作　　者： ; ; ;

机构地区： 重庆邮电大学计算机科学与技术学院

出　　处： 《计算机仿真》 2011年第2期329-332,共4页

摘　　要： 研究图像修补问题,为了适应自然图像纹理信息的多样性,提高图像的修复质量,待修复图像的填充顺序和匹配准则极为重要。合适的填充顺序和匹配准则,对于待修复图像的修复质量具有重要意义。提出样图的纹理合成技术的图像修复算法,在Criminisi A的算法基础上,提高待修复图像中已知像素的权重,加入颜色信息,并采用新的匹配搜索方法和自定义的结构-距离权重匹配准则对相似块进行匹配。通过仿真证明,改进方法对大面积的图像缺失的修复和多余物体的去除都有很好的效果,提高了修复质量,加速修补速度。 The filling order and matching criterion of image inpainted are extremely important based on the image inpainting algorithm of exemplar-based texture synthetic technology.And it pays a significant role in the inpainting quality of image inpainted.In order to adapt the diversity of the nature image texture information and improve the inpainting quality,an algorithm is proposed on the basis of Criminisi A.the weight of the known pixels is increased,the color information is added,a new matching searching method is employed,and the customized structure-distance weight criterion is adopted to match the similar blocks.Experiments demonstrate that the improved algorithm is of very good effects for the inpainting of the large area image missing and the redundant objects removing in the complicated background,and can increase the quality of image inpainted.

关 键 词： 图像修复 匹配准则 颜色信息 结构 距离权重

领　　域： [自动化与计算机技术] [自动化与计算机技术]