碎纸片的拼接复原数学模型的构建

碎纸片的拼接复原数学模型的构建

发表时间：2014-11-27T14:26:53.797Z 来源：《价值工程》2014年第9月上旬供稿作者：毕楷明[导读] 以纵横方式破碎纸片，利用同行文字行间距一致性的主要特性可解决横向碎纸片的拼接复原问题。Construction of Mathematical Model of Splicing Scrap Recovery毕楷明BI Kai-ming（东北大学理学院，沈阳110819）（NEU College of Sciences，Shenyang 110819，China）

摘要院本文讨论在碎纸机以不同方式破碎纸片的情况下建立碎纸片的拼接复原模型，以解决碎片数量巨大时人工拼接的难题，本文建立了三个具有针对性的模型。

模型一：方差分析法下的碎纸片拼接模型。在以纵切方式破碎纸片的情况下，提取碎纸片左右边缘的灰度列向量，利用碎纸片边缘处为单边同宽空白区域的特殊性对碎纸片进行定位，再利用方差分析法和欧式距离解决了纵切碎纸片的拼接复原问题。模型二：文字行间距一致性的碎纸片拼接模型。以纵横方式破碎纸片，利用同行文字行间距一致性的主要特性可解决横向碎纸片的拼接复原问题，简化了模型，将离散的像素灰度矩阵平均化处理，进而利用欧氏距离对碎纸片进行匹配，得到了碎纸片复原后的完整图片。模型三：二值化Otsu 算法的碎纸片拼接复原模型。本文从双面纵横破碎纸片的问题出发，建立了纸片二值化Otsu 法拼接模型，先对碎纸片分组预处理，为将复杂模型简单化，再利用全局阈值方法中典型的Otsu 法求取碎纸片的最佳阈值，以该阈值对碎纸片中所含灰度值信息进行划分实现二值化处理，将边缘区域明显化，利用统计学方法求取拼接后的纸片间成功匹配的像素点占纸片边缘的概率，最终双面纵横破碎纸片的拼接复原问题得以解决。Abstract: This paper discusses the construction of splicing scrap recovery model under the condition of shredder breaking paper

intopieces in different ways, so as to solve the problem of artificial splicing when there is a great amount of pieces. This paper establishes threecorresponding model.Model One: Paper Scrap Splicing Model under Analysis of Variance.Shredding paper through longitudinal mode, the paper selects the gray scraps of paper around the edge extraction column vector,locates the paper scrap by using edge of paper scraps as blank area with same width, then solves the problem of reconstruction of thelongitudinal cutting paper splicing through analysis of variance method and Euclid Distance.Model Two: Paper Scrap Splicing Model with Consistency of Text Line Spacing.Shredding paper through vertical and horizontal mode, its main characteristics of peer text line spacing consistency can solve theproblem of reconstruction of splicing transverse paper scraps, simplifies the model, processes the pixel matrix of discrete in average andmatches the paper scraps through Euclid Distance and then gets the complete picture of paper scrap after recovery.Model Three: Paper Scrap Splicing Model Based on Binaryzation Otsu Algorithm.This paper firstly expounds the double side's vertical and horizontal mode, establishes the paper scrap splicing model based onbinaryzation Otsu algorithm. The paper firstly does preconditioning for paper scraps into groups, simplifies the complex model, and then getsthe optimal threshold of the paper scraps by using typical Otsu algorithm of global threshold method. The paper classifies the gray valueinformation of paper scraps through this threshold to realize binaryzation processing, specifies the edge area, evaluates the probability ofsuccessful matching pixels on edge of splicing paper, and finally solves the mosaic and restoration problems of double side's vertical andhorizontal mode. 关键词院离散；方差分析；置信区间；阈值；Otsu 算法Key words: discrete；analysis of variance；confidence interval；threshold；Otsu algorithm中图分类号院TQ018 文献标识码院A 文章编号院1006-4311（2014）25-0238-031 模型一考虑以为空间拼接情况，为了获取拼接图像所必须的数据，文章以像素为单位离散所得碎片：利用VC++使用了Windows.H 头文件并调用RGB 等结构定义获得不同像素点的g 值[1]，生成了多个灰度矩阵。由于本题主要研究碎片的拼接，故只需考虑碎片的边缘部分，故分别提取全部碎片的最左侧和最右侧的g 值列向量：文章分别找出其中最左侧g 值列向量的值全为255（即像素全白）的和最右侧g 值列向量的值全为255 的两个碎片，于是左侧g 值全为255 的碎片对应左一位置，同理右侧g 值全为255 的碎片对应左一位置。再考虑剩余的碎片（本文中考虑18 个碎片）的对号入座问题，使最左侧碎片分别与其他碎片的最左侧灰度g 值列向量进行相同y 值下作差，得到不同碎片的G 差。先求出左一位置碎片最右侧g 值列向量：