【正文】 ...........4 模式識(shí)別系統(tǒng) .......................................................................................................4 相關(guān)值計(jì)算 ...........................................................................................................4 .........................................................................................................................5 軟件的介紹 ................................................................................................5 Matlab 的主要優(yōu)缺點(diǎn) ...........................................................................................6 圖像類型及轉(zhuǎn)換分析 ................................................................................73 手寫特征的提取與選擇 .....................................................................................................9 特征的種類與篩選 ......................................................................................................9 筆劃密度特征 ........................................................................................................9 傅立葉變換特征 ....................................................................................................9 輪廓特征 ..............................................................................................................11 投影特征 ..............................................................................................................12 重心及重心矩特征 ..............................................................................................14 首個(gè)黑點(diǎn)位置特征 ..............................................................................................14 粗網(wǎng)格特征 ..........................................................................................................15 特征提取方法 ............................................................................................................15 結(jié)構(gòu)特征提取方法 ..............................................................................................15 統(tǒng)計(jì)特征提取方法 ..............................................................................................16 手寫特征模式識(shí)別方法 ............................................................................................174 BAYES 分類器在手寫特征中應(yīng)用 .................................................................................19 BAYES 分類器 .............................................................................................................19 基于概率的 BAYES 決策 ............................................................................................20 基于最小錯(cuò)誤率 BAYES 的手寫數(shù)字字符分類 ........................................................21 樣品均值 ..............................................................................................................21 協(xié)方差矩陣 ..........................................................................................................22 先驗(yàn)概率 .............................................................................................................22 協(xié)方差矩陣的行列式 .........................................................................................22 協(xié)方差矩陣的逆矩陣 .........................................................................................23 判別函數(shù) .............................................................................................................24 基于最小風(fēng)險(xiǎn)的 BAYES 分類的實(shí)現(xiàn) ........................................................................24 與最小錯(cuò)誤 Bayes 決策的相同之處 .................................................................24 后驗(yàn)概率 .............................................................................................................25 損失函數(shù) .............................................................................................................255 分類實(shí)驗(yàn)與信息 ...............................................................................................................26 特征提取方法的軟件實(shí)現(xiàn) ........................................................................................26 截圖并說(shuō)明仿真過(guò)程 ................................................................................................26 獲得實(shí)驗(yàn)結(jié)果 ............................................................................................................28 分類結(jié)果分析與評(píng)價(jià) ................................................................................................296 總 結(jié) .................................................................................................................................30致 謝 ..................................................................................................................................31參考文獻(xiàn) ..............................................................................................................................32附 錄 ..................................................................................................................................34附錄 1．編程代碼： .......................................................................................................34附錄 2．仿真部分截圖： ...............................................................................................391 緒 論 手寫數(shù)字特征提取與分析的背景與意義手寫數(shù)字特征提取與分析在學(xué)科上屬于模式識(shí)別和人工智能的范疇。 關(guān)鍵詞：模式識(shí)別；最小錯(cuò)誤；最小風(fēng)險(xiǎn)；特征選擇；模擬手寫；Matlab 實(shí)現(xiàn)Handwritten digital feature extraction and analysis　　　　　　　　　　　　　　　　　　　　Liang Jie，Electronic and information engineering, College of Information Science and Technology Abstract: At present, the field of pattern recognition in everyday life has been more and more widely used, such as the face, fingerprint recognition, character recognition, vehicle license plate recognition. Therefore, the digital identification of learning and research is very necessary.The topic for the digital character recognition simulation demo system. Mainly using normal distribution under the minimum error rate of Bayes method and Bayes method to achieve the minimum risk, handwritten digits from 0 to9 of the identification. The system first is to realize the simulation of handwritten numeral。使用最小錯(cuò)誤率 Bayes 方法，在判別過(guò)程中能使錯(cuò)誤率達(dá)到最小，即使錯(cuò)分類出現(xiàn)的可能性最小，