基于機(jī)器視覺(jué)的紋理識(shí)別

電子信息工程論文編號(hào):DZXX236  論文字?jǐn)?shù):13151,頁(yè)數(shù):28

目錄

摘要 I
Abstract II
第1章  緒論 1
1.1 數(shù)字圖像處理的定義及其發(fā)展歷程 1
1.2 研究的背景，目的和意義 2
1.3 圖像紋理信息的概念 3
1.4 紋理識(shí)別模型的分類和比較 4
1.5 紋理識(shí)別的應(yīng)用領(lǐng)域 5
第2章 Microsoft Visual C++ 6.0 編譯環(huán)境介紹 7
2.1 Visual C++ 6.0簡(jiǎn)介 7
2.2 Visual C++ 6.0的特點(diǎn) 7
2.3 Visual C++ 6.0的常用向?qū)Ш?jiǎn)介 8
2.4 MFC類庫(kù)簡(jiǎn)介 9
第3章 紋理分析的概念和算法的實(shí)現(xiàn) 11
3.1 紋理的定義 11
3.2 紋理分析的方法 11
3.3 灰度共生矩陣法的算法原理 12
3.4 灰度共生矩陣算法的程序?qū)崿F(xiàn) 15
3.5 模式識(shí)別的實(shí)現(xiàn) 16
3.6 實(shí)驗(yàn)結(jié)果分析 17
第4章 總結(jié)和展望 22
致謝 23
參考文獻(xiàn) 24

摘要

 紋理是一種很重要的視覺(jué)信息, 通過(guò)對(duì)物件表面紋理的分析，可以讓計(jì)算機(jī)識(shí)別出該物件所屬的材料。因此圖像紋理識(shí)別成為數(shù)字圖像識(shí)別的一個(gè)重要分支。紋理識(shí)別在織物紋理自動(dòng)檢測(cè)、醫(yī)學(xué)圖像分析、場(chǎng)景分析和遙感圖像處理等領(lǐng)域中的廣泛應(yīng)用，進(jìn)一步證明紋理識(shí)別的重大研究意義和實(shí)用價(jià)值。通過(guò)比較各種算法的精度及其功能實(shí)現(xiàn)的復(fù)雜程度，本文最后采用統(tǒng)計(jì)法中的灰度共生矩陣法實(shí)現(xiàn)紋理識(shí)別功能。灰度共生矩陣法，是建立在估計(jì)圖像的二階組合條件概率密度函數(shù)的數(shù)學(xué)模型基礎(chǔ)之上。在VC++編譯環(huán)境下，先通過(guò)灰度圖計(jì)算出4個(gè)不同角度的灰度共生矩陣，然后計(jì)算出每個(gè)像元對(duì)在灰度共生矩陣中的概率，最后根據(jù)能量，熵，慣性矩，相關(guān)，局部平穩(wěn)的計(jì)算公式計(jì)算出這5個(gè)特征值。實(shí)驗(yàn)結(jié)果表明，以這種方法做出來(lái)的識(shí)別系統(tǒng)，能夠較準(zhǔn)確地計(jì)算出每個(gè)特征值，從而區(qū)別不同的材料。同時(shí)也證明了，灰度共生矩陣法是一種實(shí)踐性較高的算法。

關(guān)鍵字：紋理，紋理識(shí)別，灰度共生矩陣，VC++

Abstract

 Texture is one of the most important visual information, which makes the computer recognizes the material of the objects by analyzing the surface texture of them. Therefore, image texture recognition becomes one of the key branches of digital image recognition. Texture recognition has been widely used in the fields of fabric texture automatic detection, medical image analysis and remote sensing image processing, which further proves its research importance and practical value. With comparisons of different methods and the complex degree of function, the thesis finally applies gray level co-occurrence matrix to functionalize the texture recognition. Gray level co-occurrence matrix method is based on the second-order conditional probability density function of the estimated image. Under the condition of Visual C++ compilation, firstly the computer computes the gray level co-occurrence matrix of 4 angles from the gray image. Then, it will compute the probability of every pair pixel in the gray level co-occurrence matrix. Finally, it will compute the 5 Characteristic values according to the formula of energy, entropy, inertia matrix, correlation, and local balance. According to the experiment, this recognized system of that method can precisely compute those five characteristic values leading to the distinguishing of different materials. At the same time, the gray level co-occurrence matrix is proved to be a precisely practical method.

Keywords: Texture, Texture recognition, Gray level co-occurrence matrix, VC++