摘要
残缺档案因其边缘受到各种因素的影响,具有不规则与模糊的特性,增加了准确捕捉边缘轮廓的难度,导致无法逼近实际轮廓曲线,使修复区域与周围区域不协调,进而影响了整体的修复成效。为此,提出了一种基于水平集图像恢复的残缺档案智能修复方法。利用自适应边缘检测方法提取档案残缺区域的粗略轮廓,初步确定边缘范围;基于此,利用水平集方法对该轮廓曲线展开演化,使其逐渐逼近实际轮廓曲线,提高轮廓提取精度,为后续的修复提供更精确的修复区域界定,确保修复区域与周围区域相协调;最后通过Criminisi算法对轮廓内部区域的信息展开修复,以完成残缺档案图像的智能修复。结果表明,所提方法具有较佳的轮廓提取效果,且修复后的档案清晰度较高,重叠率在95%以上。
Due to the influence of various factors on its edges,the damaged archives have the characteristics of ir-regularity and blurriness,which increases the difficulty of accurately capturing the edge contours.As a result,it is im-possible to approximate the actual contour curve,making the restored area incongruous with the surrounding area,thus affecting the overall restoration effect.Therefore,an intelligent restoration method for damaged archives based on level set image restoration is proposed.An adaptive edge detection method is used to extract the rough contours of the in-complete document areas and initially determine the edge range;Based on this,the level set method is employed to e-volve the contour curve,gradually approaching the actual contour curve,improving the accuracy of contour extraction,and providing a more precise definition of the restoration area for subsequent repairs,ensuring that the restoration area is coordinated with the surrounding areas;Finally,the Criminisi algorithm is used to restore the information within the contour area to complete the intelligent restoration of the incomplete document image.The results indicate that the proposed method has a better contour extraction effect,and the clarity of the repaired documents is relatively high,with an overlap rate of over 95%.
作者
汤宁
支亚京
张淑莹
胡兴炜
TANG Ning;ZHI Ya-jing;ZHANG Shu-ying;HU Xing-wei(Guizhou Meteorological Data Center,Guiyang Guizhou 550002,China)
出处
《计算机仿真》
2025年第9期293-296,325,共5页
Computer Simulation
基金
基于数字标识符水印的气象数据安全技术研究(黔气科合SS[2023])。
关键词
水平集方法
图像恢复
残缺档案修复
轮廓提取
Level Set Method
Image Restoration
Incomplete Document Repair
Contour Extraction
