基于能量信息耦合梯度调节机制的图像修复算法设计与应用  

Design and Application of Image Inpainting Algorithm Based on Energy Information and Gradient Adjustment Mechanism

作　　者：杜媛[1] DU Yuan(School of Big Data Application,Xi'an Vocational and Technical College,Xi'an 710072,China)

机构地区：[1]西安职业技术学院大数据应用学院,西安710072

出　　处：《计算机测量与控制》2023年第10期147-152,181,共7页Computer Measurement &Control

基　　金：陕西职业教育乡村振兴研究院课题(22YB022)。

摘　　要：为了克服当前图像修复算法主要依靠图像的置信度信息来获取优先修复块,忽略了图像的能量信息,导致修复结果中存在不连续及伪吉布斯现象等缺陷,设计了基于能量信息与梯度调节机制的图像修复算法;首先,通过区域能量函数来求取图像的能量信息,以计算待修复块的优先权信息,得到优先修复块;然后,基于图像梯度模值,建立梯度调节机制,以调节样本块的大小,获取与图像纹理相适应的样本块尺寸;引入平方差求和(SSD)函数,以确定最优匹配块;最后,通过像素点间的差异性,构造相似惩罚因子,以更新置信度项,完成图像的修复;实验结果显示,在所提算法中将变化阈值取值为0.5时,较当前图像修复方案而言,所提算法具备更好的修复性能,所得到的修复图像拥有更好的纹理连贯性与更高的结构相似(SSIM)值;当对像素丢失比例达到50%的图像进行修复时,较对照组算法而言,所提算法修复图像的SSIM值分别提高了6.5%和17.5%。In order to overcome the current image restoration algorithms mainly rely on the image confidence information to obtain the priority repair block,ignoring the image energy information,making the algorithm's repair performance decline,resulting in the repair image discontinuity and pseudo Gibbs phenomenon and other defects.Energy information and gradient adjustment mechanism are used to repair the image.Firstly,the energy information of the image is obtained by the region energy function,and the priority information of the block to be repaired is calculated by the data and confidence terms.Then,on the basis of image gradient modulus,a gradient adjustment mechanism is established to adjust the size of the sample block and obtain the sample block size corresponding to the image texture.Finally,the sum function of square difference is introduced to calculate the similarity between the block to be repaired and the matching block,so as to obtain the optimal matching block.Through the difference between pixels,construct the similarity penalty factor to update the confidence term and complete the image restoration.The experimental results show that when the change threshold value is set in the proposed algorithm,the proposed algorithm has better repair performance than the current image repair scheme,and the repaired image has better texture coherence and higher structural similarity value.When repairing an image with a pixel loss ratio of 50%,the SSIM value of the repaired image of the proposed algorithm is 6.5%and 17.5%higher than that of the control group.

关 键 词：图像修复 能量信息 梯度模值 梯度调节机制 平方差求和函数 相似惩罚因子 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]