一种改进的轮廓提取评估算法

吉林大学学报(工学版) ›› 2011, Vol. 41 ›› Issue (增刊1): 179-182.

一种改进的轮廓提取评估算法

赵宏伟^1,2, 刘静¹, 陈华程³, 崔弘睿¹

1. 吉林大学计算机科学与技术学院, 长春 130012;
2. 吉林大学符号计算与知识工程教育部重点实验室, 长春 130012;
3. 吉林省交通规费管理局, 长春 130061

收稿日期:2011-03-10 出版日期:2011-09-01 发布日期:2011-09-01
通讯作者: 陈华程(1967 ),男,高级工程师.研究方向:信息系统.E-mail:cchtgs@yahoo.com.cn. E-mail:cchtgs@yahoo.comcn
作者简介:赵宏伟(1962 ),男,教授.博士生导师.研究方向:智能信息系统与眠人式技术.E-mail:zhaohw@jlu.edu.cn.
基金资助:
吉林省自然科学基金项目(20101504);吉林省教育厅科学基金项目(2009604)

Improved evaluation algorithm of contour detection

ZHAO Hong-wei^1,2, LIU Jing¹, CHEN Hua-cheng³, CUI Hong-rui¹

1. College of Computer Science and Technology, Jilin University, Changchun 130012, China;
2. Key Laboratory of Symbolic Computation and Knowledge Engineering for Ministry of Education, Jilin University, Changchun 130012, China;
3. Scientific Information Bureau, Jilin Transportion Chares Adminstration, Changchun 130061, China

Received:2011-03-10 Online:2011-09-01 Published:2011-09-01

摘要/Abstract

摘要：

针对带有Ground Truth图的自然图像轮廓评估算法的轮廓提取准确率低并且不精确的问题,本文提出了一种邻域匹配算法,并且与原算法相结合,再将评估的准确率公式进行修改,得到了本文的轮廓提取评估算法。通过对Canny算子处理的轮廓提取图进行评估,并与原算法相比可知,本文算法提高了评估的轮廓提取准确率。

关键词: 图像处理技术, Ground Truth图, 自然图像, 轮廓提取评估算法

Abstract:

For the low accuracy and imprecise of contour detection evaluation algorithm for a natural image with Ground Truth,we proposed a neighborhood matching algorithm,which had been combined with the original algorithm,and then modified the accuracy of the formula,at last we got the contour detection evaluation algorithm of this paper.Through assessing the contour detection images processed by Canny operator,compared with the original algorithm,The algorithm improved the accuracy value of assessment of contour detection.

Key words: image processing, Ground Truth, nature image, contour detection evaluation algorithm

中图分类号:

TP391

赵宏伟, 刘静, 陈华程, 崔弘睿. 一种改进的轮廓提取评估算法[J]. 吉林大学学报(工学版), 2011, 41(增刊1): 179-182.

ZHAO Hong-wei, LIU Jing, CHEN Hua-cheng, CUI Hong-rui. Improved evaluation algorithm of contour detection[J]. 吉林大学学报(工学版), 2011, 41(增刊1): 179-182.

参考文献

[1] Ando S.Image field categorization and edge/cornerd etection from gradient covariance
[J].IEEE Transo n Pattern Anal Machine Intell,2000,22(2):179-190.

[2] Bowyer K,Kranenburg C,Dougherty A.Edge detector evaluation using empirical ROC curves
[J].Comput Vis Image Understand,2001,84(1):77-103.

[3] Catanzaro Bryan,Su Bor-Yiing,Sundaram Narayanan,et al.Efficient,high-quality image contour detection
[J].Computer Vision,2009,2381-2388.

[4] Wang Hong-zhi,John Oliensis.Generalizing edge detection to contour detection for image segmentation
[J].Computer Vision and Image Understanding, 2010,114(7):731-744.

[5] Pablo A rbeláez,Michael Maire,Charless Fowlkes,eta l.Contour detection and hierarchical image segmentation
[J].IEEE Transaction on Pattern Analysis andM achine Intelligence,2011,33(5):898-916.

[6] Zhou Y T,Venkateshwar V,Chellappa R.Edge detection and linear feature extraction using a 2-D random field model
[J].IEEE Trans on Pattern AnalM achine Intell,1989,11:84-95.

[7] Kitchen L,Rosenfeld A.Edge evaluation using localedge coherence
[J].IEEE Trans on Syst,Man,Cy-b ern,1981,11(9):597-605.

[8] Shin M,Goldgof D,Bowyer K W.An objective comparison methodology of edge detection algorithms fors tructure from motion task
[C] ∥New York:Empirical Evaluation Techniques in Computer Vision,I EEE Press,1998:235-254.

[9] Shin M C,Bowyer K W,Goldgof D B.Comparison of edge detectors through use in an object recognitiont ask
[J].Comput Vision Image Understand,2001,84 (1):160-178.

[10] Cosmin Grigorescu,Nicolai Petkov,Michel A Westenberg.Improved contour detection by non-classicalr eceptive field inhibition biologically motivated computer vision
[J].Lecture Notes in Computer Science, 2010,2525:31-57.

[11] Grigorescu C,Petkov N,Westenberg M A.Contourd etection based on nonclassical receptive field inhibition
[J].IEEE Trans on Image Processing,2003,12 (7):729-739.

[12] Bowyer K,Kranenburg C,Dougherty A.Edge detector evaluationusing empirical ROC curves
[J].Comput Vision Image Understand,2001,84(1):77-103.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed