两个图像分割算法的研究与实现
|
摘要
图像工程是近几年发展起来的一门学科，它的研究内容非常丰富，根据抽象程度和研究方法等的不同可分为三个层次：图像处理、图像分析和图像理解。
图像分割是一种重要的和关键的图像分析技术，对图像目标进行提取、测量等都离不开它，图像分割的结果是图像特征提取和识别等图像理解的基础，对它的研究一直是图像技术研究中的热点和焦点。本文针对图像的分割问题进行了研究与实现。
针对飞行物图像的特点，提出了一种新的基于边界特征的一维最大熵图像二值化分割方法，该方法可以自适应地选取阈值。实验表明，本方法可以克服目前已有的一维最大熵取阈值方法进行图像分割时，由于阈值取的太低而丢失细节信息的缺点，有效地保留了边界的细节特征。该算法简单，易于实现，执行速度快，特别适合于质量低和边缘模糊的图像。笔者对实验结果进行了详细的讨论，得出了有意义的结论。
基于信息融合的观点，本文提出了融合灰度信息与纹理信息的图像分割方法，并且采取像素级融合技术，使得图像中的目标得到了增强。实验结果证明了算法的有效性。
Image engineering is a subject that has been developed in recent years, and it has many contents. According to the degree of abstract and the investigate methods, the research on it can be divided into three levels: image processing, image analysis and image comprehension.
Image segmentation is a critical image analysis technique. It is necessary in the process of image object' s extraction and measurement. The result of the image segmentation is the basis of the feature extraction and pattern recognition. The research of the image segmentation is the hot spot and focus of the image technology. This paper make study of and implement the image segmentation.
According to the character of the flyers, a one-dimension maximum entropy image binary conversion method based on edge features is proposed, which can implement adaptive threshold selection. Our experiments show that the new method overcomes the disadvantage of image segmentation with one-dimension maximum entropy and keeps the original edge features well. This method is simple. It is implementing. Especially, it is efficient for processing low quality and edge fuzzy images. We discuss the findings of the experiments in detail and draw good conclusions.
Based on the idea of the information fusion, this paper proposes the image segmentation algorithm, which uses fusion of gray information and texture information. And it takes the pixel fusion technology. All this boosts up the image object. The validity of this algorithm is verified in the experiments.
图像分割是一种重要的和关键的图像分析技术，对图像目标进行提取、测量等都离不开它，图像分割的结果是图像特征提取和识别等图像理解的基础，对它的研究一直是图像技术研究中的热点和焦点。本文针对图像的分割问题进行了研究与实现。
针对飞行物图像的特点，提出了一种新的基于边界特征的一维最大熵图像二值化分割方法，该方法可以自适应地选取阈值。实验表明，本方法可以克服目前已有的一维最大熵取阈值方法进行图像分割时，由于阈值取的太低而丢失细节信息的缺点，有效地保留了边界的细节特征。该算法简单，易于实现，执行速度快，特别适合于质量低和边缘模糊的图像。笔者对实验结果进行了详细的讨论，得出了有意义的结论。
基于信息融合的观点，本文提出了融合灰度信息与纹理信息的图像分割方法，并且采取像素级融合技术，使得图像中的目标得到了增强。实验结果证明了算法的有效性。
Image engineering is a subject that has been developed in recent years, and it has many contents. According to the degree of abstract and the investigate methods, the research on it can be divided into three levels: image processing, image analysis and image comprehension.
Image segmentation is a critical image analysis technique. It is necessary in the process of image object' s extraction and measurement. The result of the image segmentation is the basis of the feature extraction and pattern recognition. The research of the image segmentation is the hot spot and focus of the image technology. This paper make study of and implement the image segmentation.
According to the character of the flyers, a one-dimension maximum entropy image binary conversion method based on edge features is proposed, which can implement adaptive threshold selection. Our experiments show that the new method overcomes the disadvantage of image segmentation with one-dimension maximum entropy and keeps the original edge features well. This method is simple. It is implementing. Especially, it is efficient for processing low quality and edge fuzzy images. We discuss the findings of the experiments in detail and draw good conclusions.
Based on the idea of the information fusion, this paper proposes the image segmentation algorithm, which uses fusion of gray information and texture information. And it takes the pixel fusion technology. All this boosts up the image object. The validity of this algorithm is verified in the experiments.
引文
[1] 章毓晋 图像处理和分析 清华大学出版社
[2] 章毓晋 图像分割 科学出版社
[3] K．R．Castleman著 朱志刚 林学訚 石定机等译 数字图像处理 电子工业出版社
[4] 田捷 罗希平等 “图像分割方法综述” 模式识别与人工智能 vol．12 No．3
[5] Gonzalez R C, Woods R E. Digital Image Processing. 3rd. ed., Addison-Wesley, 1992
[6] Pal N.R. Pal and S.K.A Review on Image Segmentation Techniques. Pattern Recognition, vol. 26, pp.1277-1294,1993
[7] Zhang, Y.J.A survey of evaluation methods for image segmentation, Pattern Recognition, 1996,29(8):1335-1346
[8] 冯晓毅 周凤歧 “图像分割的一种神经网络实现方法” 计算机工程与科学 Vol．22，No．2，2000
[9] Abutaleb A S. Automation Thresholding of Gray-Level Pictures Using Two-Dimensional Entropy, Computer Vision, Graphics, and Image Processing, 1989, 47:22-32
[10] 华长发 范建平 高传善 吴立德 “基于两维熵阈值的图像分割及其快速算法” 模式识别与人工智能 Vol．13 No．1 March 2000
[11] 谌海新 沈震康 夏放怀 “一种基于目标特征的多门限图像分割方法” 电子学报 No．3 Vol．27 1999
[12] 安玮 徐晖 李宏 孙仲康 “奇异点多门限图像分割算法” 计算机工程与科学 Vol．20，No．3，1998
[13] 杨恬 李德芳 “灰度图像的二维OTSU自动阈值分割研究” 西南师范大学学报(自然科学版) 23卷第6期
[14] M.E. Sieracki, S.E. Reichenbach, and K.L. Webb, Evaluation of Automated Threshold Selection Methods for Accurately Sizing Microscopic Fluoprescent Cells by Image Analysis, Applied and Environmental Microbiology, 55, 11: 2762-2772, Nov. 1989
[15] 张新明 沈兰荪 沈波 “基于特征距离的阈值法及其在眼科图像分割中的应用” 中国图像图形学报 Vol．6(A) No．2 Feb 2001
[16] 刘文萍 吴立德 “图像分割中阈值选取方法比较研究” 模式识别与人工智能 Vol．10 No．3 1997
[17] Theo Pavlidis ans Yuh-Tay Liow, Integrating Region Growing and Edge Detection, IEEE Trans on PAMI, 12, (1990), 225-233
[18] Uchiyama T, Arbib M A. Color image segmentation using competitive learning. IEEE-PAMI, 1994, 16:1197-1206
[19] 柏子游等 “一种彩色图像的色彩分割方法” 模式识别与人工智能12卷第2期
[20] Lim Y W, Lee S U. On the color image segmentation algorithm based on the thresholding and fuzzy C-means techniques. Pattern Recognition, 1990, 23(9): 935-952
[21] 郭国栋 马颂德 “彩色图像分割” 中国图像图形学报 Vol．3 No，11 Nov． 1998 918-921
[22] Ohlander R, Price K, Reddy D R. Picture segmentation using a recursive region splitting method. CGIP, 1978, 8(3): 313-333
[23] 林通 石青云 “一种基于生长的灰度和彩色图像分割方法” 中国图像图形学报 Vol．5(A) No．11 Nov 2000
[24] 粟海华 王睿斌 庄镇泉 “一种基于判决神经网络的彩色图像分色算法” 模式识别与人工智能 1999，12(3)：354—358
[25] 张二虎 刘世昌 胡涛 “一种基于熵的图像分割方法” 微电子学与计算机 1998年第4期
[26] Shanbhag A G. Utilization of information measure as a means of image thresholding. CVGIP-GMIP, 1994,56(5): 414-419
[27] Kunt M et al. Second-generation Image Coding Scheme. Proc IEEE, 1985,73:549-574
[28] Fan J, et al. Automatic Moving Object Extraction Towerd Compact Video Representation. Optical Engineering, 2000, 39:438-452
[29] 刘俊、付敬奇、董新平，“数据融合在目标识别中的应用” 传感器技术 2001年第26卷第6期
[30] B Solaiman, R Debon Pipelier. Information Fusion: Application to data and model fusion for ultrasound image segmentation. IEEE Transactions on Biomedical Engineering, 1999, 46(10): 3-8
[31] 刘等第 傅惠民 殷刚 张少波 “一种新的数据融合方法” 中国安全科学学报 Vol8 No．6，Dec．，1998
[32] 汪培庄、李洪兴 模糊系统理论与模糊计算机 北京科学出版社，1996
[33] 赵锋 赵荣椿 纹理分割及特征提取方法综述 中国体视学与图像分析 Vol．3，
No.4, Dec 1998
[34] De Souza P. Edge based segmentation and texture separation, Proceedings, 5th Int. Con. on Pattern Recognition, 1980.554
[35] Wermser D. Unsupervise segmentation by use of a texture gradient Proceedings, 7th Int. Con. on Pattern Recognition, 1984,1114
[36] 朱慈盛 天文学教程(中) 北京教育出版社，1987
[37] Dinstein I, Fong AC. Fast discrimination between homogeneous and textured regions, Proceedings, 7th Int. Conf Pattern Recognition, 1984,361
[38] U rage S, Zerubia J, Berthed M. A markovian model for contour grouping. Pattern Recognition, 1995, 28(5): 683-693
[39] Maeda Jetal. Segmentation of natural images using anisotropic diffusion and linking of boundary edges. Pattern Recognition, 1998, 31(12): 1993-1999
[40] Farag A A, Delp E J. Edge linking by sequential search. Pattern Recognition, 1995, 28(5): 611-633
[41] Tsai D M, Wang H J. Segmenting focused objects in complex visual images. Pattern Recognition Letters, 1998, 10:929-940
[2] 章毓晋 图像分割 科学出版社
[3] K．R．Castleman著 朱志刚 林学訚 石定机等译 数字图像处理 电子工业出版社
[4] 田捷 罗希平等 “图像分割方法综述” 模式识别与人工智能 vol．12 No．3
[5] Gonzalez R C, Woods R E. Digital Image Processing. 3rd. ed., Addison-Wesley, 1992
[6] Pal N.R. Pal and S.K.A Review on Image Segmentation Techniques. Pattern Recognition, vol. 26, pp.1277-1294,1993
[7] Zhang, Y.J.A survey of evaluation methods for image segmentation, Pattern Recognition, 1996,29(8):1335-1346
[8] 冯晓毅 周凤歧 “图像分割的一种神经网络实现方法” 计算机工程与科学 Vol．22，No．2，2000
[9] Abutaleb A S. Automation Thresholding of Gray-Level Pictures Using Two-Dimensional Entropy, Computer Vision, Graphics, and Image Processing, 1989, 47:22-32
[10] 华长发 范建平 高传善 吴立德 “基于两维熵阈值的图像分割及其快速算法” 模式识别与人工智能 Vol．13 No．1 March 2000
[11] 谌海新 沈震康 夏放怀 “一种基于目标特征的多门限图像分割方法” 电子学报 No．3 Vol．27 1999
[12] 安玮 徐晖 李宏 孙仲康 “奇异点多门限图像分割算法” 计算机工程与科学 Vol．20，No．3，1998
[13] 杨恬 李德芳 “灰度图像的二维OTSU自动阈值分割研究” 西南师范大学学报(自然科学版) 23卷第6期
[14] M.E. Sieracki, S.E. Reichenbach, and K.L. Webb, Evaluation of Automated Threshold Selection Methods for Accurately Sizing Microscopic Fluoprescent Cells by Image Analysis, Applied and Environmental Microbiology, 55, 11: 2762-2772, Nov. 1989
[15] 张新明 沈兰荪 沈波 “基于特征距离的阈值法及其在眼科图像分割中的应用” 中国图像图形学报 Vol．6(A) No．2 Feb 2001
[16] 刘文萍 吴立德 “图像分割中阈值选取方法比较研究” 模式识别与人工智能 Vol．10 No．3 1997
[17] Theo Pavlidis ans Yuh-Tay Liow, Integrating Region Growing and Edge Detection, IEEE Trans on PAMI, 12, (1990), 225-233
[18] Uchiyama T, Arbib M A. Color image segmentation using competitive learning. IEEE-PAMI, 1994, 16:1197-1206
[19] 柏子游等 “一种彩色图像的色彩分割方法” 模式识别与人工智能12卷第2期
[20] Lim Y W, Lee S U. On the color image segmentation algorithm based on the thresholding and fuzzy C-means techniques. Pattern Recognition, 1990, 23(9): 935-952
[21] 郭国栋 马颂德 “彩色图像分割” 中国图像图形学报 Vol．3 No，11 Nov． 1998 918-921
[22] Ohlander R, Price K, Reddy D R. Picture segmentation using a recursive region splitting method. CGIP, 1978, 8(3): 313-333
[23] 林通 石青云 “一种基于生长的灰度和彩色图像分割方法” 中国图像图形学报 Vol．5(A) No．11 Nov 2000
[24] 粟海华 王睿斌 庄镇泉 “一种基于判决神经网络的彩色图像分色算法” 模式识别与人工智能 1999，12(3)：354—358
[25] 张二虎 刘世昌 胡涛 “一种基于熵的图像分割方法” 微电子学与计算机 1998年第4期
[26] Shanbhag A G. Utilization of information measure as a means of image thresholding. CVGIP-GMIP, 1994,56(5): 414-419
[27] Kunt M et al. Second-generation Image Coding Scheme. Proc IEEE, 1985,73:549-574
[28] Fan J, et al. Automatic Moving Object Extraction Towerd Compact Video Representation. Optical Engineering, 2000, 39:438-452
[29] 刘俊、付敬奇、董新平，“数据融合在目标识别中的应用” 传感器技术 2001年第26卷第6期
[30] B Solaiman, R Debon Pipelier. Information Fusion: Application to data and model fusion for ultrasound image segmentation. IEEE Transactions on Biomedical Engineering, 1999, 46(10): 3-8
[31] 刘等第 傅惠民 殷刚 张少波 “一种新的数据融合方法” 中国安全科学学报 Vol8 No．6，Dec．，1998
[32] 汪培庄、李洪兴 模糊系统理论与模糊计算机 北京科学出版社，1996
[33] 赵锋 赵荣椿 纹理分割及特征提取方法综述 中国体视学与图像分析 Vol．3，
No.4, Dec 1998
[34] De Souza P. Edge based segmentation and texture separation, Proceedings, 5th Int. Con. on Pattern Recognition, 1980.554
[35] Wermser D. Unsupervise segmentation by use of a texture gradient Proceedings, 7th Int. Con. on Pattern Recognition, 1984,1114
[36] 朱慈盛 天文学教程(中) 北京教育出版社，1987
[37] Dinstein I, Fong AC. Fast discrimination between homogeneous and textured regions, Proceedings, 7th Int. Conf Pattern Recognition, 1984,361
[38] U rage S, Zerubia J, Berthed M. A markovian model for contour grouping. Pattern Recognition, 1995, 28(5): 683-693
[39] Maeda Jetal. Segmentation of natural images using anisotropic diffusion and linking of boundary edges. Pattern Recognition, 1998, 31(12): 1993-1999
[40] Farag A A, Delp E J. Edge linking by sequential search. Pattern Recognition, 1995, 28(5): 611-633
[41] Tsai D M, Wang H J. Segmenting focused objects in complex visual images. Pattern Recognition Letters, 1998, 10:929-940