【正文】 結(jié)及深入分析圖像模型的基礎(chǔ)上，基于陰影檢測(cè)的基本假設(shè)和一般框架，針對(duì)不同的應(yīng)用要求設(shè)計(jì)了一種算法，并取得了較好的效果。1）基于邊緣信息的陰影抑制算法適用于目標(biāo)體邊緣信息豐富，陰影邊緣信息相對(duì)簡(jiǎn)單的陰影去除。針對(duì)上述缺陷，后續(xù)要做的工作有：研究變化背景下的目標(biāo)對(duì)象分割技術(shù)，使用圖像配準(zhǔn)技術(shù)來(lái)糾正攝像機(jī)擺動(dòng)而造成的誤差；深入地分析產(chǎn)生陰影的光學(xué)原理、各種材料的反射特性、以及材料呈現(xiàn)不同顏色的物理機(jī)理，為各光源發(fā)射光線、場(chǎng)景表面材料反射光線、人眼接收光線、大腦產(chǎn)生場(chǎng)景顏色的反應(yīng)等建立合適的模型，有可能弄清陰影的本質(zhì)特性，從而充分運(yùn)用目標(biāo)的顏色信息，提高陰影的抑制效果。參考文獻(xiàn)1.《MATLAB使用詳解》 董霖 電子工業(yè)出版社 2. MATLAB數(shù)字圖像處理／張德豐等編著．—北京：機(jī)械工業(yè)出版社，20093. 楊林 圖像分割及陰影抑制算法的研究. 20074. （上冊(cè)）：清華大學(xué)出版社 20065. 參考網(wǎng)站：程序員聯(lián)合開發(fā)網(wǎng) 基于混合高斯模型的運(yùn)動(dòng)陰影抑制算法研究2006 高斯核密度估計(jì)背景建模及噪聲與陰影抑制 2005 視頻分割技術(shù)研究 200710 Shafarenko L,Petrou M,Kittler Segmentation in aPerceptually Uniform Color Trans on Image 11 Vlachos T,Constantinides A Graphtheoretic Approach to Color ImageSegmentation and Contour 4th on ImageProcessing and its 12 Loncaric Survey of Shape Analysis Recognition.200113席礪莼,，2003致 謝衷心感謝欒慶磊指導(dǎo)老師在畢業(yè)設(shè)計(jì)期間給予的有力支持和無(wú)微不至的關(guān)懷。在課題的研究期間，欒老師為課題研究所提出的寶貴指導(dǎo)意見，是保證課題順利完成的關(guān)鍵，也使我豐富了知識(shí)，開闊了視野。欒老師的悉心指導(dǎo)及日常生活中的無(wú)私幫助是本課題得以順利進(jìn)行的堅(jiān)強(qiáng)后盾。在此，我對(duì)欒老師表示誠(chéng)摯的謝意。另外還要感謝曾經(jīng)給予我關(guān)心和幫助的所有同學(xué)。在我遇到困難時(shí)，是他們不斷地同我探討和實(shí)驗(yàn)。在他們的無(wú)私幫助下，本次的畢業(yè)設(shè)計(jì)得以順利完成。是老師的教誨，同學(xué)的幫助，集體的溫暖和力量使我得以順利完成了學(xué)習(xí)任務(wù)。再次誠(chéng)摯地謝謝大家！附 錄一Location and edge information based on shadow shadow removal algorithm CHEN Hua 1, Guan Yu Dong 2, Lin Yang 2, Han Wendy Law 2, Xu Xiao Yu Abstract This paper presents the location and based on the shadow of the shadow edge removal algorithm. First, the estimated light direction to determine the direction of the shadow。 Then, based on edge detection to eliminate the outline of the shadow edge, using the extracted foreground image minus the prospect of building, by the shadow area。 Finally, the more the shadow direction, the direction angle if a certain threshold value range, pared with the actual shadows. The results show that removal of the shadow method is rapid and accurate. Key words shadow removal。 edge detection。 shadow direction 1 Introduction Target detection and segmentation for the target image, usually containing the shadow of the human eye can easily distinguish the moving target with the goal of projecting moving shadows, but how to make puter identify the movement shadow is a challenging research problem [12 ]. On shadow detection and suppression directly affect the video image postprocessing, such as target classification, tracking and understanding behavior in such processes [3], also seriously affected the intelligent monitoring system, intelligent transportation system (Intelligent Transportation Systems, ITS), etc. Computer Vision System and promotion. In the field of film and television production, the removal of the shadow can be separated from the background characters and other flexible synthetic scene of great significance. How to better prospects in the detected target, and it casts a shadow separated from the domestic and foreign scholars has bee a hot issue, but also focus on the content of this article. Existing shadow suppression methods fall into two categories, namely the shadow suppression algorithm based on attributebased model of shadow suppression algorithm. Attributebased approach is the use of the geometric characteristics of the shadow, brightness, color and other information to identify the shadow region, the existing shadow suppression algorithm using the shadow spectral properties than the more mon geometric features, shadow suppression algorithm based on attributes of different scenes and lighting conditions There is robust。 modelbased approach is the use of scenes, moving objects, light conditions of prior information, the establishment of the shadow model, threedimensional model of moving object edge, line, angle to match, this method is usually in a specific scenario under the conditions of use, such as aerial image understanding and video surveillance. Although there are a variety of shadow suppression algorithm has been there, but existing algorithms are only scenario applies to a specific application, how to target a shadow and Yun Dong effective segmentation, Kaifa out of the shadow suppression algorithm Bijiao general is still challenging the problem of a rich . Shadow Removal Algorithm 2 Remove the shadow of the edge information [4], adapted to the target of the edge informationrich, relatively simple shadow of the shadow edge removal. For the shadow of Figure 1 Inhibition of the algorithm simulation result is not satisfactory. This article made the following improvements on such issues. Figure 1 target body and shadow determine the location of the shadow Determine the location of the shadow in the image can be better to remove the shadow. To this end, we must first estimate illuminant direction [1]. Specific algorithm is as follows: 1) edge detection Have been divided on good edge detection, extraction of edge pixels. 2) The division of light and dark pixels to determine the threshold Here this paper, the classical Otsu method to determine the threshold of the extracted edge pixels to operate, greater than the threshold value is the bright pixels, or a dark pixel. 3) the image center of gravity G (! X, y!) Prevail, the image is divided into upper and lower part of about 4, there are many ways to strike the center of gravity, this paper cvMoments (OpenCV usage) method to obtain focus. 4) that u, d, l, r, respectively, up and down about 4 directions. 5) N = (u + d + r + l) / 4。 x, y is the u, d, r, l the largest two values. 6) define the direction of light 8, respectively, said digital 0,1,2,3,4,5,6,7. 7) a, b correspond to x, y values. 8) A = (a + b) / 2, B = ab. 9) is defined as the direction of light If (x, y N) or (B = 4), then o = l。 If (x ≥ N, y N), then o = a。 If (x N, Y ≥ N), then o = b。 If (B = 6), then o = B + l。 Otherwise o = A. When o = 1 when it shows the direction the shadow is not defined, the process error. The edge detection Information based on edge detection purpose is to try to eliminate the shadow of the outline of the edge, the edg