改进的基于弦切变换的目标检测方法

何莲; 蔡敬菊; 张启衡

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

改进的基于弦切变换的目标检测方法

何莲, 蔡敬菊, 张启衡

文章导航 > 红外与激光工程 > 2012 > 41(1): 239-247

何莲, 蔡敬菊, 张启衡. 改进的基于弦切变换的目标检测方法[J]. 红外与激光工程, 2012, 41(1): 239-247.

引用本文:

何莲, 蔡敬菊, 张启衡. 改进的基于弦切变换的目标检测方法[J]. 红外与激光工程, 2012, 41(1): 239-247.

HE Lian, CAI Jing-Ju, ZHANG Qi-Heng. Improved method of object detection based on chord-tangent transformation[J]. Infrared and Laser Engineering, 2012, 41(1): 239-247.

Citation:

HE Lian, CAI Jing-Ju, ZHANG Qi-Heng. Improved method of object detection based on chord-tangent transformation[J]. Infrared and Laser Engineering, 2012, 41(1): 239-247.

改进的基于弦切变换的目标检测方法

1.
中国科学院光电技术研究所,四川成都 610209;
2.
中国科学院研究生院,北京 100049

Improved method of object detection based on chord-tangent transformation

1.
Institute of Optics and Electronics,Chinese Academy of Sciences,Chengdu 610209,China;
2.
Graduate University of the Chinese Academy of Sciences,Beijing 100049,China

摘要: 在计算机视觉中形状是目标识别和检测的重要特征，针对目前许多基于形状特征的检测方法信息不够丰富，容易受边缘缺损变形等方面的影响，不具有局部特性，尤其是在许多复杂环境下很难实现对目标的正确检测等不足，提出了一种基于弦切变换理论在有限的目标边缘点信息基础上提取几何形状特征及相应的目标检测方法。该特征具有平移、旋转以及缩放不变性，基于此特征进行的目标检测能有效的得到目标的中心位置以及相关的二维运动参数，即使在一些复杂环境以及目标边缘部分失真或缺损的情况下也具有一定的鲁棒性。但由于边缘本身容易受到图像质量、对比度以及量化误差等影响，从而影响算法的精度。因此，文中通过融合丰富的灰度信息，使表征目标的特征更加丰富和完善，在形状和灰度的共同约束下提高检测的正确率和精确性。通过对多组图像序列进行仿真实验，结果表明了算法的有效性，及其在准确性和精确性上的提高，改进后待测目标与模板之间的匹配率可达90%以上。
- 弦切变换 /
- 几何特征 /
- 特征匹配 /
- 参数估计 /
- 目标检测
Abstract: Shape is a kind of important feature to detect or recognize objects in computer vision. However, some disadvantages still exist in many methods based on shape feature at present, such as having no abundant information, easily affected by the default and distortion of edge, or having no local property. Especially, it is very difficult to detect object correctly under some complicated environments. In order to overcome these defects, a kind of shape feature extracted by using the chord-tangent transformation and corresponding algorithm of target detection was presented in this paper, which was obtained according to the finite edge point. This geometric feature had the invariant character of translation, rotation and scale. Some important parameters about movement including the location of object could be obtained through this method, even under some complicated environments. However,because the edge extraction is usually affected by the quality of image, contrast between the object and background and quantized error, the precision can be decreased. Therefore, some gray information was added to the feature in order to improve the algorithm. Finally, experimental results indicate the effectivity of the algorithm. The matching rate between the object and model can be extended to more than 90% after improving.
- chord-tangent transformation /
- geometric feature /
- feature matching /
- parameter evaluation /
- object detection

[1]
[2]	Dong-Gyu Sim, Oh-Kyu Kwon, Rae-Hong Park. Object matching algorithms using robust Hausdorff distance measures[J]. IEEE Transaction on Image Processing, 1999, 8(3): 425-429.
[3]
[4]	Ballard D H. Generalizing the Hough transform to detect arbitrary shapes [J]. Pattern Recognition, 1981, 13: 111-122.
[5]	Ye Zhouhai, Chen Fumin. An improving generalized Hough transform[J]. Microcomputer Applications, 2006, 22(6): 23-25. (in Chinese)
[6]
[7]	Zahn C T, Roskies R Z. Fourier descriptors for plane closed curves[J]. IEEE Trans on Computers, 1972, 21: 269-281.
[8]	叶州海, 陈福民. 一种广义霍夫变换的改进[J]. 微型电脑应用, 2006. 22(6): 23-25.
[9]
[10]	Selvage J E, Chenoweth D L, Gold V E. Geometric feature extraction using the chord transformation [C]//Proceedings of Aerospace Applications Conference, 1996, 4: 399-405.
[11]	Ayache N, Faugeras O D. HYPER: A new approach for the recognition and positioning of two-dimensional objects [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1986, 8: 44-54.
[12]
[13]
[14]	Dufresne T E, Dhawan A P. Chord-tangent transformation for object recognition[J]. Pattern Recognition, 1995, 28: 1321-1332.
[15]	Kass M, Witkin A, Terzopouls D. Snake: Active contour models[J]. International Journal of Computer Vision, 1987, 1(4): 321-331.
[16]
[17]	Xu C, Prince J L. Snake, shapes, and gradient vector flow[J]. IEEE Transaction on Image Processing, 1998, 7(1): 359-369.
[18]
[19]	Evans A N, Liu X U. A morphological gradient approach to color edge detection[J]. Image Processing, 2006, 15(6): 1454-1463.
[20]
[21]	Xia Liangzheng, Li Jiuxian. Digital Image Processing[M]. Nanjing: Southwast University Press, 2005. (in Chinese)
[22]
[23]
[24]	Canny J. A computational approach to edge detection[J]. IEEE Transactions on Pattern Analysis and MachineIntelligence, 1986, 8(6): 679-698.
[25]
[26]	Wu Y Y, Liu Q. Novel object contour extraction: multi-scale edge detection based on bilateral filtering[J]. Application Research of Computers, 2002, 24: 396-398. (in chinese)
[27]	夏良正, 李久贤. 数字图像处理[M]. 南京: 东南大学出版社, 2005.
[28]	Tomasi C, Manduchi R. Bilateral filtering for gray and color images[C]//Proc of the 6th International Conference on Computer Vision, 1998: 839-846.
[29]
[30]	Liu Jinfeng. Study on fast omage template matching algorithm[D]. Changsha: Central South University of China, 2007, 5. (in Chinese)
[31]
[32]
[33]	吴莹莹, 刘桥. 新颖的物体轮廓提取: 基于双边滤波的多级边缘检测[J]. 计算机应用研究, 2002, 24(12): 396-398.
[34]
[35]
[36]
[37]	刘锦峰, 图像模板匹配快速算法研究[D]. 长沙: 中南大学, 2007.

[1]	周志洪, 陈秀真, 马进, 夏正敏. 烟花算法在SAR图像属性散射中心参数估计中的应用 . 红外与激光工程, 2022, 51(8): 20210581-1-20210581-7. doi: 10.3788/IRLA20210581
[2]	王中军, 晁艳锋. 采用SURF特征和局部互相关信息的图像配准算法 . 红外与激光工程, 2022, 51(6): 20210950-1-20210950-6. doi: 10.3788/IRLA20210950
[3]	孙鹏, 于跃, 陈嘉欣, 秦翰林. 基于深度空时域特征融合的高动态空中多形态目标检测方法(特邀) . 红外与激光工程, 2022, 51(4): 20220167-1-20220167-8. doi: 10.3788/IRLA20220167
[4]	王晓艳, 徐高魁. 基于立体视觉与特征匹配的点云目标识别算法 . 红外与激光工程, 2022, 51(9): 20210596-1-20210596-7. doi: 10.3788/IRLA20210596
[5]	陈丛. 基于狼群算法的SAR图像属性散射中心参数估计 . 红外与激光工程, 2021, 50(4): 20200409-1-20200409-7. doi: 10.3788/IRLA20200409
[6]	陈明, 赵连飞, 苑立民, 徐峰, 韩默. 基于特征选择YOLOv3网络的红外图像绝缘子检测方法 . 红外与激光工程, 2020, 49(S2): 20200401-20200401. doi: 10.3788/IRLA20200401
[7]	王保平, 马健钧, 张研, 方阳. 基于最小能量准则的参数估计圆周SAR三维成像方法 . 红外与激光工程, 2019, 48(7): 726001-0726001(7). doi: 10.3788/IRLA201948.0726001
[8]	南天章, 耿建君, 陈旭, 陈颖. 基于邻域特征的红外低慢小目标检测 . 红外与激光工程, 2019, 48(S1): 174-180. doi: 10.3788/IRLA201948.S128002
[9]	程全, 樊宇, 刘玉春, 程朋. 分块投影匹配的运动目标检测方法 . 红外与激光工程, 2018, 47(10): 1026004-1026004(4). doi: 10.3788/IRLA201847.1026004
[10]	郭力仁, 胡以华, 王云鹏. 相干激光探测中微动参数估计的克拉美-罗界 . 红外与激光工程, 2017, 46(7): 706002-0706002(7). doi: 10.3788/IRLA201746.0706002
[11]	蒋立辉, 章典, 陈星, 熊兴隆. 基于谱估计的Fabry-Perot标准具透过率曲线参数估计方法 . 红外与激光工程, 2016, 45(9): 906006-0906006(8). doi: 10.3788/IRLA201645.0906006
[12]	胡梦婕, 魏振忠, 张广军. 基于对象性测度估计和霍夫森林的目标检测方法 . 红外与激光工程, 2015, 44(6): 1936-1941.
[13]	张采芳, 田岩, 李江. 基于几何特征分析的路面裂缝分类算法研究 . 红外与激光工程, 2015, 44(4): 1359-1364.
[14]	赵爱罡, 王宏力, 杨小冈, 陆敬辉, 何星. 融合几何特征的压缩感知SIFT描述子 . 红外与激光工程, 2015, 44(3): 1085-1091.
[15]	韩艳丽, 刘峰. 基于三角形匹配的空间小目标检测算法 . 红外与激光工程, 2014, 43(9): 3134-3140.
[16]	杨晟, 李学军, 朱诗兵, 刘涛. 抗仿射形变异构金字塔复合描述点特征匹配算法 . 红外与激光工程, 2014, 43(7): 2387-2392.
[17]	谢易辰, 陈健, 闫镔, 童莉, 曾磊, 崔明明. 三维特征点距离特征集合求交匹配算法 . 红外与激光工程, 2014, 43(8): 2728-2732.
[18]	陈骁, 石志广, 杨卫平, 凌建国. 利用三维点云的圆锥状弹头目标参数估计 . 红外与激光工程, 2014, 43(10): 3480-3485.
[19]	邹盼, 刘晖, 张闻文, 陈钱, 顾国华, 张连东. 利用期望最大化算法的EMCCD噪声分布模型的参数估计 . 红外与激光工程, 2013, 42(1): 268-272.
[20]	杨亚威, 李俊山, 杨威, 赵方舟. 利用稀疏化生物视觉特征的多类多视角目标检测方法 . 红外与激光工程, 2012, 41(1): 267-272.

点击查看大图

计量

文章访问数: 450
HTML全文浏览量: 69
PDF下载量: 83
被引次数: 0

改进的基于弦切变换的目标检测方法

1. 中国科学院光电技术研究所,四川成都 610209;

2. 中国科学院研究生院,北京 100049

刊出日期: 2012-01-25

关键词:

摘要: 在计算机视觉中形状是目标识别和检测的重要特征，针对目前许多基于形状特征的检测方法信息不够丰富，容易受边缘缺损变形等方面的影响，不具有局部特性，尤其是在许多复杂环境下很难实现对目标的正确检测等不足，提出了一种基于弦切变换理论在有限的目标边缘点信息基础上提取几何形状特征及相应的目标检测方法。该特征具有平移、旋转以及缩放不变性，基于此特征进行的目标检测能有效的得到目标的中心位置以及相关的二维运动参数，即使在一些复杂环境以及目标边缘部分失真或缺损的情况下也具有一定的鲁棒性。但由于边缘本身容易受到图像质量、对比度以及量化误差等影响，从而影响算法的精度。因此，文中通过融合丰富的灰度信息，使表征目标的特征更加丰富和完善，在形状和灰度的共同约束下提高检测的正确率和精确性。通过对多组图像序列进行仿真实验，结果表明了算法的有效性，及其在准确性和精确性上的提高，改进后待测目标与模板之间的匹配率可达90%以上。

English Abstract

Improved method of object detection based on chord-tangent transformation

1. Institute of Optics and Electronics,Chinese Academy of Sciences,Chengdu 610209,China;

2. Graduate University of the Chinese Academy of Sciences,Beijing 100049,China

Publish Date: 2012-01-25

Keywords:

Abstract: Shape is a kind of important feature to detect or recognize objects in computer vision. However, some disadvantages still exist in many methods based on shape feature at present, such as having no abundant information, easily affected by the default and distortion of edge, or having no local property. Especially, it is very difficult to detect object correctly under some complicated environments. In order to overcome these defects, a kind of shape feature extracted by using the chord-tangent transformation and corresponding algorithm of target detection was presented in this paper, which was obtained according to the finite edge point. This geometric feature had the invariant character of translation, rotation and scale. Some important parameters about movement including the location of object could be obtained through this method, even under some complicated environments. However,because the edge extraction is usually affected by the quality of image, contrast between the object and background and quantized error, the precision can be decreased. Therefore, some gray information was added to the feature in order to improve the algorithm. Finally, experimental results indicate the effectivity of the algorithm. The matching rate between the object and model can be extended to more than 90% after improving.