张小寒,杨东刚.基于平滑度测量与互相关制约的图像修复算法[J].包装工程,2019,40(7):222-228.
ZHANG Xiao-han,YANG Dong-gang.Image Inpainting Algorithm Based on Smoothness Measurement andCross-correlation Constraints[J].Packaging Engineering,2019,40(7):222-228. |
| 基于平滑度测量与互相关制约的图像修复算法 |
| Image Inpainting Algorithm Based on Smoothness Measurement andCross-correlation Constraints |
| 投稿时间：2018-12-31 修订日期：2019-04-10 |
| DOI：10.19554/j.cnki.1001-3563.2019.07.034 |
| 中文关键词: 图像修复 Laplace算子 优先权 等照度线 平滑度测量因子 互相关制约 |
| 英文关键词: image inpainting Laplace operator priority equal illumination line smoothness measurement factor cross-correlation constraints |
| 基金项目:四川省自然科学基金（2017JSQ487） |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 为了解决当前较多图像修复算法在对损坏区域实施填充修复时没有考虑图像块之间的相关性，导致修复图像中存在振铃以及连接间断等不足，拟设计平滑度测量因子耦合互相关制约的图像修复算法。方法 将图像的Laplace算子引入到优先权的计算过程中，以增加图像的边缘信息，优化数据项的成分，并利用置信度、图像的Laplace算子和数据项构造优先权计算模型，以获取优先修复块。利用等照度线的二阶导数，构造平滑度测量因子获取图像的平滑度，并以图像的平滑度为导向，对最优匹配块的搜索区域进行定位。通过误差平方和函数（Sum of Squared Differences，SSD）在定位的搜索区域中搜寻最优匹配块，并采用互相关系数函数对最优匹配块的唯一性进行制约，以提高所获最优匹配块的准确性。结果 实验结果显示，与当前修复技术相比，所提方法具有更高的连接完整性和清晰度。结论 所提方案可对损坏图像完成较好的视觉复原，在图像信息处理领域具有一定的参考价值。 |
| 英文摘要: |
| To solve the problem as ringing and discontinuous connection effect in repaired images induced by ignoring the correlation between image blocks when filling and repairing damaged areas in current image inpainting algorithm, this paper aims to design an image inpainting algorithm based on smoothness measurement factor coupling and cross-correlation constraints. The Laplace operator was introduced into the priority calculation process to increase the edge information of the image, optimize the components of the data items, and construct the priority calculation model by using the confidence and the Laplace operator and data items of the image to obtain the priority repair module. The smoothness measurement factor was constructed with the second-order differential of the equal illumination line to obtain the smoothness of the image. And the search area of the optimal matching block was located based on the smoothness of the image. In order to improve the accuracy of the optimal matching block, the error square sum function was used to search the optimal matching block in the searching area of the location, and the cross-correlation coefficient function was used to restrict the uniqueness of the optimal matching block. The experimental results showed that the image restored by the proposed method had more complete edge continuity and clarity than the image restored by the current method, and the visual effect of the restored image can be effectively improved. The proposed scheme can achieve better visual restoration of damaged images, which has a certain reference value in the field of image information processing. |
|
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