曾康铭,吴杏.基于引导因子与曲率惩罚模型的图像修复算法[J].包装工程,2018,39(23):209-215.
ZENG Kang-ming,WU Xing.Image Inpainting Algorithm Based on Guided Factor Coupling Curvature Penalty Model[J].Packaging Engineering,2018,39(23):209-215. |
| 基于引导因子与曲率惩罚模型的图像修复算法 |
| Image Inpainting Algorithm Based on Guided Factor Coupling Curvature Penalty Model |
| 投稿时间：2018-08-22 修订日期：2018-12-10 |
| DOI：10.19554/j.cnki.1001-3563.2018.23.034 |
| 中文关键词: 图像修复 引导因子 梯度特征 搜索范围 最小绝对差平方和函数 曲率惩罚模型 |
| 英文关键词: image inpainting guided factor gradient feature search range sum of squared differences function curvature penalty model |
| 基金项目:广西高校科学技术研究项目（KY2015YB530）；广西邕宁区科学研究与技术开发计划（20160321A） |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 解决当前图像修复方法主要通过待修复像素点的法向量来确定修复过程，无法保证其修复顺序从破损区域的周边至中心进行，导致修复图像中存在块效应和不连续效应等问题。方法 将引导因子与曲率惩罚模型相结合，设计新的图像修复方法。利用破损区域的中心像素点与其他任意待修复像素点之间的距离来构造引导因子，并将其与置信度项以及数据项结合，形成优先权模型，用于选取优先修复块。利用待修复块的梯度特性对其平滑度进行判断，以明确该待修复块对应的最优匹配块的搜索范围，使其通过最小绝对差平方和（SSD）函数来搜索最优匹配块，从而将最优匹配块中像素点扩散填充至待修复块。最后，基于像素点间的等照度线曲率来建立曲率惩罚模型，以更新置信度项，从而实现图像的修复。结果 测试数据表明，与已有修复方案相比，所提算法可以更好地兼顾修复质量与效率。结论 所提方案具有较好的修复质量，可用于损坏面积较大图像的复原。 |
| 英文摘要: |
| The work aims to solve such problems as block effect and discontinuity effect in the restored image induced by the current image inpainting method that mainly determines the restoration process with the normal vector of pixels to be repaired, so that the repairing order is not guaranteed from the periphery to the center of the damaged area. A new image inpainting method was designed based on guided factor coupled with curvature penalty model. The distance between the center pixel of the damaged area and other arbitrary pixels to be repaired was used to construct the guided factor, and the guided factor was combined with the confidence degree and data item to form the priority model for the selection of the priority repair block. The smoothness of the block to be repaired was judged by its gradient feature to determine the search range of the optimal matching block corresponding to such block, and the optimal matching block was searched through the sum of squared differences (SSD) function, so that the pixel point diffusion in the optimal matching block was filled to the block to be repaired. Finally, the curvature penalty model was constructed based on the curvature of the equal illumination line between pixels, which was used to update the confidence term and thus realize image restoration. The test data showed that, compared with the existing image inpainting scheme, the proposed algorithm could better give consideration to restoration quality and efficiency. The proposed scheme has better restoration quality and can be used for the restoration of extensively damaged image. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
渝公网安备 50010702501716号