摘要
当前较多图像修复算法采用单一大小样本块进行图像修复,不能适应图像不同差异的纹理丰富度变化,使得修复结果存在块效应以及模糊效应等不足。本文利用图像的梯度值,设计了基于梯度调节规则的图像修复算法。将图像的梯度信息引入优先权计算,联合数据项、置信度项目构造优先权计算函数,以计算优先修复块。利用图像的梯度变化率,建立梯度调节规则,用以调节样本块大小,适应不同的纹理丰富度。引入SSD(Sumofsquareddifferences)函数从源区域中寻找最优匹配块,实现图像修复。实验结果显示,所设计方法修复的图像具有良好的视觉效果。
At present,many image restoration algorithms use only one size sample block for image restoration,which is difficult to adapt to the different texture richness of the image,resulting in the defect of blocking and blurring effect.In this paper,we use the gradient value of image to design an image inpainting algorithm based on gradient adjustment rule.The gradient information of image is introduced into priority calculation,and priority calculation function is constructed by combining data items and confidence items to calculate priority repair blocks.The gradient adjustment rules are established to adjust the size of the sample block and adapt to different texture richness by using the gradient change rate of the image.The SSD function is introduced to find the best matching block from the source area and repair the repair block.Experimental results show that the image repaired by this algorithm has better visual effect.
作者
马凤娟
宋大伟
赵华
MA Feng-juan;SONG Da-wei;ZHAO Hua(Weifang engineering Career Academy,Weifang,Shandong 262500,China;Shandong University of Science and Technology,Qingdao,Shandong 266590,China)
出处
《井冈山大学学报(自然科学版)》
2019年第1期34-38,共5页
Journal of Jinggangshan University (Natural Science)
基金
山东省自然科学基金项目(ZR2013FQ030)
关键词
图像修复
梯度信息
优先权计算
样本块大小
纹理丰富度
SSD函数
image inpainting
gradient information
priority computation
sample block size
texture richness
SSD function