王瑞祥
【摘要】本文基于花崗巖和砂巖數(shù)字圖像特征,利用小波分析理論及Bayes決策理論建立起巖石中幾種成份(云母、石英、長(zhǎng)石)的頻譜圖。首先利用巖石圖像灰度統(tǒng)計(jì)函數(shù)存在多個(gè)極小值的特點(diǎn),將其灰度級(jí)劃分成若干個(gè)子區(qū)間,并利用迭代算法對(duì)區(qū)間進(jìn)行優(yōu)化,根據(jù)優(yōu)化所得區(qū)間來(lái)建立起各類(lèi)的樣本集及其分布域。然后用小波理論對(duì)圖像進(jìn)行多重分解,按塔式原則將其各級(jí)系數(shù)矩陣還原成與原圖像大小一致的矩陣,并對(duì)各矩陣進(jìn)行均一化處理,經(jīng)處理之后的小波系數(shù)矩陣為圖像的波段。最后,以樣本集為基樣本,求出小波分解的各級(jí)分解系數(shù)與對(duì)應(yīng)點(diǎn)的坐標(biāo)集及其分解系數(shù)集,利用Bayes算法建立花崗巖和砂巖中各成份的頻譜圖。本文中頻譜圖是建立在先驗(yàn)基礎(chǔ)之上的,在對(duì)頻譜圖的應(yīng)用時(shí),只需將一幅圖片進(jìn)行小波分解,同時(shí)對(duì)分解系數(shù)做還原及均一化處理,根據(jù)先驗(yàn)所得的頻譜對(duì)樣本進(jìn)行計(jì)算,便可確定出被分析圖像的各種成份及其分布情況。
【關(guān)鍵詞】小波分析;巖石圖像分類(lèi);頻譜;波段;樣本;樣本集;聚類(lèi)中心
【Abstract】Based on the characteristics of granite and sandstone digital images, this paper builds up the spectrum of several components (mica, quartz and feldspar) in rock by wavelet analysis theory and Bayes decision theory. Firstly, the gray level is divided into several subintervals by using the gray level statistical function of the rock image. The iterative algorithm is used to optimize the interval. According to the optimized range, the sample sets are set up. Its distribution domain. Then the wavelet is used to decompose the image, and the matrix of the coefficients is reduced to the same size as the original image according to the tower principle, and the matrix is processed uniformly. The wavelet coefficients matrix after processing is the band of the image The Finally, the spectral set of granite and sandstone is established by Bayes algorithm, and the spectral set of the decomposition coefficient of the wavelet decomposition and the corresponding coordinate set and its decomposition coefficient are obtained. In this paper, the spectrum is based on the transcendental basis, in the application of the spectrum, only a picture of the wavelet decomposition, while the decomposition factor to do the reduction and uniform processing, according to a priori spectrum pairs The samples are calculated to determine the various components of the image being analyzed and their distribution.
【Key words】Wavelet analysis;Rock image classification;Spectrum;Band;Sample;Sample set;Clustering center
1. 前言
(1)對(duì)于圖像的分類(lèi),過(guò)去有很多學(xué)者對(duì)此做了很多的研究。在傳統(tǒng)方法上,人們利用對(duì)象與圖像背景之間的差別來(lái)識(shí)別對(duì)象,這些差別主要體現(xiàn)在圖像函數(shù)f(x)的一階導(dǎo)數(shù)和梯度沿圖像邊緣切線(xiàn)方向變化的趨勢(shì)較緩,而沿垂直圖像邊緣方向的變化趨勢(shì)較陡,經(jīng)典的算法有:Roberts算子、Prewitt算子、Sobel算子、LOG算子、Canny算子等[1]。另外在利用邊緣檢測(cè)與圖像的數(shù)學(xué)形態(tài)學(xué)相接合,也能較好的識(shí)別出圖像中的對(duì)象[2,3]。
(2)巖土材料是由不同成份的物質(zhì)組成,它們緊密交織,且其間隙十分小,因此邊緣檢測(cè)和數(shù)學(xué)形態(tài)學(xué)很難將它們分離出來(lái)[4,5]。Seungcheol Shin等[6]人利用小波理論對(duì)巖土材料進(jìn)行分解,并用各級(jí)分解系數(shù)的能量特征,對(duì)土顆粒的尺度進(jìn)行研究,取得了較好的效果。但是這一方法沒(méi)有考慮到圖像中不同物質(zhì)成份的概率分布。
(3)花崗巖和砂巖圖像的灰度統(tǒng)計(jì)函數(shù)存在著若干個(gè)極小值,以這些極小值點(diǎn)為分界點(diǎn),將灰度函數(shù)的定義域分成若干個(gè)子區(qū)間,用迭代算法對(duì)各區(qū)間進(jìn)行優(yōu)化;以?xún)?yōu)化所得子區(qū)間為依據(jù),將像素值屬于同一區(qū)間的點(diǎn)歸為一類(lèi);求出各類(lèi)的分布區(qū)域、聚類(lèi)中心等參數(shù)。同時(shí)用小波分解將圖像沿垂直、水平、對(duì)角三個(gè)方向分解;按塔式放大原則[7]將各級(jí)分解系數(shù)矩陣還原成與原始圖像大小相同的矩陣;根據(jù)所得分布域,把所有放大后的系數(shù)矩陣劃分成若干個(gè)子域;將某系數(shù)矩陣中位于同一分布域的點(diǎn)集視為類(lèi)的一個(gè)波段。最后利用Bayes算法[8~10]求出每一類(lèi)的先驗(yàn)概率及其判別式方程,將方程中未知項(xiàng)視為某一類(lèi)成份的頻譜。本文對(duì)若干個(gè)花崗巖和砂巖圖像進(jìn)行分析,求出各類(lèi)成份的頻譜,便可基于這些頻譜對(duì)花山崗巖和砂巖進(jìn)行成份分析。