基于区域分割和DM642的数字图像修复系统研究  

Research on digital image restoration system based on region segmentation and DM642

作　　者：曲美亭[1] 冀轶卿 QU Meiting;JI Yiqing(Northeast Agricultural University,Harbin 150030,China)

机构地区：[1]东北农业大学,黑龙江哈尔滨150030

出　　处：《现代电子技术》2024年第11期59-63,共5页Modern Electronics Technique

基　　金：黑龙江省艺术科学规划项目(2023D050)。

摘　　要：由于数字图像中颜色通道之间关联性显著,改变某颜色分量,会导致剩余分量随之变化,使数字图像整体视觉效果出现变化,针对该问题,文中研究基于区域分割和DM642的数字图像修复系统。将DM642处理器作为系统核心处理器,在该处理器协助下,针对破损数字图像采用基于分通道自体理论的数字图像区域分割方法,利用图像颜色信息和像素间的空间关系,实现图像修复区域分割。针对需修复区域,通过基于Criminisi算法的数字图像修复方法提取修复像素点生成修复块,利用图像自相似性,在未破损区域找到与修复块匹配度最高的样本块,用此样本填充修补破损区域。实验结果表明,此系统能够有效分割数字图像的修复区域,修复视觉效果较好,且修复后数字图像不存在明显失真问题,可实现多图像并行快速修复。Due to the significant correlation between color channels in digital images,changing a certain color component can lead to changes in the remaining components,which results in changes in the overall visual effect of the digital images.In view of this,a digital image restoration system based on region segmentation and DM642 is studied.The DM642 processor is used as the core processor of the system.With the assistance of this processor,a digital image region segmentation method based on the subchannel self theory is adopted for damaged digital images.The image restoration region segmentation is achieved by image color information and spatial relationships between pixels.For the regions that needs to be restored,a digital image restoration method based on the Criminisi algorithm is used to extract the pixel points that need to be restored and generate restoration blocks.The image self-similarity is used to find out the sample block with the highest matching degree with the restoration block in the undamaged region.This sample is used to fill and restore the damaged region.The experimental results show that the proposed system can effectively segment the digital image region that needs to be restored,and the visual restoration effect is good,and there is no obvious distortion in the restored digital images.Therefore,the proposed system can achieve simultaneous and fast restoration of multiple images.

关 键 词：区域分割 DM642处理器 数字图像 修复系统 分通道自体理论 Criminisi算法 置信度 修复块 

分 类 号：TN911.73-34[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]