首页 > 论文 > 激光与光电子学进展 > 57卷 > 8期(pp:81105--1)

基于多特征的彩色唐卡修复图像无参考质量评价方法

No-Reference Quality Assessment Method for Inpainting Thangka Image Based on Multiple Features

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

为了解决单一特征反映不同修复方法效果差异的局限性,结合唐卡图像结构特征和色彩特征,提出一种基于多特征的无参考彩色唐卡修复图像质量评价方法。首先,该算法利用唐卡图像纹理丰富的特性,选取高斯差分算子提取图像线描图,并结合唐卡对称特性获取图像结构特征;然后依据简单线性迭代聚类分割后不同超像素区域间的色彩熵变化差异,提取唐卡图像色彩特征;最后考虑到多尺度特征更加符合人眼视觉特性,将拉普拉斯金字塔分解后的多元特征输入自适应增强神经网络进行训练,预测得出修复图像质量客观评价分数。实验结果表明,该方法利用唐卡图像丰富的结构和色彩特征,较好地获得与主观评价一致的分数,其斯皮尔曼相关系数与皮尔逊相关系数均达到0.94以上。

Abstract

This paper proposes a new no-reference quality assessment method for inpainting Thangka image based on multiple features, and combining the structural and color characteristic of Thangka images to solve the problem that a single feature is confined to reflects the difference in the effect of restoration methods. The proposed algorithm not only uses the rich texture of Thangka images but also selects Gaussian difference operator to extract the line drawing of target image, and combining symmetry characteristic of Thangka image to obtain structural features. Secondly, the color features of Thangka images are extracted according to the difference of color entropy between each superpixels after simple linear iterative cluster segmentation. Finally, considering that the multi-scale features are more consistent with the human visual characteristics, the decomposed image features are input into the adaptive neural network for training, and the objective evaluation score of image quality is predicted. The experimental results show that this method can obtain the scores which is consistent with the subjective evaluation by utilizing the structure and color characteristics of Thangka images, and its Spearman correlation coefficient and Pearson correlation coefficient are both above 0.94.

Newport宣传-MKS新实验室计划
补充资料

中图分类号:TP391.415

DOI:10.3788/LOP57.081105

所属栏目:成像系统

基金项目:国家民委创新团队计划资助、国家自然科学基金、西北民族大学中央高校基本科研业务费专项资金资助研究生项目;

收稿日期:2019-07-19

修改稿日期:2019-09-24

网络出版日期:2020-04-01

作者单位    点击查看

叶雨琪:西北民族大学数学与计算机科学学院, 甘肃 兰州 730030
胡文瑾:西北民族大学数学与计算机科学学院, 甘肃 兰州 730030西北民族大学中国民族语言文字信息技术教育部重点实验室, 甘肃 兰州 730030

联系人作者:胡文瑾(wenjin_zhm@126.com)

备注:国家民委创新团队计划资助、国家自然科学基金、西北民族大学中央高校基本科研业务费专项资金资助研究生项目;

【1】Li N, Zhou P B, Geng G H, et al. Quality evaluation method for color restoration image [J]. Journal of Computer Applications. 2016, 36(6): 1673-1676.
李娜, 周蓬勃, 耿国华, 等. 色彩复原图像的质量评价方法 [J]. 计算机应用. 2016, 36(6): 1673-1676.

【2】Fang X J. Scientific protection and restoration of thangka[C]∥ Proceedings of the seventh annual conference of China association for the protection of cultural relics. Beijing: Science Press, 2012, 191-198.
方小济. 科学出版社: 唐卡的科学化保护与修复研究综述[C]∥中国文物保护技术协会第七次学术年会论文集. 北京, 2012, 191-198.

【3】Tang S L, Liu X J, Liu Z Q. Summary of Thangka image inpainting [J]. Software Guide. 2018, 17(9): 4-8.
唐守亮, 刘晓静, 刘志强. 唐卡图像修复综述 [J]. 软件导刊. 2018, 17(9): 4-8.

【4】Wang W L, Tang S X. Tibetan Tangka image inpainting in intricate disrepaired region [J]. Computer Engineering and Design. 2007, 28(3): 703-705.
王维兰, 唐仕喜. 唐卡图像复杂破损区域的修复 [J]. 计算机工程与设计. 2007, 28(3): 703-705.

【5】Wang Y B, Bai Y P. Image restoration algorithm based on sample block [J]. Fire Control & Command Control. 2017, 42(3): 55-58, 63.
王一卜, 白艳萍. 基于样块改进的图像修复算法 [J]. 火力与指挥控制. 2017, 42(3): 55-58, 63.

【6】Criminisi A, Pérez P, Toyama K. Object removal by exemplar-based inpainting . [C]∥2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings., June 18-20, 2003, Madison, WI, USA. New York: IEEE. 2003, 7770168.

【7】Lu X B. Research on Thangka image inpainting of comprehensive partial differential equations [D]. Lanzhou: Northwest Minzu University. 2011.
卢小宝. 综合偏微分方程的唐卡图像修复研究 [D]. 兰州: 西北民族大学. 2011.

【8】Zhou N, Zhu Z Z. Criminisi image inpainting algorithm based on rough data-deduction [J]. Laser & Optoelectronics Progress. 2019, 56(2): 021005.
周宁, 朱昭昭. 基于粗糙数据推理的Criminisi图像修复算法 [J]. 激光与光电子学进展. 2019, 56(2): 021005.

【9】Du S S, Han C. An improved image inpainting algorithm based on total variation model [J]. Laser & Optoelectronics Progress. 2018, 55(7): 071005.
杜闪闪, 韩超. 基于总变分模型的改进图像修复算法 [J]. 激光与光电子学进展. 2018, 55(7): 071005.

【10】Dang T T, Beghdadi A, Larabi M C. Inpainted image quality assessment . [C]∥European Workshop on Visual Information Processing (EUVIP), June 10-12, 2013, Paris, France. New York: IEEE. 2013, 13838803.

【11】Qureshi M A, Deriche M, Beghdadi A, et al. A critical survey of state-of-the-art image inpainting quality assessment metrics [J]. Journal of Visual Communication and Image Representation. 2017, 49: 177-191.

【12】Vijay Venkatesh M. Cheung S C S. Eye tracking based perceptual image inpainting quality analysis . [C]∥2010 IEEE International Conference on Image Processing, September 26-29, 2010, Hong Kong, China. New York: IEEE. 2010, 1109-1112.

【13】Chen Y D, Li C F, Sang Q B. Quality assessment without reference images based on convolution neural network and deep forest [J]. Laser & Optoelectronics Progress. 2019, 56(11): 111003.
陈寅栋, 李朝锋, 桑庆兵. 卷积神经网络结合深度森林的无参考图像质量评价 [J]. 激光与光电子学进展. 2019, 56(11): 111003.

【14】Ardis P A, Singhal A. Visual salience metrics for image inpainting [J]. Proceedings of SPIE. 2009, 7257: 72571W.

【15】Hu W J, Ye Y Q, Meng J H, et al. No reference quality assessment for Thangka color image based on superpixel [J]. Journal of Visual Communication and Image Representation. 2019, 59: 407-414.

【16】Hu W J, Ye Y Q, Zeng F L, et al. A new method of Thangka image inpainting quality assessment [J]. Journal of Visual Communication and Image Representation. 2019, 59: 292-299.

【17】Birch P, Mitra B, Bangalore N M, et al. Approximate bandpass and frequency response models of the difference of Gaussian filter [J]. Optics Communications. 2010, 283(24): 4942-4948.

【18】Xu L, Chen Q, Wang Q. Application of color entropy to image quality assessment [J]. Journal of Image and Graphics. 2015, 20(12): 1583-1592.
徐琳, 陈强, 汪青. 色彩熵在图像质量评价中的应用 [J]. 中国图象图形学报. 2015, 20(12): 1583-1592.

【19】Peng Y J, Wang J J, Wang Y H. Improved image fusion method based on Laplacian pyramid transform [J]. Software Guide. 2016, 15(1): 167-170.
彭延军, 王瑾瑾, 王元红. 基于拉普拉斯金字塔改进的图像融合方法 [J]. 软件导刊. 2016, 15(1): 167-170.

【20】Schapire R E, Singer Y. Improved boosting algorithms using confidence-rated predictions [J]. Machine Learning. 1999, 37(3): 297-336.

【21】Bao L L. On line drawing art of Tibetan Thangka [J]. Journal of Kangding Nationality Teachers College. 2009, 18(1): 11-16.
鲍丽丽. 藏传唐卡线描艺术浅论 [J]. 康定民族师范高等专科学校学报. 2009, 18(1): 11-16.

【22】Xia M, Zhang R, Peng R, et al. Generation of Thangka relief from line drawings [J]. Scientia Sinica Informationis. 2019, 49(2): 216-228.
夏梦, 张荣, 彭韧, 等. 用线描图生成唐卡浮雕 [J]. 中国科学:信息科学. 2019, 49(2): 216-228.

【23】Li Y. A stylistic method using extended DOG operator [D]. Kunming: Yunnan University. 2014.
李阳. 基于高斯差分的风格化方法 [D]. 昆明: 云南大学. 2014.

【24】Qiu N. Analysis on Tibetan Thangka characteristic art [D]. Beijing: Capital Normal University. 2005.
丘宁. 藏族唐卡艺术特色分析 [D]. 北京: 首都师范大学. 2005.

【25】Shen W, Cheng X Q, Zeng D. Symmetry detection in natural images via edge feature learning [J]. Journal of Shanghai University(Natural Science Edition). 2014, 20(6): 715-725.
沈为, 程小青, 曾丹. 基于边缘特征学习的自然图像对称轴检测 [J]. 上海大学学报(自然科学版). 2014, 20(6): 715-725.

【26】Shen L L, Hang N. No-reference image qua1ity assessment using joint mu1tip1e edge detection [J]. Chinese Journal of Engineering. 2018, 40(8): 996-1004.
沈丽丽, 杭宁. 联合多种边缘检测算子的无参考质量评价算法 [J]. 工程科学学报. 2018, 40(8): 996-1004.

引用该论文

Ye Yuqi,Hu Wenjin. No-Reference Quality Assessment Method for Inpainting Thangka Image Based on Multiple Features[J]. Laser & Optoelectronics Progress, 2020, 57(8): 081105

叶雨琪,胡文瑾. 基于多特征的彩色唐卡修复图像无参考质量评价方法[J]. 激光与光电子学进展, 2020, 57(8): 081105

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF