本文選題：驗證碼分割 + 復(fù)雜網(wǎng)絡(luò)　； 參考：《南京航空航天大學(xué)》2014年碩士論文

【摘要】：隨著互聯(lián)網(wǎng)的迅猛發(fā)展，網(wǎng)絡(luò)給人們的生活帶來了極大方便，同時網(wǎng)絡(luò)上的安全問題也日益突出。驗證碼作為一種區(qū)分機(jī)器和人類的手段，已廣泛應(yīng)用于網(wǎng)絡(luò)安全等領(lǐng)域。目前，基于字符的驗證碼識別研究已經(jīng)越來越成熟，而驗證碼分割由于起步比較晚并且驗證碼中的字符可變性也比較大，使得驗證碼分割比驗證碼識別研究更為困難。因此，越來越復(fù)雜的驗證碼分割研究已經(jīng)成為當(dāng)前最為棘手的問題，但研究驗證碼分割技術(shù)在增強(qiáng)網(wǎng)絡(luò)的安全性，防止對網(wǎng)站的惡意攻擊等方面有著非常重要的意義。 本文主要針對粘連字符驗證碼進(jìn)行分析研究，并將研究結(jié)果與不同的算法進(jìn)行對比，使得該研究具有一定的實際應(yīng)用價值和作用。主要的工作和研究成果如下： 1）針對粘連字符驗證碼的特點(diǎn)，，提出了基于復(fù)雜網(wǎng)絡(luò)的社區(qū)劃分分割算法，并將不同的粘連字符個數(shù)分割結(jié)果進(jìn)行分析對比。實驗結(jié)果表明該算法能夠有效的分割粘連字符，對于Authorize、京東商城、天涯社區(qū)、Windows Live和淘寶網(wǎng)驗證碼的分割成功率分別能達(dá)到98%、95%、71%、55%、33%。同時，隨著粘連字符的增多，字符分割成功率隨之降低，分割時間也隨之增長。 2）針對天涯社區(qū)驗證碼的特點(diǎn)設(shè)計出了基于蓄水池的分割算法，并與不同的分割方法進(jìn)行了對比，實驗結(jié)果表明該方法的分割成功率能達(dá)到92%，并且對于驗證碼中有字符重疊或者兩字符之間像素點(diǎn)相差過大情況的分割非常有效。該方法同時也解決了社區(qū)劃分分割算法的局限性，有效地縮短了字符分割時間。 3）針對驗證碼的識別提出了33個特征提取的方法，并用C支持向量分類機(jī)對其進(jìn)行識別，其實驗結(jié)果表明該方法對天涯社區(qū)驗證碼字符的識別率能夠達(dá)到93%。最后，介紹了本文從驗證碼獲取、圖片預(yù)處理到字符分割以及字符識別的系統(tǒng)實現(xiàn)。
[Abstract]:With the rapid development of the Internet, the network has brought great convenience to people's life, and the security problems on the network have become increasingly prominent. As a means of distinguishing machine from human, CAPTCA has been widely used in network security and other fields. At present, the research of character based verification code recognition has become more and more mature, and because of the relatively late start and the character variability in the verification code, it is more difficult to segment the verification code than the research of the verification code recognition. Therefore, more and more complex research on CAPTC-code segmentation has become the most difficult problem at present, but the research of CAPTC-code segmentation technology in enhancing the security of the network, preventing malicious attacks on websites and other aspects has a very important significance. This paper focuses on the analysis and research of the adhesive character verification code, and compares the research results with different algorithms, which makes the research have a certain practical application value and function. The main findings and findings are as follows: 1) according to the characteristics of the adhesive character verification code, a community partition segmentation algorithm based on complex network is proposed, and the results of the different number of adhesion characters are analyzed and compared. The experimental results show that the algorithm can effectively segment the adhesive characters. For Authorize, JingDong Mall, Tianya Community Windows Live and Taobao, the success rate of the segmentation of the verification codes can reach 98%, 95% and 71% and 5555%, respectively. At the same time, with the increase of adhesion characters, the success rate of character segmentation decreases and the segmentation time increases. 2) according to the characteristics of Tianya community verification code, a segmentation algorithm based on cistern is designed and compared with different segmentation methods. The experimental results show that the method can achieve a success rate of 92 and is very effective for the segmentation where there is overlap of characters in the verification code or the pixel difference between two characters is too large. The method also solves the limitation of community partitioning algorithm and effectively shortens the time of character segmentation. 3) A method of 33 feature extraction is proposed for the identification of CAPTC-code, and it is recognized by C support vector classifier. The experimental results show that the recognition rate of this method for the character of verification code in Tianya community can reach 933%. Finally, this paper introduces the implementation of the system from the acquisition of verification code, image preprocessing to character segmentation and character recognition.
【學(xué)位授予單位】：南京航空航天大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TP393.08

【參考文獻(xiàn)】

相關(guān)期刊論文 前2條

1 王曉波;王興芬;;基于MODI的驗證碼識別系統(tǒng)設(shè)計與實現(xiàn)[J];北京信息科技大學(xué)學(xué)報(自然科學(xué)版);2010年01期

2 李文s

本文編號：1906285