一種結(jié)合結(jié)構(gòu)和統(tǒng)計(jì)特征的脫機(jī)數(shù)字識(shí)別方法

張玉葉，姜彬，李開(kāi)端，王春歆

一種結(jié)合結(jié)構(gòu)和統(tǒng)計(jì)特征的脫機(jī)數(shù)字識(shí)別方法

張玉葉，姜彬，李開(kāi)端，王春歆

脫機(jī)手寫(xiě)數(shù)字識(shí)別歸根結(jié)底是數(shù)字的圖像特征匹配識(shí)別問(wèn)題。為了提高識(shí)別效率，需要降低數(shù)字的特征維數(shù)；同時(shí)要提高數(shù)字識(shí)別的準(zhǔn)確性，必須考慮手寫(xiě)數(shù)字的筆畫(huà)結(jié)構(gòu)不穩(wěn)定的特點(diǎn)。提出了一種結(jié)合字符統(tǒng)計(jì)特征和結(jié)構(gòu)特征的識(shí)別方法。首先，利用主分量分析法抽取數(shù)字字符圖像的統(tǒng)計(jì)特征，通過(guò)對(duì)主分量重建模型的誤差分析進(jìn)行數(shù)字識(shí)別；為了進(jìn)一步提高數(shù)字識(shí)別的準(zhǔn)確度，再加入數(shù)字的寬高比結(jié)構(gòu)特征進(jìn)行比對(duì)識(shí)別。最后自制訓(xùn)練樣本及測(cè)試樣本庫(kù)進(jìn)行識(shí)別實(shí)驗(yàn)，實(shí)驗(yàn)結(jié)果表明本方法識(shí)別準(zhǔn)確率較高。

脫機(jī)手寫(xiě)數(shù)字識(shí)別；主分量分析；結(jié)構(gòu)特征；統(tǒng)計(jì)特征

0 引言

脫機(jī)數(shù)字識(shí)別不能利用聯(lián)機(jī)識(shí)別可以得到的時(shí)間、筆順等動(dòng)態(tài)信息，系統(tǒng)實(shí)現(xiàn)比較困難[1]。手寫(xiě)數(shù)字識(shí)別最重要的環(huán)節(jié)是數(shù)字字符的特征提取[2]。目前，手寫(xiě)數(shù)字的特征可分為兩類(lèi)：統(tǒng)計(jì)特征和結(jié)構(gòu)特征。統(tǒng)計(jì)特征是利用字符樣本庫(kù)，找出的0到9中每類(lèi)字符空間分布的統(tǒng)計(jì)規(guī)律。結(jié)構(gòu)特征包括數(shù)字的構(gòu)造如端點(diǎn)、交叉點(diǎn)、輪廓等。兩類(lèi)特征各有優(yōu)勢(shì)，統(tǒng)計(jì)特征可以描述數(shù)字的本質(zhì)特征，適用于給定訓(xùn)練集差別不大的情況；利用結(jié)構(gòu)特征能夠精確描述數(shù)字的細(xì)節(jié)特征，對(duì)書(shū)寫(xiě)較規(guī)范的數(shù)字有較高的識(shí)別率。可以將兩類(lèi)特征結(jié)合運(yùn)用，以便更好地進(jìn)行數(shù)字識(shí)別。

本文提出了一種能夠結(jié)合字符統(tǒng)計(jì)特征和結(jié)構(gòu)特征的識(shí)別方法，利用主分量分析法抽取數(shù)字字符樣本的統(tǒng)計(jì)特征，通過(guò)對(duì)主分量重建模型的誤差分析進(jìn)行字符識(shí)別；為了進(jìn)一步提高字符識(shí)別的準(zhǔn)確度，加入寬高比結(jié)構(gòu)特征進(jìn)行字符比對(duì)識(shí)別。

1 脫機(jī)手寫(xiě)數(shù)字識(shí)別預(yù)處理

脫機(jī)手寫(xiě)數(shù)字識(shí)別，首要任務(wù)就是將紙質(zhì)載體掃描為圖像信息以便計(jì)算機(jī)處理。得到的數(shù)字圖像在進(jìn)行識(shí)別之前需要進(jìn)行圖像的預(yù)處理，以消除圖像中無(wú)關(guān)的信息，從而改進(jìn)特征提取、圖像分割、匹配和識(shí)別的可靠

An Off-line Handwritten Numeral Recognition Method Combined With the Statistical Characteristics and Structural Features

Zhang Yuye1, Jiang Bin2,Li Kaiduan1,Wang Chunxin3
(1.Naval Aeronautical and Astronantical Universing, Qingdao 266041,China; 2.Qingdao University,Qingdao 266071,China; 3. North China Sea Fleet,Qingdao 266041,China))

Off-line handwritten numeral recognition is a pattern recognition problem of the images of ten numbers. In order to improve the recognition efficiency， the character dimension of number’s image should be decreased. As well, in order to improve the recognition veracity, the character mode instability which resulted from different writing styles and habits should be considered. The article proposed a numbers recognition method combined with the statistical characteristics and structural features of numbers. Firstly, the principal component analysis (PCA) method is adopted to extract statistical characteristics of numeral image. The numeral recognition will be realized through analysis of the reconstruction error of model which is reconstructed by the principal components. In order to further determine the type of numeral, the structural features of width and height rate should be added. Finally, through experiments on the identification of numeral image, the reliability and accuracy of this method of digital recognition is verified, and the deficiency of this method in real-time recognition is analyzed.

Offline Handwritten Numeral Recognition; Principal Component Analysis; Structural Features; Statistical Characteristics

TP311

A

1007-757X(2016)08-0076-04

張玉葉（1980-），女（漢族），海軍航空工程學(xué)院，講師，研究方向：數(shù)字圖像處理，青島，266041

姜 彬（1980-），女（漢族），青島大學(xué)，信息工程學(xué)院，講師，研究方向：圖像特征識(shí)別，青島，266041

李開(kāi)端（1967-），男（漢族），海軍航空工程學(xué)院，副教授，研究方向：圖像判讀，青島，266041

王春歆（1979-），男（漢族），北海艦隊(duì)，工程師，研究方向：圖像目標(biāo)檢測(cè)，青島，266041