【正文】 ge strength (typically the gradient magnitude), the next stage is to apply a threshold, to decide whether edges are present or not at an image point. The lower the threshold, the more edges will be detected, and the result will be increasingly susceptible to noise, and also to picking out irrelevant features from the image. Conversely a high threshold may miss subtle edges, or result in fragmented edges. If the edge thresholding is applied to just the gradient magnitude image, the resulting edges will in general be thick and some type of edge thinning postprocessing is necessary. For edges detected with nonmaximum suppression however, the edge curves are thin by definition and the edge pixels can be linked into edge polygon by an edge linking (edge tracking) procedure. On a discrete grid, the nonmaximum suppression stage can be implemented by estimating the gradient direction using firstorder derivatives, then rounding off the gradient direction to multiples of 45 degrees, and finally paring the values of the gradient magnitude in the estimated gradient direction. A monly used approach to handle the problem of appropriate thresholds for thresholding is by using thresholding with hysteresis. This method uses multiple thresholds to find edges. We begin by using the upper threshold to find the start of an edge. Once we have a start point, we then trace the path of the edge through the image pixel by pixel, marking an edge whenever we are above the lower threshold. We stop marking our edge only when the value falls below our lower threshold. This approach makes the assumption that edges are likely to be in continuous curves, and allows us to follow a faint section of an edge we have previously seen, without meaning that every noisy pixel in the image is marked down as an edge. Still, however, we have the problem of choosing appropriate thresholding parameters, and suitable thresholding values may vary over the image. Some edgedetection operators are instead based upon secondorder derivatives of the intensity. This essentially captures the rate of change in the intensity gradient. Thus, in the ideal continuous case, detection of zerocrossings in the second derivative captures local maxima in the gradient. We can e to a conclusion that,to be classified as a meaningful edge point,the transition in gray level associated with that point has to be significantly stronger than the background at that we are dealing with local putations,the method of choice to determine whether a value is “significant” or not id to use a we define a point in an image as being as being an edge point if its twodimensional firstorder derivative is greater than a specified criterion of connectedness is by definition an term edge segment generally is used if the edge is short in relation to the dimensions of the key problem in segmentation is to assemble edge segments into longer alternate definition if we elect to use the secondderivative is simply to define the edge ponits in an image as the zero crossings of its second definition of an edge in this case is the same as is important to note that these definitions do not guarantee success in finding edge in an simply give us a formalism to look for derivatives in an image are puted using the derivatives are obtained using the Laplacian. 數(shù)字圖像處理與邊緣檢測(cè) 數(shù)字圖像處理 數(shù)字圖像處理方法的研究源于兩個(gè)主要應(yīng)用領(lǐng)域：其一是為了便于人們分析而對(duì)圖像信息進(jìn)行改進(jìn)：其二是為使機(jī)器自動(dòng)理解而對(duì)圖像數(shù)據(jù)進(jìn)行存儲(chǔ)、傳輸及顯示。 Digital Image Processing and Edge Detection Digital Image Processing Interest in digital image processing methods stems from two principal applica tion areas: improvement of pictorial information for human interpretation。 at a smooth object edge。數(shù)字圖像處理是指借用數(shù)字計(jì)算機(jī)處理數(shù)字圖像，值得提及的是數(shù)字圖像是由有限的元素組成的，每一個(gè)元素都有一個(gè)特定的位置和幅值，這些元素稱為圖像元素、畫面元素或像素。它們可以對(duì)非人類習(xí)慣的那些圖像源進(jìn)行加工，這些圖像源包括超聲波、電子顯微鏡及計(jì)算機(jī)產(chǎn)生的圖像。我們認(rèn)為這一定義僅是人為界定和限制。人工智能領(lǐng)域處在其發(fā)展過程中的初期階段，它的發(fā)展比預(yù)期的要慢的多，圖像分析（也稱為圖像理解）領(lǐng)域則處在圖像處理和計(jì)算機(jī)視覺兩個(gè)學(xué)科之間。低級(jí)處理是以輸入、輸出都是圖像為特點(diǎn)的處理。 根據(jù)上述討論，我們看到，圖像處理和圖像分析兩個(gè)領(lǐng)域合乎邏輯的重疊區(qū)域是圖像中特定區(qū)域或物體的識(shí)別這一領(lǐng)域。理解一頁的內(nèi)容可能要根據(jù)理解的復(fù)雜度從圖像分析或計(jì)算機(jī)視覺領(lǐng)域考慮問題。在今天的應(yīng)用中，最主要的圖像源是電磁能譜，其他主要的能源包括聲波、超聲波和電子（以用于電子顯微鏡方法的電子束形式）。如果光譜波段根據(jù)光譜能量進(jìn)行分組，我們會(huì)得到下圖 1 所示的伽馬射線（最高能量）到無線電波（ 最低能量）的光譜。通常，圖像獲取包括如設(shè)置比例尺等預(yù)處理。應(yīng)記住，增強(qiáng)是圖像處理中非常主觀的領(lǐng)域，這一點(diǎn)很重要。另一方面，增強(qiáng)以怎樣構(gòu)成好的增強(qiáng)效果這種人的主觀偏愛為基礎(chǔ)。 小波是在各種分辨率下描述圖像的基礎(chǔ)。雖然存儲(chǔ)技術(shù)在過去的十年內(nèi)有了很大改進(jìn)，但對(duì)傳輸能力我們還不能這樣說，尤其在互聯(lián)網(wǎng)上更是如此，互聯(lián)網(wǎng)是以大量的圖片內(nèi)容為特征的。 分割過程將一幅圖 像劃分為組成部分或目標(biāo)物。通常，分割越準(zhǔn)確，識(shí)別越成功。當(dāng)注意的 焦點(diǎn)是外部形狀特性（如拐角和曲線）時(shí)，則邊界表示是合適的。為了描述數(shù)據(jù)以使感興趣的特征更明顯，還必須確定一種方法。 到目 前為止，還沒有談到上面圖 2 中關(guān)于先驗(yàn)知識(shí)及知識(shí)庫與處理模塊之間的交互這部分內(nèi)容。除 了引導(dǎo)每一個(gè)處理模塊的操作，知識(shí)庫還要控制模塊間的交互。盡管在任何關(guān)于分割的討論中，點(diǎn)和線檢測(cè)都是很重要的，但是邊緣檢測(cè)對(duì)于灰度級(jí)間斷的檢測(cè)是最為普遍的檢測(cè)方法。在邊線的每一邊都有一個(gè)邊緣。在這個(gè)例子中，我們的數(shù)據(jù)是一行不同點(diǎn)亮度的數(shù)據(jù)。實(shí)際上，這也是為什么邊緣檢測(cè)不是一個(gè)簡(jiǎn)單問題的原因之一。 已發(fā)表的邊緣檢測(cè)方法應(yīng)用計(jì)算邊界強(qiáng)度的度量， 這與平滑濾波有本質(zhì)的不同 . 正如許多邊緣檢測(cè)方法依賴于圖像梯度的計(jì)算， 他們用不同種類的濾波器來估計(jì) x方向和 y方向的梯度 . 一旦我們計(jì)算出導(dǎo)數(shù)之后，下一步要做的就是給出一個(gè)閾值來確定哪里是邊緣位置。然而非最大抑制的邊緣檢測(cè)，邊緣曲線的定義十分模糊，邊緣像素可能成為邊緣多邊形通過一個(gè)邊緣連接（邊緣跟蹤）的過程。首先使用一個(gè)閾值上限去尋找邊線開始的地方。 其它一些邊緣檢測(cè)操作是基于亮度的二階導(dǎo)數(shù)。如上所述，邊線是雙重邊緣，這樣我們就可以在邊線的一邊看到一個(gè)亮度梯度，而在另一邊看到相反的梯度。由于我們用局部計(jì)算進(jìn)行處理，決定一個(gè)值是否有效的選擇方法就是使用門限。如果我們選擇使用二階導(dǎo)數(shù)，則另一個(gè)可用的定義是將圖像中的邊緣點(diǎn)定義為它的二階導(dǎo)數(shù)的零交叉點(diǎn)。 指導(dǎo)老 師：汪濟(jì)洲 06 電子 (1)班 劉崴 (0605072021)