Journal of Computer Applications ›› 2015, Vol. 35 ›› Issue (11): 3194-3197.DOI: 10.11772/j.issn.1001-9081.2015.11.3194

• DPCS 2015 Paper • Previous Articles     Next Articles

Texture image restoration model based on combined subdivision

WAN Jinliang, WANG Jian   

  1. College of Computer and Information Engineering, Henan University of Economics and Law, Zhengzhou Henan 450002, China
  • Received:2015-07-08 Revised:2015-07-08 Published:2015-11-13

基于融合细分的纹理图像重构模型

万金梁, 王健   

  1. 河南财经政法大学 计算机与信息工程学院, 郑州 450002
  • 通讯作者: 万金梁(1980-), 男, 河南郑州人, 讲师, 博士, CCF会员, 主要研究方向:数字图像处理、模式识别.
  • 作者简介:王健(1981-), 男, 河南安阳人, 讲师, 博士, 主要研究方向:云计算、智能信息处理.
  • 基金资助:
    国家自然科学基金资助项目(61201236);河南省教育厅科学技术研究重点项目(13A520034,13B510905).

Abstract: A texture image restoration model based on combined subdivision was proposed in this paper, which can solve the problems of piecewise iterative curve fitting, especially the discontinuous contour and the size error of reconstruction regions. Firstly, segmentation regions of the original image were extracted. The feature vector of region shape was got by using contour tracing and downsampling. Then the region contour curve was reconstructed by combining ternary approximating and interpolating subdivision scheme. Finally, region texture was synthesized to get texture image restoration result. The proposed model had been tested on many natural images. The experimental results show that the proposed model is valid and the restoration results are consistent with human visual system. The proposed algorithm has lower time complexity of image restoration, and its performance in subjective assessment of the quality of pictures is much better than the multi-region image reconstruction algorithm.

Key words: texture image restoration, image segmentation, combined subdivision, texture synthesis

摘要: 针对分段迭代曲线拟合存在的重建区域轮廓不连续、重建区域尺寸有误差等问题,提出了一种基于融合细分的纹理图像重构模型.首先提取原始图像的分割区域,经过轮廓跟踪与下采样得到区域形状的特征向量;然后利用三重逼近与三重插值统一的融合细分方法,重建区域轮廓曲线;最后合成区域纹理,得到纹理图像重构结果.在多幅自然场景图像上进行实验验证,并给出相应的实验结果和分析.实验结果表明,所提模型正确有效,具有和人类视觉特性相符合的重构结果; 所提算法能够减少图像重建时的处理时间,并在图像质量主观评价指标上明显优于多区域图像重建算法.

关键词: 纹理图像重构, 图像分割, 融合细分, 纹理合成

CLC Number: