外文文獻及翻譯--一種有效地自動圖像增強方法

1、<p><b>　　附錄A：外文文獻</b></p><p>　　An Effective Automatic Image Enhancement Method</p><p>　　ABSTRACT Otsu method is proper to deal with two conditions: (1) two or more classes with

2、distintive gray-values respectively; (2) classes without distinctive gray-values, but with similar areas. However, when the gray-value differences among classes are not so distinct, and the object is small relative to ba

3、ckgroud, the separabilities among classes are insufficient. In order to overcome the above problem, this paper presents an improved spatial low-pass filter with a parameter and presents an un</p><p>　　KEYWOR

4、DS image processing; automated image enhancement; image segmentation; automated visual inspection</p><p>　　1 Introduction</p><p>　　Automated visual inspection of cracked container (AVICC) is a p

5、ractical application of machine vision technology. To realize our goal, four essential operations must be dealt with – image preprocessing, object detection, feature description and final cracked object classification. I

6、mage enhancement is to provide a result more suitable than original image for specific applications. In this paper the objective of enhancement, followed by image segmentation, is to obtain an image with a higher cont<

7、;/p><p>　　Segmentation is discussed in [8]. The most simplest, represented by Otsu [9], is method using only the gray level histogram analysis to maximize the separability of the resultant classes. Kuntimad [10

8、] describes a method for segmenting digital images using pulse coupled neural networks (PCNN). Salzenstein [11] deals with a comparison of recent statistical models on fuzzy Markov random fields and chains for multispect

9、ral image segmentation. Due to ill-defined, there is no unique segmentation of a</p><p>　　In this paper, we present an improved spatial low-pass filter with a tunable parameter in the mask making all element

10、s no longer sum to unity. The optimal parameter for the filter can be determined by the improved discriminant criterion based on the one mentioned in [9]. Convolving images with this mask, the background uninteresting ca

11、n be removed easily leaving the object intact to some extent. The remainder of the paper is organized as follows: Sect.2 presents how to enhance an input image in </p><p>　　2　Image Enhancement</p><

12、;p>　　2.1 Analysis of Prior Knowledge</p><p>　　The preprocessing quality influences the latter work directly, in that, feature description. Therefore, analysis for the characteristics related to input imag

13、es should be presented. A standard image of cracked container is shown as Fig.1 (a). From the image, we see the cracked part occupies small region. Much noise, such as rust, shadow, smear etc, appears within the backgrou

14、nd. At a coarse glance, however, we find gray level of the hole is less than the other parts distinctly. Further study sho</p><p>　　Fig.1 (a) is a standard gray level image of a cracked container(b) is the

15、histogram of Fig.1 (a), indicating gray level region of the hole’s edge. </p><p>　　2.2 Formulation</p><p>　　This section discusses the principal content in the paper. Traditional spatial filter

16、uses a 3×3 mask, the elements of which sum to unity, to convolve with the input image. This method can deal with some cases shown in equation (1):</p><p><b>　?。?）</b></p><p>　　w

17、here, I is image interested, N is Gaussian white noise, (x,y) denotes each pair of coordinates. N can be deliminated by blurring G. Our objective, however, is to deliminate not only white noise, but any other background

18、uninteresting. Thus equation (1) is improved by equation (2):</p><p><b>　?。?）</b></p><p>　　where, I' is the object, N' consists of white noise and the other parts except I

19、9;. Fig.2 (c) displays an improved mask with a parameter Para. We will later illustrate that tuning Para properly is to facilitate object segmentation. The smoothing function used is shown in equation (3):</p><

20、;p><b>　　（3）</b></p><p>　　where, F(x,y) denotes the smoothing filter, in that, the mask shown as Fig.2 (c). </p><p>　　Now, we only consider gray-level images, and define Mg as the

21、maximum gray level of an image. Then the following equations are set to distinguish the object of interest and the non-object :</p><p><b>　?。?） </b></p><p>　　In essence, convolu

22、tion operator is a low-pass filtering process, which blurs an image by sliding a mask through the image and leaves the filtering response at the position corresponding to central location of the mask. One question occurs

23、 that, why not enhance value of each pixel by the same scale directly for the distinct gray levels between the object and background. The reason is that it doesn’t consider the relationship of adjacent pixels. When indiv

24、idual noise point occur, enhancing its gr</p><p>　　Now, we will search the optimal parameter Para so as to maximize the separability between object and background. Let a given image be represented in L gray

25、levels. The number of pixels at level i is denoted by ni and the total number of pixels by N. The probability of each level is denoted by Pi as follow [9]:</p><p><b>　　（5） </b></p><

26、p>　　Suppose that we partition the pixels into two classes C0 and C1 (object and background) by a threshold at level k; C0 denotes pixels with levels [1, … , k], and C1 denotes pixels with levels [k+1, … , L]. Then the

27、 probabilities of class occurrence w0,w1 and the class mean levels u0,u1 respectively,are given by</p><p><b>　?。?）</b></p><p><b>　　（7）</b></p><p><b>　

28、　（8）</b></p><p><b>　?。?）</b></p><p><b>　　（10）</b></p><p><b>　?。?1）</b></p><p><b>　?。?2）</b></p><p>　　

29、The procedure of obtaining optimal para is based on obtaining optimal threshold for every filtered image. The optimal threshold is determined by maximizing the separability between object and background using the followi

30、ng discriminant criterion measure as mentioned in [9] :</p><p><b>　　（13）</b></p><p><b>　　where</b></p><p><b>　?。?4）</b></p><p>　　and

31、are the between class variance and the total variance of levels,respectively.</p><p><b>　?。?5）</b></p><p>　　The optimal threshold k* that maximizes n is selected in the following seq

32、uential search by using equation (5)-(14):</p><p><b>　　（16）</b></p><p>　　Equation (16) is a discriminant criterion to select the gray level to maximize the separability between objec

33、t and background for a given picture. In this paper, a parameter Para is introduced, so the equations (6)~(9), (11)~(14), (16) is parameterized by Para and k and equations (10), (15) is parameterized by Para. Equation (1

34、3) can be rewritten as:</p><p><b>　?。?7）</b></p><p>　　Where is not a constant any more and is not negligible, but some computation reduction can be operated onand</p><p&g

35、t;　　Here, what we want to acquire is the proper filtered picture including vivid object by searching parameter Para, the discriminant criterion used is improved as follow :</p><p><b>　?。?8）</b>&l

36、t;/p><p>　　In the above representation, parameter Para plays an important role, because optimal Para makes the separability between object and background maximal, and make Otsu segmentation method effective to

37、segment small object from large background without distinctive gray-value between them, which can be observed later from image histogram after image enhancement</p><p>　　2.3 Existence Discussion of Para and

38、k*</p><p>　　The problem above is reduced to search for a threshold k* under the condition of Para which maximizes the discriminant criterion in equation (18). The condition discussed is the image with two cl

39、ass at least. Subsequently, the following two cases don’t occur, in that,(1) w0 or w1 is zero originally without setting Para,in which there is only one class;(2) w0 or w1 is zero with certain increasing Para,in which th

40、ere is also one class finally;</p><p>　　The above two cases are decribed as:</p><p>　　The case concerned is A,Thus,there is certain Para with proper k to make discriminant criterion maximal.<

41、/p><p>　　3 Experiments</p><p>　　This paper aims at monochrome images. First, the initial values are presented. Several values should be set: Para = 1/9 (beginning with an averaging filter for 3*3 m

42、ask),Mg=L=256 (the range of gray-level, shown in equation (4) and (5)). Using the algorithm above, we can compute each value of discriminant criterion (k*) computed, in that image I 'f that is most proper to be segme

43、nted is obtained. Here, we take images of cracked container for example. Fig.3 and Fig.4 show the experimental process,</p><p>　　4　Conclusion </p><p>　　This paper is to overcome the disadvantage

44、 of Otsu method in dealing with the condition: when the gray-value differences among classes are not so distinct, and the object is small relative to backgroud, the separabilities among classes are not sufficient. This p

45、aper proposes an effective image enhancement method in spatial domain. We define all the non-objects as noise, which urges us to design an effective filter to remove noise at one time. We propose an improved mask, accord

46、ing to the charact</p><p><b>　　譯文：</b></p><p>　　一種有效地自動圖像增強方法</p><p><b>　　1.簡介</b></p><p>　　基于集裝箱裂紋的自動視覺檢測(AVICC)是一個應用機器視覺技術。要實現(xiàn)我們的目標,必須使用四種基本的操

47、作-圖像預處理、目標檢測、特征描述和破裂對象最終分類。圖像增強是為了提供一個比原始圖像中的具體應用更合適的結果。這篇關于圖像增強文章的主要目的是獲得較高質量的感興趣的內容而最大的減少噪聲。岡薩雷斯所論述圖像增強方法分為兩大類：空間域和頻率域的方法。伯頓適的人臉識別系統(tǒng)圖像應用技術,使它能夠識別變化大面孔。Centeno提出了一種自適應圖像增強算法,該算法改變了分割圖像增強和銳化的缺點,避免了噪音和模糊邊界。Munteanu應用人工智能技

48、術,提供去噪圖像增強的功能。除了空間域方法、頻域處理的相似度也可以用來做定量比較圖像分割算法。</p><p>　　在本文中,我們提出了一種改進的空間低通濾波器。最優(yōu)參數(shù)濾波的判別準則確定見參考書[9]。這款面膜Convolving圖像,能夠簡易地去除不感興趣的背景而使感興趣的部分被保留。其余的組織提出如下:Sect.2提出了如何提高一個輸入圖像在理論的基礎的算法。Sect.3證明了Sect.2方法的有效性。最后

49、,在Sect.4提出了相關結論。</p><p><b>　　2.圖像增強</b></p><p>　　2.1對所學知識回顧</p><p>　　預處理后圖像的質量直接影響之后的工作。因此,應該給出輸入圖片的相關特性。一個標準的容器破裂圖如(A)所示。從圖像中,我們看到破裂的部分只占一個小區(qū)域。在圖片中出現(xiàn)了如鐵銹、陰影、涂片等許多噪聲。然而,

50、在粗糙的反光下我們發(fā)現(xiàn)灰度洞口比其他部分模糊。進一步研究圖像的灰度像素可以看出邊上的孔洞像素最小。</p><p>　　a)是有裂紋容器的標準灰度圖像。 b)是圖一的直方圖</p><p><b>　　2.2分析</b></p><p>　　本節(jié)主要是介紹基本的內容。傳統(tǒng)的空間過濾器使用一個3×3的模板與輸入圖像

51、進行卷積。該方法可以處理一些適用方程(1)的圖像: </p><p><b>　?。?）</b></p><p>　　I是我們感興趣的部分,N是高斯白噪聲,(x,y)表示一對坐標。通過得到G我們可以消除N .但我們的目的不僅是消除白噪聲,而且要消除其他不相關的背景噪聲。因

52、此通過方程(2)改善方程(1):</p><p><b>　?。?）</b></p><p>　　在這方程式里面I’是我們想要得到的,N '是噪聲。圖2(c)顯示一個改進的模板參數(shù)。我們稍后會說明適當調整是為了促進對象的分割。光滑函數(shù)可以用方程(3)來表示:</p><p>　?。?）

53、 </p><p>　　F(x,y)表示平滑濾波，模板顯示如圖2（c）。現(xiàn)在,我們只考慮灰度圖像,并定義Mg為一個最大灰度級。下列方程是用來區(qū)分感興趣和不敢興趣部分:</p><p>　?。?） </p><p>　　本質上,卷積算子是一個低通濾波過程,通過一個模

54、板與圖像卷積使圖像模糊。但為什么會使每一個像素的灰度值不相同幅度的提高呢？。原因是它不考慮相鄰像素間的關系。當噪聲點發(fā)生,提高其灰度，噪聲點將直接保存。實驗說明后者的方法將不能刪除許噪聲點,但前者方法可以。</p><p>　　現(xiàn)在,我們將搜索最優(yōu)參數(shù)最大限度地分離目標體和背景。并用用L灰度水平描繪一個圖像。處于i灰度級的像素數(shù)目被改成n灰度級水平每一灰度級描述如【9】所示:</p><p&g

55、t;<b>　?。?） </b></p><p>　　假設我們將圖像的像素分成兩組C0和C1（客體和背景）,K為分界點，C0表示1至K，C1表示K+1至L，可能頻率分別為w1和w2，灰度級分別用u0和u1表示，方程式如下：</p><p><b>　?。?）</b></p><p><b>　?。?）</

56、b></p><p><b>　?。?）</b></p><p><b>　　（9）</b></p><p><b>　?。?0）</b></p><p><b>　?。?1）</b></p><p><b>　　（

57、12）</b></p><p>　　其中ut是圖像像素的總平均值</p><p><b>　?。?3）</b></p><p><b>　　（14）</b></p><p><b>　　其中和都變量</b></p><p><b>

58、　?。?5）</b></p><p>　　獲得最佳效果的程序是基于為每一個過濾的圖象獲得最佳閾值。確定最優(yōu)閾值最大的使得物體和背景分離，使用下列判別準則，詳見[9]:</p><p><b>　?。?6）</b></p><p>　　方程(16)是選取灰度判別準則使一張圖片目標體和背景之間最大的分離。本文介紹了一個參數(shù),方程(6)~

59、(9),(11)~(14),(16)是參數(shù)化方程：(10),(15)是參數(shù)化條件。所以方程(13)可以改寫為:</p><p><b>　?。?7）</b></p><p>　　其中不再是一個常數(shù)，但不能忽視，一些計算可以通過和簡化。</p><p>　　我們想找到適合的參數(shù)使圖像過濾后得到更好的效果，改善判別標準如下：</p>

60、<p><b>　?。?8）</b></p><p>　　在上面的表達式中參數(shù)設置非常重要,因為最優(yōu)參數(shù)能最大的分離對象和背景,使得最后能進行有效的分割,這使得閾值分割法能更有效地從大背景中分離小目標體，這可以從增強后的圖像直方圖觀察到</p><p>　　2.3對Para和k的分析</p><p>　　以上問題尋求降低閾值k *情況

61、下,這使得在方程(18)中判別準則最大化。討論以上圖像至少要分為兩種情況。但以下兩種情況下不發(fā)生,因為（1）w0或者w1是初始值為0，這種情況下只有一類；(2)w0或者w1沒有確定的數(shù)值，在這種情況下也只有一類。以上兩種情況可如下描述：</p><p>　　這里主要是討論A，所以必須有一個確定的參數(shù)K使得標準最大化。</p><p><b>　　3.實驗</b><

62、;/p><p>　　本文是針對單色圖像的，首先初始值已經給出。其他有些值需設定：Para=1/9,Mg=L=256,使用上述算法我們可以計算K*的每個值和與之對應參數(shù)para的值，通過對比計算得到最優(yōu)K*值，在這幅圖像中,I‘f最佳的分割。在此,我們以拍攝的容器破裂圖像為例子。圖3、圖4顯示實驗過程中,第一的照片顯示經過濾波后的圖片第二行顯示和相應的直方圖第三行顯示曲線相應的判別標準。最后的是柱優(yōu)化后的圖像增強效果,

63、從中我們可以看到許多噪音如鐵銹、陰影、涂片等幾乎都被移除了,而斷裂的部分幾乎完好無損。Tab.1 Tab.2展示了不同的K * 和Para值，得到的不同結果如圖3和圖4。當P增加到5/9時,K*能計算得到最好的效果。當P不斷增加,K *會降低，并且裂紋部分將被嚴重破壞，如圖5所示。</p><p><b>　　4.結論</b></p><p>　　本文是為了處理當灰度

64、值之間的差異是不那么明顯，對象相對背景非常小時閾值分割法的缺點。本文提出了一種有效的空間域圖像增強方法。我們認為所有非對象都是噪音，這使得我們需設計一個一次性有效的濾波去除噪聲的方法。我們提出了一種改進的模板，根據(jù)破裂容器的灰度值特點，需使非目標體灰度值高于閾值并且使目標體的灰度值低于閾值。被最佳分割而得到的過濾后的圖像，可以采用改進的判別準則自動最大限度地分離感興趣和不感興趣的部分。在進行特定的增強后，隨后的操作就非常輕松了。實驗表明