基于两尺度自卷积的直方图仿射不变量提取方法

doi:10.3969/j.issn.1000-436x.2012.12.018

通信学报 ›› 2012, Vol. 33 ›› Issue (12): 140-146.doi: 10.3969/j.issn.1000-436x.2012.12.018

基于两尺度自卷积的直方图仿射不变量提取方法

黄波^1,²,赵晓晖¹,曹昕艳^1,³,赵继印¹,时公涛²

¹ 吉林大学通信工程学院，吉林长春 130012
² 空军装备研究院情报所，北京 100085
³ 长春大学电子信息工程学院，吉林长春 130022

出版日期:2012-12-25 发布日期:2017-07-15
基金资助:
国家自然科学基金资助项目;国家自然科学基金资助项目

Histogram affine invariants extraction method based on two-scale auto-convolution

Bo HUANG^1,²,Xiao-hui ZHAO¹,Xin-yan CAO^1,³,Ji-yin ZHAO¹,Gong-tao SHI²

¹ College of Communication Engineering,Jilin University,Changchun 130012,China
² Information Institute,the Equipment Academy of the Air Force,Beijing 100085,China
³ College of Electronic Information Engineering,Changchun University,Changchun 130022,China

Online:2012-12-25 Published:2017-07-15
Supported by:
The National Natural Science Foundation of China;The National Natural Science Foundation of China

摘要/Abstract

摘要：

摘要：提出了一种新的两尺度自卷积直方图仿射不变量。以图像二维坐标线性变换生成新图像为基点，建立两尺度自卷积变换，进而把新图像密度函数值转化成直方图区间划分，由此推导出新的直方图仿射不变量。另外，重点分析了新不变量的参数选择范围，并详述了其计算步骤。利用经典的“fish”和“coil-100”测试数据库，对比性能优的多尺度自卷积归一化直方图仿射不变量，从时间效率、抗图像变形、抗视角变化等方面，全面验证了所提新仿射不变量的有效性。

关键词: 目标识别, 仿射不变量, 自卷积, 直方图

Abstract:

A two-scale auto-convolution histogram affine invariant was proposed.A two-scale auto-convolution transformation was established based on the new image generated by two-dimensional linear transform of image coordinates.Then the value of the new image density function is transformed into the histogram intervals partition and the proposed histogram affine invariant is obtained.In addition,the possible parameter ranges of the new invari are analyzed and the detailed deduction is given.By using the classic "fish"and "coil-100"test database,the effectiveness of the proposed invariant is verified by comparing to the multi-scale auto-convolution normalized histogram affine invariants from the perspective of time efficiency,anti-image distortion,and anti-viewing angle changes.

Key words: object recognition, affine invariant, auto-convolution, histogram

黄波,赵晓晖,曹昕艳,赵继印,时公涛. 基于两尺度自卷积的直方图仿射不变量提取方法[J]. 通信学报, 2012, 33(12): 140-146.

Bo HUANG,Xiao-hui ZHAO,Xin-yan CAO,Ji-yin ZHAO,Gong-tao SHI. Histogram affine invariants extraction method based on two-scale auto-convolution[J]. Journal on Communications, 2012, 33(12): 140-146.

图/表 8

图1

图2

表1

表2

表3

表4

图3

图4

参考文献 16

[1]	HU M K . Visual pattern recognition by moment invariant[J]. IEEE Trans on Information Theory, 1962,8(2): 179-187.
[2]	TEAGUE M R . Image analysis via the general theory of moments[J]. Journal of Optimal Society of American, 1980,70(8): 920-930.
[3]	ZAHN C T , ROSKIES R Z . Fourier descriptors for plane closed curves[J]. IEEE Transactions on Computers, 1972,21(3): 269-281.
[4]	RAHTU E , HEIKKIL? J . Object classification with multi-scale autoconvolution[A]. Proc of the 17th International Conference on Pattern Recognition[C]. Cambridge, 2004. 37-40.
[5]	RAHTU E , SALO M , HEIKKIL? J . Affine invariant pattern gnition using multi-scale autoconvolution[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005,27(6): 908-918.
[6]	RAHTU E . Generalized affine moment invariants for object recognition[A]. Proc 18th International Conference on Pattern Recognition[C]. Hong Kong,China, 2006. 634-637.
[7]	RAHTU E , SALO M , HEIKKIL? J . Multiscale autoconvolution histograms for affine invariant pattern recognition[A]. Proc 16th British Machine Vision Conference[C]. Edinburgh, 2006. 1039-1048.
[8]	张洁玉, 陈强, 白小晶 ,等. 基于广义典型相关分析的仿射不变特征提取方法[J]. 电子与信息学报, 2009,31(10): 2465-2469. ZHANG J Y , CHEN Q , BAI X J ,et al. Affine invariant feature extraction algorithm based on generalized canonical correlation analysis[J]. Journal of Electronics ＆Information Technology, 2009,31(10): 2465-2469.
[9]	张洁玉, 陈强, 白小晶 ,等. 仿射不变的多尺度自卷积熵提取方法[J]. 计算机辅助设计与图形学报, 2009,21(9): 1328-1332. ZHANG J Y , CHEN Q , BAI X J ,et al. Affine invariant feature extraction based on multi-scale auto- convolution entropy[J]. Journal of Computer-Aided Design ＆Computer Graphics, 2009,21(9): 1328-1332.
[10]	黄波, 赵继印, 郑蕊蕊 ,等. 基于多尺度自卷积归一化直方图的仿射不变量模式识别[J]. 电子学报, 2011,39(1): 64-69. HUNG B , ZHAO J Y , ZHENG R R ,et al. Affine invariant pattern recognition based on multiscale autoconvolution normal zed histograms[J]. Acta Electronica Sinica, 2011,39(1): 64-69.
[11]	谢锦, 蔡自兴, 唐琎 . 基于MSA不变矩的道路导向标线分类[J]. 中国图象图形学报, 2011,16(8): 1418-1423. XIE J , CAI Z X , TANG J . Road oriented marking classifi ion based on MSA invariant moment[J]. Journal of Image and Graph, 2011,16(8): 1418-1423.
[12]	纪华, 韩广良, 陈小林 ,等. 基于MSA直方图的景象匹配算法研究[J]. 激光与红外, 2009,39(7): 788-792. JI H , HAN G L , CHEN X L ,et al. Study on scene matching algorithm based on MSA histogram[J]. Laser ＆Infrared, 2009,39(7): 788-792.
[13]	雷琳, 蔡红苹, 唐涛 ,等. 基于MSA特征的遥感图像多目标关联算法[J]. 遥感学报, 2008,12(4): 586-592. LEI L , CAI H P , TANG T ,et al. A MSA feature-based multiple targets association algorithm in remote sensing images[J]. Journal of Remote Sensing, 2008,12(4): 586-592.
[14]	KANNALA J , RAHTU E , HEIKKIL? J . Affine registration with multi-scale autoconvolution[A]. Proc International Conference on Image Processing[C]. Genoa, 2005. 1064-1067.
[15]	ALEXANDER K , MARIA P . The trace transform and its application[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005,23(8): 811-828.
[16]	Columbia University . Coil-100 image database[EB/OL]. , 2010.

算法名称	高斯噪声方差v										平均值
算法名称	0.01	0.02	0.03	0.04	0.05	0.06	0.07	0.08	0.09	0.1	平均值
1-不变量	99.3	100	100	98.7	99.3	98.7	98.0	94.0	92.7	92.0	97.3
TSAH3	100	100	100	100	100	99.3	100	98.7	97.3	96.7	99.2
TSAH20	100	100	100	100	99.3	99.3	100	100	98.7	100	99.7

算法名称	遮挡尺寸d						平均值
算法名称	5	10	15	20	25	30	平均值
1-不变量	99.3	98.0	94.0	82.7	68.0	58.7	83.4
TSAH3	100	98.7	96.7	92.7	82.0	67.3	89.6
TSAH20	100	100	95.3	90.0	78.0	68.0	88.6

算法名称	亮度变化c							平均值
算法名称	-27	-18	-9	0	+9	+18	+27	平均值
1-不变量	87.3	98.7	100	100	99.3	99.3	93.3	96.8
TSAH3	96.7	100	100	100	100	100	98.7	99.3
TSAH20	99.3	100	100	100	100	99.3	93.3	99.8

算法名称	亮度值变化斜率s							平均值
算法名称	0.03	0.06	0.09	0.12	0.15	0.18	0.21	平均值
1-不变量	100	100	100	96.7	95.3	92.0	86.7	95.8
TSAH3	100	100	100	99.3	99.3	96.7	95.3	98.8
TSAH20	100	100	99.3	97.3	95.3	94.0	86.7	96.1

基于两尺度自卷积的直方图仿射不变量提取方法

Histogram affine invariants extraction method based on two-scale auto-convolution

在线阅读

PDF下载

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 16

相关文章 9

Metrics

推荐阅读 0

[1]	李红光, 郭英, 眭萍, 齐子森. 基于时频能量谱纹理特征的跳频调制方式识别[J]. 通信学报, 2019, 40(10): 20-29.
[2]	卫颜俊,冯博琴,伍卫国. 基于多项式一致逼近的多阈值图像分割算法[J]. 通信学报, 2016, 37(10): 56-64.
[3]	余晓东,雷英杰,宋亚飞,岳韶华,胡军红. 基于直觉模糊核匹配追踪集成的目标识别方法[J]. 通信学报, 2015, 36(10): 165-171.
[4]	邓小鸿,陈志刚,梁涤青,毛伊敏. 分区域的医学图像高容量无损信息隐藏方法[J]. 通信学报, 2015, 36(1): 189-198.
[5]	雷阳,孔韦韦,雷英杰. 基于直觉模糊c均值聚类核匹配追踪的弹道中段目标识别方法[J]. 通信学报, 2012, 33(11): 136-143.
[6]	陈文飞,廖斌,许雪峰,黄志勇,董文永. 基于Piecewise直方图均衡化的图像增强方法[J]. 通信学报, 2011, 32(9): 153-160.
[7]	孙欣尧,王雪,王晟. 无线传感网络协同概率多模识别方法[J]. 通信学报, 2011, 32(6): 141-147.
[8]	高海英. 针对非均匀DCT的量化索引调制隐写算法的分析[J]. 通信学报, 2011, 32(11A): 209-212.
[9]	吴一全,潘喆,吴文怡. 二维直方图区域斜分阈值分割及快速递推算法[J]. 通信学报, 2008, 29(4): 77-84.