaboration[J].Journal of Image and Graphics,2013,18(12):1601-1609
6 Ahmed K T,Iqbal M A.Region and texture based effective image extraction[EB/OL].https://doi.org/10.1007/s10586-017-0915-3
7李英,田春娜,颜建强,等.一种图像中的文字区域检测新方法[J].西安电子科技大学学报(自然科学版),2013,40(6):187-192Li Ying,Tian Chunna,Yan Jianqiang,et al.New technique for text region location in images[J].Journal of Xidian University(Natural science edition),2013,40(6):187-192
8 Matas J,Chum O,Urban M,et al.Robust wide-baseline stereo from maximally stable extremal regions[J].Image and Vision Computing,2004,22(10):761-767
9张鹏伟,张伟伟.用卷积神经网络分类最大稳定极值区域实现汉字区域定位[J].国防科技大学学报,2017,39(3):91-96Zhang Pengwei,Zhang Weiwei.Scene Chinese text localization by convolutional neural network classifying maximum stable extremal regions[J].Journal of National University of Defense Technology,2017,39(3):91-96
10高晶,吴育峰,吴昆,等.基于角点检测的图像匹配算法[J].仪器仪表学报,2013,34(8):1717-1725Gao Jing,Wu Yufeng,Wu Kun,et al.Image matching method based on corner detection[J].Chinese Journal of Scientific Instrument,2013,34(8):1717-1725
11 Matas J,Chum O,Urban M,et al.Robust wide-baseline stereo from maximally stable extremal regions[J].Image and Vision Computing,2004,22(10):761-767
12 Smith S M,Brady J M.SUSAN-A new approach to low level image processing[J].International Journal of Computer Vision,1997,23(1):45-78