易良玲,
许海霞,
张莹
1.湘潭大学信息工程学院 ??湘潭 ??411105
2.机器人视觉感知与控制国家工程实验室 ??长沙 ??410012
基金项目:国家自然科学基金(61602397),湖南省自然科学基金(2017JJ2251, 2017JJ3315),湖南省重点学科建设项目
详细信息
作者简介:张东波:男,1973年生,博士,教授,研究方向为计算机视觉、模式识别
易良玲:女,1993年生,硕士,研究方向为计算机视觉、机器学习
许海霞:女,1979年生,博士,副教授,研究方向为机器视觉、模式识别
张莹:男,1972年生,博士,副教授,研究方向为机器人控制、模式识别、高维可视化处理
通讯作者:张东波 zhadonbo@163.com
中图分类号:TP391.4计量
文章访问数:1123
HTML全文浏览量:355
PDF下载量:51
被引次数:0
出版历程
收稿日期:2018-05-25
修回日期:2018-12-18
网络出版日期:2018-12-25
刊出日期:2019-04-01
Multi-scale Local Region Structure Dominant Binary Pattern Learning for Image Representation
Dongbo ZHANG,,Liangling YI,
Haixia XU,
Ying ZHANG
1. College of Information Engineering, Xiangtan University, Xiangtan 411105, China
2. Robot Visual Perception & Control Technology National Engineering Laboratory, Changsha 410012, China
Funds:The National Natural Science Foundation of China (61602397), The Natural Science Foundation of Hunan Province (2017JJ2251, 2017JJ3315), The Key Discipline Construction Project of Hunan Province
摘要
摘要:通过零均值化的微观结构模式二值化(ZMPB)处理,该文提出一种立足于局部图像多尺度结构二值模式提取的图像表示方法。该方法能够表达图像中可能出现的各种具有视觉意义的重要模式结构,同时通过主导二值模式学习模型,可以获得适应于图像数据集的主导特征模式子集,在特征鲁棒性、鉴别力和表达能力上达到优异性能,同时可以有效降低特征编码的维度,提高算法的执行速度。实验结果表明该算法性能优异,具有很强的鉴别能力和鲁棒性,优于传统LBP和GIMMRP方法,和很多最新算法结果相比,也具有竞争优势。
关键词:目标识别/
零均值化的微观结构模式二值化/
主导二值模式学习/
局部结构
Abstract:By means of Zero-mean Microstructure Pattern Binarization (ZMPB), an image representation method based on image local microstructure binary pattern extraction is proposed. The method can express all the important patterns with visual meaning that may occur in the image. Moreover, through the dominant binary pattern learning model, the dominant feature pattern set adapted to the different data sets is obtained, which not noly achieves excellent ability in feature robustness, discriminative and representation, but also can greatly reduce the dimension of feature coding and improve the execution speed of the algorithm. The experimental results show that the proposed method has strong discriminative power and outperformes the traditional LBP and GIMMRP methods. Compared with many recent algorithms, the proposed method also presents a competitive advantage.
Key words:Object recognition/
Zero-mean Microstructure Pattern Binarization (ZMPB)/
Dominant binary pattern learning/
Local region structure
PDF全文下载地址:
https://jeit.ac.cn/article/exportPdf?id=28326572-a15b-4dc7-b6ef-29558e0b1cc8