| 陶兆胜,王磊,张敬寒,王彪.强化边缘结构分辨的分段自适应图像修复[J].光电子激光,2018,29(12):1350~1357 |
| 强化边缘结构分辨的分段自适应图像修复 |
| Enhanced edge resolution image inpainting based on piecewise adaptive algorithm |
| 投稿时间：2018-05-08 |
| DOI： |
| 中文关键词: 图像修复 优先权 边缘分辨因子 特征判断因子 特征项 |
| 英文关键词:image inpainting priority edge resolution factor feature judgment factor fea ture term |
| 基金项目:国家自然科学基金资助项目(51374007)资助项目 (安徽工业大学 机械工程学院 安徽 马鞍山 243032) |
|
| 摘要点击次数: 837 |
| 全文下载次数: 0 |
| 中文摘要: |
| 针对Criminisi算法对边缘纹理分辨不足、无法识 别图像局部区域的直线和曲线的特征以及通过颜色进行块匹配而造成修复效果失真等问题, 提出一种局部特征与边缘纹理分辨相结合的分段自适应图像修复 算法。首先在分段自适应图像修复算法的优先权中引入边缘分辨因子和特征判断因子,增强 对边缘、局部 区域的直线和曲线的分辨能力,克服破损边缘不合理的修复顺序；其次在块匹配准则中引入 特征项,提高 样本块的匹配准确率,避免块匹配的颜色匹配不足；然后采用分段自适应算法进行置信项更 新,解决置信 项快速趋于0的问题；最后采用主客观的评价体系对图像修复质量进行评价。评价结果显示 分段自适应图 像修复算法的图像修复质量优于其他算法,信噪比和峰值信噪比的评价值均提高在 0~3.8之间；在结构相 似度的优化程度上提高了0~0.7%。实验结果证明,分段自适应图像修 复算法有效地修复了破损图像,获得较好的图像视觉效果。 |
| 英文摘要: |
| Because the Criminisi algorithm could not distinguish the edge texture sufficiently or identify the features about the straight line and the curve of local area,besides the distortion of the image inpainting was brought out due to the improper matching patch by color,a piecewise adaptive image inpainting algorithm combining the local feature and the edge texture was proposed.Firstly,in order to overcome the unreasonable inpainting sequence of the damaged edge,the edge resolution factor and the feature judgment factor were introduced into the priority functio n to enhance the inpainting ability of distinguishing the edge and the features about the straigh t line and the curve of the local area.Secondly,the feature term was introduced to improve the matching accu racy of the sample patch in the patch matching criterion.Then the piecewise adaptive algorithm was adopted to update the confidence term to solve the problem of the rapid reduction of the confidence te rm.Finally,the subjective evaluation system and the objective evaluation system were applied to the image inpainting quality evaluation.The evaluation results show that the inpainting quality of th e proposed algorithm is better than that of other algorithms.Not only the signal-to-noise ratio and the peak signal-to-noise ratio are increased resp ectively by 0~3.8,but also the optimization degree of the structural similarity is increased by 0~0.7% .The experimental results indicate that the proposed algorithm could inpaint damaged image effectively an d obtain better image visual effect. |
| 查看全文 下载PDF阅读器 |
| 关闭 |
|
|
|
|
|