人臉圖像的二維線性判別分析研究

2.2萬字
本人今年最新原創(chuàng)的畢業(yè)設(shè)計(jì)，僅在本站獨(dú)家提交，大家放心使用


摘要 隨著圖像處理技術(shù)在實(shí)際工程中的不斷應(yīng)用，圖像處理分析日漸成為一門引人注目、前景遠(yuǎn)大的技術(shù)。人工智能的提出，使得人們對(duì)于計(jì)算機(jī)能夠像自己一樣具有識(shí)別分析能力有了很大的期待。而如何使得計(jì)算機(jī)具有識(shí)別理解圖像能力成為近年來模式識(shí)別以及圖像處理領(lǐng)域的研究熱點(diǎn)。
人臉圖像的二維線性分析是運(yùn)用基于二維線性方法分析歸類人臉圖像的研究，其中對(duì)于人臉圖像的分析具有重要意義，其豐富的信息量為區(qū)別個(gè)體信息邊界，目標(biāo)識(shí)別創(chuàng)造了優(yōu)越的條件，與其他方法相比，人臉圖像識(shí)別具有極大的優(yōu)越性。隨著人臉圖像的分辨率大幅度的提高以及信息集成化的迅速發(fā)展，使得對(duì)人臉圖像處理的要求越來越高。
本文采用了二維線性分析方法，其中運(yùn)用比較了多種二維圖像處理方法，例如：二維線性判別分析(2DLDA)、二維主成分分析(2DPCA)、二維非相關(guān)判別分析(2DUDA)、雙向二維線性判別分析(2D2LDA)等方法，通過比較采用了雙向二維線性判別分析法對(duì)人臉圖像進(jìn)行特征提取，其任務(wù)是從大量的原始數(shù)據(jù)中找到最能代表該圖像的少量特征。其次是運(yùn)用最近鄰方法將人臉圖像分類，其工作原理是尋找到被分類對(duì)象訓(xùn)練、測(cè)試數(shù)據(jù)集中的k個(gè)最近鄰域數(shù)據(jù)，然后根據(jù)這些最近鄰域數(shù)據(jù)分類屬性對(duì)目標(biāo)進(jìn)行預(yù)測(cè)，將得到的預(yù)測(cè)值賦給被分類對(duì)象的分類屬性。從而在類別決策時(shí)，只有極少量的相鄰樣本相關(guān)。其具體的操作是通過MATLAB的編程、調(diào)試、運(yùn)行來實(shí)現(xiàn)的，得到二維線性壓縮的圖像，與原有的圖像比較，并輸出最終的識(shí)別率。

關(guān)鍵詞： 二維線性分析 最近鄰法 圖像識(shí)別 MATBLE


Abstract Along with image processing technology in the practical engineering applications, image processing analysis has become a compelling, prospects great technology.Artificial intelligence is put forward that enables the people for the computer to like themselves with identification analysis ability have great expectations.And how to make computer has the identification ability to read images become in recent years the research focus in the field of pattern recognition and image processing.
Face image of two-dimensional linear analysis is using the method based on two-dimensional linear classification research of face image, which is of great significance for face image analysis, its abundant information for the difference between individual information boundaries, target recognition has created favorable conditions, compared with other methods, human face image recognition has a great advantage.With the improvement of the resolution of the face image greatly, and the rapid development of information integration, making more and more high to the requirement of human face image processing.
This paper adopts the two-dimensional linear analysis method, which using the comparison of the variety of two-dimensional image processing methods, such as: 2DLDA, 2DPCA, 2DUDA, 2D2LDA method, through comparing the 2D2LDA method of face images for feature extraction, the mission from lots of original data to find the most can represent the image of a small amount of features.Using the nearest neighbor method followed by face image classification, its working principle is to look for to be classified object of training and testing data set k nearest neighbor domain data, and then based on the nearest neighbor domain data classification properties forecast target, will get the predicted values assigned to the category attributes of the object being classified.To the decision, the category only very small amounts of the adjacent sample correlation.Its concrete operation is accomplished by MATLAB programming, debugging and running, two-dimensional linear compressed image, compared with the original image, and the output of the final recognition rate.

Key words: Two dimensional linear analysis Nearest neighbor method Image recognition MATLAB


目 錄
摘要…………………………………………………………………………………………I
目錄…………………………………………………………………………………………III
第一章 緒論………………………………………………………………………………1
1.1 生物特征識(shí)別技術(shù)概述……………………………………………………………1
1.2 人臉識(shí)別技術(shù)優(yōu)缺點(diǎn)…………………………………………………………………2
1.2.1 人臉識(shí)別技術(shù)優(yōu)勢(shì)………………………………………………………………2
1.2.2 人臉識(shí)別技術(shù)劣勢(shì)………………………………………………………………3
1.3人臉識(shí)別研究背景及研究意義……………………………………………………3
1.4 國(guó)內(nèi)外研究概述………………………………………………………………………4
1.5 本論文研究簡(jiǎn)介及組織結(jié)構(gòu)…………………………………………………………5
第二章 二維人臉圖像的分析……………………………………………………………6
2.1 二維人臉圖像處理系統(tǒng)組成………………………………………………………6
2.2 二維人臉圖像的預(yù)處理……………………………………………………………7
2.3 二維人臉圖像檢測(cè)…………………………………………………………………8
2.4 二維人臉圖像的特征提取…………………………………………………………8
2.5 二維人臉識(shí)別分類…………………………………………………………………9
第三章 二維線性分析方法以及最近鄰分類法……………………………10
3.1..