改进区域生长算法在视杯图像分割中的应用
|
摘要
目的:视杯图像分割对于通过眼底图像检测青光眼具有重要意义,在传统的区域生长算法基础上进行改进,提出了基于眼底图像的视杯自动检测分割方法.方法:首先,对眼底主要生理结构进行特征分析,为分割目标选取了绿色通道并根据阈值法粗略提取出感兴趣区域(ROI);其次,考虑到传统的区域生长算法在选取种子点时不精确、自适应性差等缺点,通过计算ROI的几何中心并结合中心亮度作为选取种子点的标准进行改进;最后,用5*5模板对眼底图像进行均值滤波,应用山谷差值准则和8邻域连通准则对眼底图像进行种子合并,最终准确分割出视杯.结果:应用这种方法,对高分辨率眼底图像(HRF)数据库中15张青光眼眼底图像和15张健康眼眼底图像逐张进行检测,准确率达到93.3%.结论:实验结果表明,该算法能快速、有效地自动检测出眼底图像中的视杯并将其正确的分割出来,与传统算法相比较该算法稳定可靠,有较高的分割灵敏度、特异度以及准确性.
Purpose: Optic cup image segmentation has great significance to glaucoma measurement by retinal image. This article is based on the traditional region growing algorithm to improve,and put forward the optic cup automatic detection based on retinal image segmentation method. Methods: Firstly,the characteristics of fundus main physiological structure is analized,and then the green channel is chosen,thus the region of interest( ROI) roughly for target segmentation is extracted according to the threshold value method; Secondly,considering that the traditional region growing algorithm is not accurate and has poor adaptability when selecting the seed point. To improve this situation,this articleselect seed points by calculating geometric center of ROI and combining the central brightness; Finally,filter the retinal image by 5 * 5 template,merger seed by valley difference criterion and use 8 neighborhood connected rules to retinal image. Then optic cup is accurately extracted finally. Results: This method applied in this article tests 15 pieces of glaucoma retinal image and 15 pieces healthy eye retinal image in the highresolution retinal image( HRF) database,and the accuracy is 93. 3%. Conclusion: The experimental result shows that the algorithm can rapidly and effectively automatically detect the optic cup of retinal images and extracts the optic cup correctly. Compared with the traditional algorithm,this one is stable and reliable,and has higher sensitivity,specificity and accuracy.
Purpose: Optic cup image segmentation has great significance to glaucoma measurement by retinal image. This article is based on the traditional region growing algorithm to improve,and put forward the optic cup automatic detection based on retinal image segmentation method. Methods: Firstly,the characteristics of fundus main physiological structure is analized,and then the green channel is chosen,thus the region of interest( ROI) roughly for target segmentation is extracted according to the threshold value method; Secondly,considering that the traditional region growing algorithm is not accurate and has poor adaptability when selecting the seed point. To improve this situation,this articleselect seed points by calculating geometric center of ROI and combining the central brightness; Finally,filter the retinal image by 5 * 5 template,merger seed by valley difference criterion and use 8 neighborhood connected rules to retinal image. Then optic cup is accurately extracted finally. Results: This method applied in this article tests 15 pieces of glaucoma retinal image and 15 pieces healthy eye retinal image in the highresolution retinal image( HRF) database,and the accuracy is 93. 3%. Conclusion: The experimental result shows that the algorithm can rapidly and effectively automatically detect the optic cup of retinal images and extracts the optic cup correctly. Compared with the traditional algorithm,this one is stable and reliable,and has higher sensitivity,specificity and accuracy.
引文
[1]郑姗,范慧杰,唐延东,等.多相主动轮廓模型的眼底图像杯盘分割[J].中国图象图形学报,2014,19(11):1604-1612.
[2]赵秋红.彩色眼底图像中视杯识别算法研究[D].北京:北京工业大学,2014:1-8.
[3]Irene Fondón,Jose Francisco Valverde,Auxiliadora Sarmiento,et al.Automatic Optic Cup Segmentation Algorithm for Retinal Fundus Images based on Random Forest Classifier[J].IEEE International Conference on Computer as a Tool:from Salamanca,2015:1-6.
[4]Ashish Issac,M.Parthasarthi,Malay Kishore Dutta.An Adaptive Threshold Based Algorithm for Optic Disc and Cup Segmentation in Fundus Images[J].IEEE International Conference on Signal Processing and Integrated Netw orks:from Noida,2015:143-147.
[5]Dwarikanath Mahapatra,Joachim M.Buhmann.A Field of Experts Model for Optic Cup and Disc Segmentation from Retinal Fundus Images[J].IEEE 12th International Symposium on Biomedical Imaging:New York,2015:218-221.
[6]Hanan Alghmdi,Hongying Lilian Tang,Morten Hansen.Measurement of Optical Cup-to-Disc Ratio in Fundus Images For Glaucoma Screening[J].IEEE Computational Intelligence for M ultimedia Understanding:Prague,1-5.
[7]Nataraj A.Vijapur,R.Srinivasa Rao Kunte.Glaucoma Detection by Using Pearson-R Correlation Filter[J].IEEE Communications and Signal Processing,2015 International Conference on M elmaruvathur,2015:1194-1198.
[8]Mahaman Sani Chaibou,Karim Kalti,Basel Solaiman,et al.A combined Approach Based on Fuzzy Classification and Contextual Region Grow ing To Image Segmentation[J].IEEE 13th International Conference on Computer Graphics,Beni M ellal,2016,172-177.
[9]Kishor Kumar Reddy C,Anisha P R,Raju G V S.A Novel Approach for Detecting the Tumor Size and Bone Cancer Stage using Region Grow ing Algorithm[J].International Conference on Computational Intelligence and Communication Netw orks,Jabalpur,2015:228-233.
[10]HRF database available on URL:https://www5.cs.fau.de/research/data/fundus-images/
[2]赵秋红.彩色眼底图像中视杯识别算法研究[D].北京:北京工业大学,2014:1-8.
[3]Irene Fondón,Jose Francisco Valverde,Auxiliadora Sarmiento,et al.Automatic Optic Cup Segmentation Algorithm for Retinal Fundus Images based on Random Forest Classifier[J].IEEE International Conference on Computer as a Tool:from Salamanca,2015:1-6.
[4]Ashish Issac,M.Parthasarthi,Malay Kishore Dutta.An Adaptive Threshold Based Algorithm for Optic Disc and Cup Segmentation in Fundus Images[J].IEEE International Conference on Signal Processing and Integrated Netw orks:from Noida,2015:143-147.
[5]Dwarikanath Mahapatra,Joachim M.Buhmann.A Field of Experts Model for Optic Cup and Disc Segmentation from Retinal Fundus Images[J].IEEE 12th International Symposium on Biomedical Imaging:New York,2015:218-221.
[6]Hanan Alghmdi,Hongying Lilian Tang,Morten Hansen.Measurement of Optical Cup-to-Disc Ratio in Fundus Images For Glaucoma Screening[J].IEEE Computational Intelligence for M ultimedia Understanding:Prague,1-5.
[7]Nataraj A.Vijapur,R.Srinivasa Rao Kunte.Glaucoma Detection by Using Pearson-R Correlation Filter[J].IEEE Communications and Signal Processing,2015 International Conference on M elmaruvathur,2015:1194-1198.
[8]Mahaman Sani Chaibou,Karim Kalti,Basel Solaiman,et al.A combined Approach Based on Fuzzy Classification and Contextual Region Grow ing To Image Segmentation[J].IEEE 13th International Conference on Computer Graphics,Beni M ellal,2016,172-177.
[9]Kishor Kumar Reddy C,Anisha P R,Raju G V S.A Novel Approach for Detecting the Tumor Size and Bone Cancer Stage using Region Grow ing Algorithm[J].International Conference on Computational Intelligence and Communication Netw orks,Jabalpur,2015:228-233.
[10]HRF database available on URL:https://www5.cs.fau.de/research/data/fundus-images/