CN101853286A - Intelligent selection method of video thumbnails - Google Patents

Intelligent selection method of video thumbnails Download PDF

Info

Publication number
CN101853286A
CN101853286A CN 201010180153 CN201010180153A CN101853286A CN 101853286 A CN101853286 A CN 101853286A CN 201010180153 CN201010180153 CN 201010180153 CN 201010180153 A CN201010180153 A CN 201010180153A CN 101853286 A CN101853286 A CN 101853286A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
thumbnail
score
value
image
rgb
Prior art date
Application number
CN 201010180153
Other languages
Chinese (zh)
Other versions
CN101853286B (en )
Inventor
刘子枫
连惠城
Original Assignee
上海全土豆网络科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Abstract

The invention discloses an intelligent selection method of video thumbnails, comprising the following steps: 1) acquiring N numbered thumbnails of a video, wherein N is more than or equal to 2 and less than or equal to 100, and N is a constant; 2) calculating the characteristics of each thumbnail, such as frontal face position and size, profile face position and size, whole body position and size, upper body position and size, lower body position and size, gray level histogram distribution value, gray level histogram variance distribution value, colorful RGB histogram variance distribution value, difference value that a colorful RGB image is changed into a gray level image, the variance value of a gray level image pixel point, the variance value of the colorful RGB image pixel point, the dispersion distance value between the colorful RGB blocks, and the like; 3) calculating the characteristic values of each thumbnail respectively through ambiguity functions; 4) weighting and calculating the characteristic values of each thumbnail to obtain a final fraction; 5) sequencing the N numbered thumbnails according to the fraction of each thumbnail; and 6) selecting the thumbnail with the highest fraction as the thumbnail representing the video.

Description

视频缩略图智能选取方法 Intelligent video thumbnail selection method

技术领域 FIELD

[0001] 本发明涉及图片选取方法,特别涉及一种通过计算机图像处理进行的视频缩略图智能选取方法。 [0001] The present invention relates to a method of selecting images, and particularly relates to a video image by a computer for performing intelligent processing method of selecting a thumbnail.

背景技术 Background technique

[0002] 在视频网站的日常应用中,有一项重要的技术就是从视频文件中抽取一帧图像作为该视频文件的代表图片,我们称该图片为视频文件的缩略图。 [0002] In daily use video sites, there is an important technique to extract an image from a video file as a video file on behalf of the picture, we call the picture as a thumbnail of the video file. 目前我们的做法是从视频文件中抽取八张缩略图,并设置第四张作为默认的缩略图。 Currently, our approach is to extract the eight thumbnail images from video files, and the fourth set as the default thumbnail. 这种采用固定位置方法来默认而得到的缩略图通常不是八张缩略图中最具有代表性的缩略图。 This fixed position to the default method is usually not obtained thumbnail eight thumbnails of the most representative. 我们采用了一些方法可以解决这个问题,比如(1)用户可以通过自定义的方法来确定哪张图片作为默认的缩略图, We use a number of ways to solve this problem, such as (1) the user can determine which thumbnail images as the default method by custom,

[2]赋予网站工作人员选择哪张图片作为默认缩略图的权利。 [2] gives the site staff choose which image rights as the default thumbnails. 虽然上述方法可以解决这个问题,但是实际仍然无法满足要求,这是因为(1)通过数据分析,我们知道用户通常不会或很少自定义缩略图(2)由于每日上传到视频网站的视频文件数量十分众多,通过网站工作人员来定义缩略图的工作量十分繁重。 Although the above methods can solve this problem, but still can not meet the actual requirements, because (1) through data analysis, we know that users often little or no custom thumbnail image (2) due to daily upload video website video very many number of files, thumbnails are defined by the site staff workload is very heavy. 基于上述两个原因,我们发明一种能够通过计算机图像处理的方法来自动的选择具有代表性的缩略图作为默认的缩略图的方法。 For these two reasons, our invention is capable of automatically selecting a representative thumbnail of the thumbnail as the default method of computer image processing method.

[0003] P. Viola和M. Jones等人在2001年的计算机视觉和模式识别国际会议上(Proceedings ofComputer Vision and Pattern Recognition, 2001)提出的《米用简单特征的层叠式提升方法进行快速物体检测》(Rapid Object Detection Using a Boosted Cascade of SimpleFeatures.)方法进行检测的。 [0003] P. Viola and M. Jones et al in 2001, computer vision and pattern recognition of the International Conference on (Proceedings ofComputer Vision and Pattern Recognition, 2001) "The method of rice stacked lifting proposed wherein a simple fast object detection "(Rapid Object detection Using a Boosted Cascade of SimpleFeatures.) the method of detecting.

[0004] 有鉴于此,本领域技术人员针对上述问题,提供了一种能够通过计算机图像处理方法来自动的选择具有代表性的缩略图作为默认的缩略图的视频缩略图智能选取方法。 [0004] Accordingly, the present art for the above-described problems in the art, provided that can automatically select the representative thumbnail image processing method by a computer as a default thumbnail video thumbnail smart selection method.

发明内容 SUMMARY

[0005] 本发明提供了一种视频缩略图智能选取方法,克服了现有技术的困难,以达到通过计算机图像处理方法来自动的选择具有代表性的缩略图作为默认的缩略图的目的。 [0005] The present invention provides an intelligent method of selecting a video thumbnail, overcomes the difficulties of the prior art, to achieve the purpose of automatically selecting a representative thumbnail of the image processing method by a computer as a default thumbnail.

[0006] 本发明采用如下技术方案: [0006] The present invention adopts the following technical solution:

[0007] 本发明提供了一种视频缩略图智能选取方法,用于从一段视频中选出一张代表该视频的缩略图,包括以下步骤: [0007] The present invention provides an intelligent method of selecting a video thumbnail for a video selected from the representative thumbnail of a video, comprising the steps of:

[0008] (1)获取一段视频的N张缩略图,其中,2 100,且N是常数; [0008] (1) obtaining a video thumbnails of N, wherein, 2100, and N is a constant;

[0009] (2)对每张缩略图计算其正面人脸的位置和大小、侧面人脸的位置和大小、全身人体的位置和大小、上身的位置和大小和下身的位置和大小、灰度直方图分布值、灰度直方图方差分布值、彩色RGB直方图方差分布值、彩色RGB图像变成灰度图像的差异值、灰度图像像素点的方差值、彩色RGB图像像素点的方差值、彩色RGB块之间的离散距离值等等特征; [0009] (2) the position and size of the lower body and of calculating the frontal face of the position and size of each thumbnail, the position and size of the side face, the location and size of the body of the human body, the position and size of the upper body, the gradation histogram distribution values, the variance histogram distribution value, the variance of the color distribution histogram of RGB values, the color difference value of the RGB image into a gray image, a variance value of the pixel gray scale image, color image RGB pixels square difference discrete distance between the characteristic value and the like RGB color block;

[0010] (3)对每张缩略图分别通过模糊函数求得上述各特征的分值; [0010] (3) above are determined by the fuzzy value of each characteristic function of each thumbnail;

[0011] (4)对每张缩略图的特征分值通过加权求值得到最终的一个分数; [0011] (4) wherein each of the thumbnails to the final value is worth a fraction weighted request;

[0012] (5)根据分数的大小对N张缩略图进行排序;[0013] (6)选用分数最高的缩略图作为代表该视频的缩略图。 [0012] (5) on the N thumbnails are sorted according to the size fraction; [0013] (6) With the highest score as a thumbnail of the thumbnail images representative of the video.

[0014] 优选地,所述步骤(1)中,采用等间隔抽取方法,根据预先得到的视频时长,将它平均分成N等份,然后每一等份抽取一帧作为缩略图,且N = 8。 [0014] Preferably, the step (1), using the method of thinning the like, according to pre-obtained long video, it is equally divided into N aliquots, each aliquot was then extracted as a thumbnail, and N = 8.

[0015] 优选地,所述步骤(2)中包括以下步骤: [0015] Preferably, the step (2) comprises the steps of:

[0016] (21)输入缩略图,检测人脸特征,包括正面人脸的位置和大小、侧面人脸的位置和大小,并计算分数值; [0016] (21) an input thumbnail position detecting facial features, including a front face and size, position and size of a side face, and calculating a score value;

[0017] (22)输入缩略图,检测人体特征,包括全身人体的位置和大小、上身的位置和大小和下身的位置和大小,并计算分数值; [0017] (22) an input thumbnail, detection of human features, including the location and size of the human body, the upper body and the position and size of the position and size of the lower body, and calculating a score value;

[0018] (23)输入缩略图,计算灰度直方图分布值; [0018] (23) an input thumbnail calculated histogram distribution value;

[0019] (24)输入缩略图,计算灰度直方图方差分布值; [0019] (24) an input thumbnail calculates the variance histogram distribution value;

[0020] (25)输入缩略图,计算彩色RGB直方图方差分布值; [0020] (25) an input thumbnail RGB color histogram variances calculated distribution value;

[0021] (26)输入缩略图,计算彩色RGB图像变成灰度图像的差异值; [0021] (26) an input thumbnail calculates RGB color image into a grayscale image difference value;

[0022] (27)输入缩略图,计算灰度图像像素点的方差值; [0022] (27) an input thumbnail calculating a variance value of a gray pixel image;

[0023] (28)输入缩略图,计算彩色RGB图像像素点的方差值; [0023] (28) an input thumbnail calculating a variance value of the RGB color image pixels;

[0024] (29)输入缩略图,计算彩色RGB块之间的离散距离值。 [0024] (29) an input thumbnail calculated discrete distance values ​​between RGB color blocks.

[0025] 优选地,所述步骤(21)中: [0025] Preferably, said step (21):

'0, 当η = 0 '0, η = 0 when

[0026]则正面人脸的分值 score_face_frontal = j “ ; Scores score_face_frontal [0026] the front face = j ";

. ί=ι . Ί = ι

[0027] 其中Wi为正面人脸的宽,η为人脸个数,设W为缩略图的宽; [0027] where Wi is the width of the front face, the number of [eta] a human face, is provided as a thumbnail width W;

[0029] 其中Wi为侧面人脸的宽,η为人脸个数,设W为缩略图的宽。 [0029] where Wi is the width of the side face, the number of [eta] a human face, set width W as a thumbnail.

[0030] 优选地,所述步骤(22)中设(Xi,yi; Wi,比),i = 1,. . .,η,其中Xi,yi代表人脸位置,Wi为人脸的宽,η为人脸个数,设W为缩略图的宽, [0030] Preferably, said step (22) is provided (Xi, yi; Wi, ratio), i = 1 ,., η, wherein Xi, yi representative face location, Wi human face width, η.. the number of human faces, thumbnail width W is set,

[0031]则全身的分值 [0031] the body's scores

[0032] 其中Wi为全身人体的宽,η为人数,设W为缩略图的宽; [0032] in which the body's systemic Wi wide, η is the number, let W thumbnails wide;

[0033]上身的分值 [0033] upper body scores

[0034] 其中Wi为全身人体上肢的宽,η为人数,设W为缩略图的宽; [0034] where Wi systemic human upper limb width, η is the number, let W thumbnails wide;

[0035]下身的分值; [0035] The value of the lower body;

[0036] 其中Wi为全身人体下肢的宽,η为人数,设W为缩略图的宽。 [0036] where Wi is the width of the lower limbs of human body, η is the number, let W thumbnails wide. [0037] 优选地,所述步骤(23)中输入缩略图,设其灰度直方图为gray_hist = (gh_l, gh_2,. .,gh_n),η为直方图的个数,tl,t2为整数阈值,Tl和T2为直方图和的阈值,base_ score为基准分数,则灰度直方图分布值SCOre_hiSt_diStrib的计算公式为: Input [0037] Preferably, said step (23) of the thumbnail, which histogram is provided gray_hist = (gh_l, gh_2 ,.., Gh_n), η is the number of histogram, tl, t2 is an integer threshold, Tl, and T2 histogram and threshold value, base_ score for the benchmark score, the gray value histogram distribution is calculated SCOre_hiSt_diStrib:

[0038] [0038]

[0039] 其中,η = 10,tl = 2,t2 = 8,T1 = 0. 7,T1 = 0. 8,base_score = 1. 0。 [0039] wherein, η = 10, tl = 2, t2 = 8, T1 = 0. 7, T1 = 0. 8, base_score = 1. 0.

[0040] 优选地,所述步骤(24)中输入缩略图,设其灰度直方图为gray_hist = (gh_l, [0040] Preferably, the input of the thumbnail image in step (24), which is provided for the histogram gray_hist = (gh_l,

gh—2,· ·,gh—n),η 为直方图的个数, gh-2, · ·, gh-n), η is the number of histogram,

max—num = max(gh—1,gh—2,· ·,gh—η), max-num = max (gh-1, gh-2, · ·, gh-η),

base—score为基准分数,则计算灰度直方图方差分布值的score—gray—hist—std计算公式 base-score for the benchmark score, score-gray-hist-std histogram distribution values ​​calculated variance is calculated

为: for:

[0041] [0041]

[0042] 其中,η = 64,base—score = 1. 0。 [0042] wherein, η = 64, base-score = 1. 0.

[0043] 优选地,所述步骤(25)中分别在R平面、G平面、B平面上直方图方差分布值,然后求均值得到彩色RGB直方图方差分布值score—RGB—hist—std ; [0043] Preferably, said step (25), respectively, the variance of the histogram distribution value on the R plane, G plane, B plane and then averaging variances obtained RGB color histogram distribution value score-RGB-hist-std;

[0044] [0044]

[0045] [0045]

[0046] [0046]

[0047] [0047]

[0048] ;其中,η = 64,base_score = 1. 0。 [0048]; wherein, η = 64, base_score = 1. 0. [0049] 优选地,所述步骤(26)中输入缩略图,设灰度图像的像素值为Xij,i = 1,. . .,W, j = 1,. . .,H,彩色RGB 图像的像素值(、.,giJ, b^.),i = 1,. . .,W,j = 1,. . .,H,其中W,H 分别为缩略图的宽和高,base_scorel和baSe_SCOre2为基准分数,tl为阈值,设则计算彩色RGB图像变成灰度图像的差异值SCOre_rgb2gray_diff的计算公式为: Input [0049] Preferably, the step (26) thumbnail image pixel gray scale value provided Xij, i = 1 ,..., W, j = 1 ,..., H, RGB color image the pixel values ​​(,., giJ, b ^.), i = 1 ,..., W, j = 1 ,..., H, where W, H are the width and height of the thumbnail, base_scorel and baSe_SCOre2 for the benchmark score, tl is the threshold value, the calculated set of the RGB image into a color difference value calculated SCOre_rgb2gray_diff gradation image is:

[0050] [0050]

[0051]其中, [0051] wherein,

[0052] 优选地,所述步骤(27)中输入缩略图,设灰度图像的像素值为Xij,i = 1,. . .,W, j = 1,...,H,其中W,H分别为缩略图的宽和高,则计算灰度图像像素点的方差值的score— gray—std计算公式为: Input [0052] Preferably, said step (27) thumbnail image pixel gray scale value provided Xij, i = 1 ,..., W, j = 1, ..., H, wherein W, score- gray-std formula H are the width and height of the thumbnail image, the gradation is calculated variance value of pixels in the image:

[0053] [0053]

[0054] 优选地,所述步骤(28)中分别在R平面、G平面、B平面上图像像素点的方差值,然后求均值得到彩色RGB图像像素点的方差值score—RGB—std ; Variance [0054] Preferably, said step (28), respectively, in the R plane, G plane, the variance values ​​of the image pixels on the plane B, and then averaging the RGB image pixels to obtain a color point score-RGB-std ;

[0055] [0055]

[0056] [0056]

[0057] [0057]

[0058] [0058]

[0059] 优选地,所述步骤(29)中输入缩略图,设彩色RGB图像的像素值gi,b,), i = 1,. . .,WXH,其中W,H分别为缩略图的宽和高,base—score为基准分数,则计算彩色RGB块之间的离散距离值score—rgb—dist的计算公式为: Input [0059] Preferably, said step (29) in the thumbnail image pixel RGB color values ​​provided gi, b,), i = 1 ,..., WXH, where W, H are the width of the thumbnail and high, base-score for the benchmark score, calculated discrete distance values ​​between RGB color blocks score-rgb-dist is calculated as:

[0060] [0060]

[0061] 优选地,所述步骤(4)中最终的分值finalscore由下列公式计算得到: [0061] Preferably, the step (4) finalscore final score is calculated from the following formula:

[0062] f inal_score = W1X score_face_frontal+w2 X score_face_prof ile+w3X score_ ful l_body+w4X score_upper_body+w5X score_lower_body+w6X score_hi st_ distrib+w7X score_gray_hist_std+w8X score_RGB_hist_std+w9X score_rgb2gray_ diff+w10X score—gray—std+wnX score—RGB—std ; [0062] f inal_score = W1X score_face_frontal + w2 X score_face_prof ile + w3X score_ ful l_body + w4X score_upper_body + w5X score_lower_body + w6X score_hi st_ distrib + w7X score_gray_hist_std + w8X score_RGB_hist_std + w9X score_rgb2gray_ diff + w10X score-gray-std + wnX score- RGB-std;

[0063] 其中,W1 = 1. 0,W2 = 0. 5,W3 = 0. 5,W4 = 0. 5,W5 = 0. 5,W6 = 0. 5,W7 = 0. 5,W8=1. 2,W9 = 1. 2,W10 = 0. 5,W11 = 1. 2。 [0063] wherein, W1 = 1. 0, W2 = 0. 5, W3 = 0. 5, W4 = 0. 5, W5 = 0. 5, W6 = 0. 5, W7 = 0. 5, W8 = 1 . 2, W9 = 1. 2, W10 = 0. 5, W11 = 1. 2.

[0064] 由于采用了上述技术,与现有技术相比,本发明具有如下优点:本发明能够通过计算机图像处理方法来自动的选择具有代表性的缩略图作为默认的缩略图。 [0064] By adopting the technique, as compared with the prior art, the present invention has the following advantages: the present invention is capable of automatically selecting a representative thumbnail default thumbnail image processing method by a computer.

[0065] 以下结合附图及实施例进一步说明本发明。 [0065] The following examples further illustrate the present invention and in conjunction with the accompanying drawings.

附图说明 BRIEF DESCRIPTION

[0066] 图1为本发明的视频缩略图智能选取方法的流程图。 [0066] Figure 1 a flowchart of a thumbnail video of the present invention, the smart selection method. 具体实施方式 detailed description

[0067] 下面通过图1来介绍本发明的一种具体实施例。 [0067] Next, the present invention is to introduce a specific embodiment 1 through FIG.

[0068] 如图1所示,一种视频缩略图智能选取方法,包括以下步骤: [0068] As shown in FIG 1 A video thumbnail intelligent selection method, comprising the steps of:

[0069] (1)获取一个视频的N张缩略图,N张缩略图的获取方法是采用等间隔抽取方法, 即根据预先得到的视频时长,将它平均分成N等份,然后每一等份抽取一帧作为缩略图。 [0069] (1) Get a thumbnail video of N, N thumbnail acquisition method is the use of thinning and other methods, i.e., the length of the video obtained in advance, it is equally divided into N equal parts, and then each aliquot extracted as a thumbnail. 这里取N = 8,但不局限于N = 8,N可以在2到100之间。 Here take N = 8, but not limited to N = 8, N may be between 2-100.

[0070] (2)对每张缩略图计算其正面人脸的位置和大小、侧面人脸的位置和大小、全身人体的位置和大小、上身的位置和大小和下身的位置和大小、灰度直方图分布值、直方图方差分布值、彩色RGB直方图方差分布值、彩色RGB图像变成灰度图像的差异值、灰度图像像素点的方差值、彩色RGB图像像素点的方差值、彩色RGB块之间的离散距离值等等特征。 [0070] (2) the position and size of the lower body and of calculating the frontal face of the position and size of each thumbnail, the position and size of the side face, the location and size of the body of the human body, the position and size of the upper body, the gradation variance value histogram distribution, the variance value histogram distribution, the variance of the color distribution histogram RGB values, RGB color image into a grayscale image difference value, the variance value of pixel grayscale image, the color pixels of the RGB image , discrete distance values ​​between RGB color blocks, etc. features.

[0071] (3)分别通过模糊函数求得上述特征的分值。 [0071] (3) The above features are obtained by the fuzzy value function.

[0072] (4)通过加权求值得到最终的一个分数。 [0072] (4) It is worth to a final weighted score requirements.

[0073] (5)根据分数的大小对N张缩略图进行排序。 [0073] (5) on the N thumbnails are sorted according to size fractions.

[0074] (6)分值最高的缩略图作为默认的缩略图。 [0074] (6) the highest score as the default thumbnail thumbnail.

[0075] 上述步骤(2),按如下步骤进行: [0075] The step (2), the following steps:

[0076] (21)输入缩略图,检测人脸特征,包括有正面人脸的位置和大小、侧面人脸的位置和大小;计算分数值; [0076] (21) an input thumbnail detecting facial features, including the frontal face location and size, position and size of a side face; calculated score value;

[0077] (22)输入缩略图,检测人体特征,包括有全身人体的位置和大小、上身的位置和大小和下身的位置和大小;计算分数值; [0077] (22) an input thumbnail, detection of human features, including the location and size of the human body, the upper body and the position and size of the position and size of the lower body; calculated score value;

[0078] (23)输入缩略图,计算灰度直方图分布值; [0078] (23) an input thumbnail calculated histogram distribution value;

[0079] (24)输入缩略图,计算灰度直方图方差分布值; [0079] (24) an input thumbnail calculates the variance histogram distribution value;

[0080] (25)输入缩略图,计算彩色RGB直方图方差分布值; [0080] (25) an input thumbnail RGB color histogram variances calculated distribution value;

[0081] (26)输入缩略图,计算彩色RGB图像变成灰度图像的差异值; [0081] (26) an input thumbnail calculates RGB color image into a grayscale image difference value;

[0082] (27)输入缩略图,计算灰度图像像素点的方差值; [0082] (27) an input thumbnail calculating a variance value of a gray pixel image;

[0083] (28)输入缩略图,计算彩色RGB图像像素点的方差值; [0083] (28) an input thumbnail calculating a variance value of the RGB color image pixels;

[0084] (29)输入缩略图,计算彩色RGB块之间的离散距离值。 [0084] (29) an input thumbnail calculated discrete distance values ​​between RGB color blocks.

[0085] 本发明的实施方式如下: [0085] Embodiments of the invention are as follows:

[0086] 继续参见图1,本发明的一种视频缩略图智能选取方法,用于从一段视频中选出一张代表该视频的缩略图,包括以下步骤: [0086] With continued reference to FIG. 1, the present invention in a video thumbnail smart selection method for selecting a thumbnail image representing the video from a video, comprising the steps of:

[0087] 步骤(1)获取一段视频的N张缩略图,其中,2 < NS 100,且N是常数,采用等间隔抽取方法,根据预先得到的视频时长,将它平均分成N等份,然后每一等份抽取一帧作为缩略图,且N = 8。 Acquiring a video of the [0087] Step (1) N thumbnails, wherein, 2 <NS 100, and N is a constant, equal intervals extraction method employed, in accordance with previously obtained long video, it is equally divided into N equal parts, and then each aliquot extracted as a thumbnail, and N = 8.

[0088] 步骤(2)对每张缩略图计算其正面人脸的位置和大小、侧面人脸的位置和大小、 全身人体的位置和大小、上身的位置和大小和下身的位置和大小、灰度直方图分布值、灰度直方图方差分布值、彩色RGB直方图方差分布值、彩色RGB图像变成灰度图像的差异值、灰度图像像素点的方差值、彩色RGB图像像素点的方差值、彩色RGB块之间的离散距离值等等特征; [0088] Step (2) the position and size of the lower body and the position and size calculated on the frontal face of the position and size of each thumbnail, the position and size of the side face, the location and size of the body of the human body, the upper body, gray histogram distribution value, the variance histogram distribution value, the variance of the distribution of RGB color value histogram, RGB color image into a grayscale image difference value, the variance value of pixel grayscale image, the color pixels of the RGB image variance, discrete distance between the characteristic value and the like RGB color block;

[0089 [0090 [0089 [0090

位置和大小,并计算分数值; Position and size, and calculates a score value;

[0091 [0091

和大〃 [0092 [0093 [0094 [0095 [0096 [0097 [0098 [0099 And large 〃 [0092 [0093 [0094 [0095 [0096 [0097 [0098 [0099

所述步骤(2)中包括以下步骤: Said step (2) comprises the steps of:

步骤(21)输入缩略图,检测人脸特征,包括正面人脸的位置和大小、侧面人脸的 Step (21) input thumbnail detecting facial features, including a front face position and size of the side face

步骤(22)输入缩略图,检测人体特征,包括全身人体的位置和大小、上身的位置、和下身的位置和大小,并计算分数值; Step (22) input thumbnail, detection of human features, including the location and size of the human body, the upper position, and the position and size of the lower body, and calculating a score value;

步骤(23)输入缩略图,计算灰度直方图分布值; Step (23) input thumbnail calculated histogram distribution value;

步骤(24)输入缩略图,计算灰度直方图方差分布值; (24) an input step thumbnail calculating the variance histogram distribution value;

步骤(25)输入缩略图,计算彩色RGB直方图方差分布值; Step (25) input thumbnail RGB color histogram variances calculated distribution value;

步骤(26)输入缩略图,计算彩色RGB图像变成灰度图像的差异值; Step (26) input thumbnails calculated RGB color image into a grayscale image difference value;

步骤(27)输入缩略图,计算灰度图像像素点的方差值; Step (27) input thumbnail calculating a variance value of a gray pixel image;

步骤(28)输入缩略图,计算彩色RGB图像像素点的方差值; Step (28) input thumbnail calculating a variance value of the RGB color image pixels;

步骤(29)输入缩略图,计算彩色RGB块之间的离散距离值。 Step (29) input thumbnail calculated discrete distance values ​​between RGB color blocks.

所述步骤(2 1 )中:则正面人脸的分值 The step (21) in which: the frontal face score

'0, 当n = 0 '0, when n = 0

;其中&为正面人脸的宽,n为人脸个数,设 ; Wherein & broad number, n-human face is a front face, provided

W为缩略图的宽;侧面人脸的分值, W as thumbnails wide; the side of the face value,

[0100] 其中Wi为侧面人脸的宽,n为人脸个数,设W为缩略图的宽。 [0100] wherein the number of wide, n-Wi human face is a side face, set width W as a thumbnail.

[0101] 所述步骤(22)中设 [0101] The step (22) is provided

,...,11,其中\,71代表人脸位置,〜为人脸的宽,n为人脸个数,设W为缩略图的宽, , ..., 11, where \, 71 representative of the position of the face, the number of face width, n-man - human face, set width W as a thumbnail,

「0, 当《 = 0 "0, when the" = 0

[0102] 则全身的分值score_full—body = [0102] the body score score_full-body =

[0103] 其中Wi为全身人体的宽,n为人数,设W为缩略图的宽; [0103] in which the body's systemic Wi wide, n is the number, let W thumbnails wide;

'0, 当n = 0 '0, when n = 0

[0104]上身的分值 [0104] upper body scores

. i=l . I = l

[0105] 其中Wi为全身人体上肢的宽,n为人数,设W为缩略图的宽 [0105] where Wi systemic human upper limb width, n is the number, set the width W as thumbnails

11[0106] 下身的分值 11 [0106] score lower body

[0107] 其中Wi为全身人体下肢的宽,n为人数,设W为缩略图的宽。 [0107] where Wi systemic human lower limb wide, n is the number, let W thumbnails wide.

[0108] 所述步骤(23)中输入缩略图,设其灰度直方图为gray_hist = (gh_l, gh_2, • •, gh_n),n为直方图的个数,tl,t2为整数阈值,T1和T2为直方图和的阈值,base_score为基准分数,则灰度直方图分布值SCOre_hiSt_diStrib的计算公式为: Input [0108] the step (23) thumbnails, which histogram is provided gray_hist = (gh_l, gh_2, • •, gh_n), n is the number of histogram, tl, t2 is an integer threshold, Tl and T2 histogram and threshold value, base_score for the benchmark score, the gray value histogram distribution is calculated SCOre_hiSt_diStrib:

[0109] [0109]

[0110] 其中,n = 10,tl = 2,t2 = 8,Tx = 0. 7,Tx = 0. 8,base—score = 1. 0。 [0110] where, n = 10, tl = 2, t2 = 8, Tx = 0. 7, Tx = 0. 8, base-score = 1. 0.

[0111] 所述步骤(24)中输入缩略图,设其灰度直方图为gray—hist = (gh—1,gh—2,.., Input [0111] the step (24) the thumbnail, which is provided for the histogram gray-hist = (gh-1, gh-2, ..,

为直方图的个数, Is the number of histogram,

为基准分数,则计算灰度直方图方差分布值的score—gray—hist—std计算公式为: A reference score is calculated histogram score-gray-hist-std is calculated as the variance of the distribution of values:

[0112] [0112]

[0113] 其中,n = 64,base_score = 1. 0。 [0113] where, n = 64, base_score = 1. 0.

[0114] 所述步骤(25)中分别在R平面、G平面、B平面上直方图方差分布值,然后求均值得到彩色RGB直方图方差分布值score—RGB—hist—std ; [0114] The step (25), respectively, the variance of the histogram distribution value on the R plane, G plane, B plane and then averaging variances obtained RGB color histogram distribution value score-RGB-hist-std;

[0115] [0115]

[0118] [01]

[0119] ;其中,n = 64, base_score = 1. 0。 [0119]; wherein, n = 64, base_score = 1. 0.

[0120] 所述步骤(26)中输入缩略图,设灰度图像的像素值为Xij,i = 1,...,W,j = 1,. . .,H,彩色RGB 图像的像素值(i^.,giJ,、),i = 1,. . .,W,j = 1,. . .,H,其中W,H 分别为缩略图的宽和高,base_scorel和baSe_SCOre2为基准分数,tl为阈值,设则计算彩色RGB图像变成灰度图像的差异值SCOre_rgb2gray_diff的计算公式为: [0120] The step (26) enter the thumbnail image pixel gray value provided Xij, i = 1, ..., W, j = 1 ,..., Pixel values ​​of H, a color RGB image (i ^., giJ ,,), i = 1 ,..., W, j = 1 ,..., H, where W, H are the width and height of the thumbnail, base_scorel baSe_SCOre2 and reference points, tl the threshold value, the calculated set of the RGB image into a color difference value calculated SCOre_rgb2gray_diff gradation image is:

[0121] [0121]

[0122]其中, [0122] wherein,

[0123] 所述步骤(27)中输入缩略图,设灰度图像的像素值为Xij,i = 1,...,W,j = 1,...,H,其中W,H分别为缩略图的宽和高,则计算灰度图像像素点的方差值的SCOre_ gray_std计算公式为: Input [0123] the step (27) of the thumbnail, the pixel value is provided grayscale image Xij, i = 1, ..., W, j = 1, ..., H, where W, H are SCOre_ gray_std formula thumbnail width and height, the calculated variance value of pixels in the gray image is:

[0125] 所述步骤(28)中分别在R平面、G平面、B平面上图像像素点的方差值,然后求均值得到彩色RGB图像像素点的方差值score—RGB—std ; [0125] The step (28), respectively, in the R plane, G plane, the variance values ​​of the image pixels on the plane B, and then averaging the RGB image pixels to obtain a color variance value score-RGB-std;

[0130] 所述步骤(29)中输入缩略图,设彩色RGB图像的像素值gi,bi),i = 1,..., WXH,其中W,H分别为缩略图的宽和高,base—score为基准分数,则计算彩色RGB块之间的离散距离值score—rgb—dist的计算公式为: Input [0130] the step (29) in the thumbnail image pixel RGB color values ​​provided gi, bi), i = 1, ..., WXH, where W, H are the width and height of the thumbnail, base -score as benchmark score, calculated discrete distance values ​​between RGB color blocks score-rgb-dist is calculated as:

[0131] [0131]

[0132] 步骤(3)对每张缩略图分别通过模糊函数求得上述各特征的分值; [0132] Step (3) of each thumbnail and each value obtained by the above-described various features ambiguity function;

[0133] 步骤(4)对每张缩略图的特征分值通过加权求值得到最终的一个分数,最终的分值finalscore由下列公式计算得到:[0134] f inal_score = X score_face_frontal+w2X score_face_prof ile+w3X score_ ful l_body+w4X score_upper_body+w5X score_lower_body+w6X score_hi st_ distrib+w7X score_gray_hist_std+w8X score_RGB_hist_std+w9X score_rgb2gray_ diff+w10 X score_gray_std+wn X score_RGB_std ; [0133] Step (4) wherein each of the thumbnails to the final value is worth a fraction by weighting the final score is calculated by the following formula finalscore obtained: [0134] f inal_score = X score_face_frontal + w2X score_face_prof ile + w3X score_ ful l_body + w4X score_upper_body + w5X score_lower_body + w6X score_hi st_ distrib + w7X score_gray_hist_std + w8X score_RGB_hist_std + w9X score_rgb2gray_ diff + w10 X score_gray_std + wn X score_RGB_std;

[0135] 其中,Wj = 1. 0, w2 = 0. 5, w3 = 0. 5, w4 = 0. 5, w5 = 0. 5, w6 = 0. 5, w7 = 0. 5, w8 =1. 2,ff9 = 1. 2,w10 = 0. 5,wn = 1. 2。 [0135] wherein, Wj = 1. 0, w2 = 0. 5, w3 = 0. 5, w4 = 0. 5, w5 = 0. 5, w6 = 0. 5, w7 = 0. 5, w8 = 1 . 2, ff9 = 1. 2, w10 = 0. 5, wn = 1. 2.

[0136] 步骤(5)根据分数的大小对N张缩略图进行排序; [0136] Step (5) of the N thumbnails are sorted according to the size fraction;

[0137] 步骤(6)选用分数最高的缩略图作为代表该视频的缩略图。 [0137] Step (6) With the highest score as a representative thumbnail of the thumbnail video.

[0138] 综上可知,由于采用了上述技术,本发明具有如下优点:本发明能够通过计算机图像处理方法来自动的选择具有代表性的缩略图作为默认的缩略图。 [0138] To sum up, the use of the techniques described above, the present invention has the following advantages: the present invention is capable of automatically selecting a representative thumbnail default thumbnail image processing method by a computer.

[0139] 以上所述的实施例仅用于说明本发明的技术思想及特点,其目的在于使本领域内的技术人员能够了解本发明的内容并据以实施,不能仅以本实施例来限定本发明的专利范围,即凡依本发明所揭示的精神所作的同等变化或修饰,仍落在本发明的专利范围内。 Example [0139] The above-described technical idea is for illustration only and features of the invention, its object is to enable one skilled in the art to understand the present invention and accordingly embodiment, not only the present embodiment is defined the patentable scope of the present invention, i.e., where equivalent variations or modifications of the disclosed under this invention made by the spirit, still fall within the scope of the present invention.

Claims (13)

  1. 一种视频缩略图智能选取方法,用于从一段视频中选出一张代表该视频的缩略图,其特征在于:包括以下步骤:(1)获取一段视频的N张缩略图,其中,2≤N≤100,且N是常数;(2)对每张缩略图计算其正面人脸的位置和大小、侧面人脸的位置和大小、全身人体的位置和大小、上身的位置和大小和下身的位置和大小、灰度直方图分布值、灰度直方图方差分布值、彩色RGB直方图方差分布值、彩色RGB图像变成灰度图像的差异值、灰度图像像素点的方差值、彩色RGB图像像素点的方差值、彩色RGB块之间的离散距离值等等特征;(3)对每张缩略图分别通过模糊函数求得上述各特征的分值;(4)对每张缩略图的特征分值通过加权求值得到最终的一个分数;(5)根据分数的大小对N张缩略图进行排序;(6)选用分数最高的缩略图作为代表该视频的缩略图。 A video thumbnail intelligent selection method for selecting a thumbnail image representing the video from a video, which is characterized in that: comprising the steps of: (1) obtaining a video thumbnails of N, wherein, 2 ≦ N≤100, and N is a constant; position (2) to calculate the frontal face of the position and size of each thumbnail, the position and size of a side face, body position and size of a human body, the upper body and lower body size and the position and size distribution value histogram, histogram distribution value variance, the variance of the color distribution histogram RGB values, the color difference value of the RGB image into a gray image, a variance value of the pixel gray scale image, color RGB image pixel variance value, the discrete distance between the RGB color value of a block like characteristics; (3) the respective scores are determined by characteristics of the ambiguity function of each thumbnail; (4) reduction of each thumbnail worth by weighting the characteristic value to a final score; (5) N thumbnails are sorted according to the size fraction; (6) With the highest score thumbnail as a representative video thumbnail.
  2. 2.如权利要求1所述的视频缩略图智能选取方法,其特征在于:所述步骤(1)中,采用等间隔抽取方法,根据预先得到的视频时长,将它平均分成N等份,然后每一等份抽取一帧作为缩略图,且N = 8。 The video thumbnail according to claim 1 smart selection method, wherein: said step (1), the extraction method employed at equal intervals, the duration of the video obtained in advance, it is equally divided into N equal parts, and then each aliquot extracted as a thumbnail, and N = 8.
  3. 3.如权利要求2所述的视频缩略图智能选取方法,其特征在于:所述步骤(2)中包括以下步骤:(21)输入缩略图,检测人脸特征,包括正面人脸的位置和大小、侧面人脸的位置和大小,并计算分数值;(22)输入缩略图,检测人体特征,包括全身人体的位置和大小、上身的位置和大小和下身的位置和大小,并计算分数值;(23)输入缩略图,计算灰度直方图分布值;(24)输入缩略图,计算灰度直方图方差分布值;(25)输入缩略图,计算彩色RGB直方图方差分布值;(26)输入缩略图,计算彩色RGB图像变成灰度图像的差异值;(27)输入缩略图,计算灰度图像像素点的方差值;(28)输入缩略图,计算彩色RGB图像像素点的方差值;(29)输入缩略图,计算彩色RGB块之间的离散距离值。 3. The video thumbnails according to claim 2 smart selection method, wherein: said step comprises (2) the following steps: (21) input thumbnail position detecting facial features, including a front face and size, position and size of a side face, and calculating a score value; (22) input thumbnail, detection of human features, including the location and size of the human body, the position and size of the position and size of the lower body and upper body, and calculating a score value ; (23) input thumbnail calculated histogram distribution value; (24) input thumbnail calculates the variance histogram distribution value; (25) input thumbnail RGB color histogram variances calculated distribution value; (26 ) input thumbnail calculates RGB color image into a grayscale image difference value; (27) input thumbnail calculating a variance value of a gray pixel image; (28) input thumbnails calculated color RGB pixels of the image variance; (29) input thumbnail calculated discrete distance values ​​between RGB color blocks.
  4. 4.如权利要求3所述的视频缩略图智能选取方法,其特征在于:所述步骤(21)中则正面人脸的分值 The video thumbnail according to claim 3 intelligent selection method, wherein: said step value (21) of the frontal face 其中Wi为正面人脸的宽,η为人脸个数,设W为缩略图的宽;侧面人脸的分值 Wherein Wi is the width of the front face, the number of [eta] a human face, is provided as thumbnails width W; side face score 其中Wi为侧面人脸的宽,η为人脸个数,设W为缩略图的宽。 Wherein Wi is the width of the side face, the number of [eta] a human face, set width W as a thumbnail.
  5. 5.如权利要求3所述的视频缩略图智能选取方法,其特征在于:所述步骤(22)中设(Xi, Yi, W1,K),! = I,..., Π,其中Xi,Yi代表人脸位置,Wi为人脸的宽,η为人脸个数,设W为缩略图的宽, 则全身的分值 5. The video thumbnails according to claim 3 intelligent selection method, wherein: said step (22) is provided in (Xi, Yi, W1, K) ,! = I, ..., Π, wherein Xi , Yi representative face location, Wi width of a human face, a human face number [eta], set the width W as a thumbnail, the value of the body 其中Wi为全身人体的宽,η为人数,设W为缩略图的宽; 上身的分值 Wi human body which is wide, η is the number, let W thumbnails wide; upper body scores 其中Wi为全身人体上肢的宽,η为人数,设W为缩略图的宽; 下身的分值 Where Wi systemic human upper limb width, η is the number, let W thumbnails wide; lower body scores 其中K为全身人体下肢的宽,η为人数,设W为缩略图的宽。 Where K is the lower limb of the human body width, η is the number, let W thumbnails wide.
  6. 6.如权利要求3所述的视频缩略图智能选取方法,其特征在于:所述步骤(23)中输入缩略图,设其灰度直方图为gray_hist = (gh_l, gh_2, . .,gh_n), η为直方图的个数,tl, t2为整数阈值,Tl和T2为直方图和的阈值,baSe_SCOre为基准分数,则灰度直方图分布值score_hist_distrib 的计算公式为:- 6. The video thumbnails according to claim 3 intelligent selection method, comprising: (.. Gh_l, gh_2,, gh_n) inputted in said step (23) thumbnails, which histogram is provided gray_hist = , [eta] is the number of histogram, tl, t2 is an integer threshold, Tl, and T2 histogram and threshold value, baSe_SCOre benchmark score is, the histogram distribution value score_hist_distrib calculated as follows: -
  7. 7.如权利要求3所述的视频缩略图智能选取方法,其特征在于:所述步骤(24)中输入缩略图,设其灰度直方图为gray_hist = (gh_l, gh_2, ..,gh_n),η为直方图的个数, 7. The video thumbnails according to claim 3 intelligent selection method, comprising: an input thumbnail said step (24), which is provided as a histogram gray_hist = (gh_l, gh_2, .., gh_n) , [eta] is the number of histogram,
  8. 8.如权利要求3所述的视频缩略图智能选取方法,其特征在于:所述步骤(25)中分别在R平面、G平面、B平面上直方图方差分布值,然后求均值得到彩色RGB直方图方差分布值score—RGB—hist—std ; 8. The video thumbnails according to claim 3 intelligent selection method, wherein: said step R plane, G plane, the variance of the distribution of the histogram value (25), respectively, on the B plane, then averaging the obtained RGB color the variance of the histogram distribution value score-RGB-hist-std;
  9. 9.如权利要求3所述的视频缩略图智能选取方法,其特征在于:所述步骤(26)中输入缩略图,设灰度图像的像素值为xu,i = 1,...,W,j = 1,...,H,彩色RGB图像的像素值(^,gij^ij),! = 1,· · ·,W,j = 1,· · ·,H,其中1,!1分别为缩略图的宽和高^386_8(;0仪1和base_score2为基准分数,tl为阈值,设则计算彩色RGB图像变成灰度图像的差异值sCOre_ rgb2gray_diff的计算公式为: 9. The video thumbnails according to claim 3 intelligent selection method, wherein: the input step (26) thumbnail image pixel gray scale value provided xu, i = 1, ..., W , j = 1, ..., H, the pixel values ​​of RGB color image (^, gij ^ ij) ,! = 1, · · ·, W, j = 1, · · ·, H, wherein a,! 1 respectively, the width and height of the thumbnail images 386_8 ^ (; 0 base_score2 instrument 1 and the threshold value is the benchmark score, tl, provided the calculated RGB image into a color difference value calculated sCOre_ rgb2gray_diff gradation image is:
  10. 10.如权利要求3所述的视频缩略图智能选取方法,其特征在于:所述步骤(27)中输入缩略图,设灰度图像的像素值为Xij, i = 1,...,W,j = 1,...,H,其中W,H分别为缩略图的宽和高,则计算灰度图像像素点的方差值的SCOre_gray_Std计算公式为: 10. The video thumbnails according to claim 3 intelligent selection method, comprising: an input thumbnail said step (27), the pixel gray scale image is provided Xij, i = 1, ..., W , j = 1, ..., H, where W, H are the width and height of the thumbnail image, the variance of the gray scale image SCOre_gray_Std pixel calculation formula is calculated as:
  11. 11.如权利要求3所述的视频缩略图智能选取方法,其特征在于:所述步骤(28)中分别在R平面、G平面、B平面上图像像素点的方差值,然后求均值得到彩色RGB图像像素点的方差值score—RGB—std ; 11. The video thumbnails according to claim 3 intelligent selection method, wherein: said step (28), respectively, in the R plane, G plane, the variance values ​​of the image pixels on the plane B, then averaging to give color RGB image pixel variance value score-RGB-std;
  12. 12.如权利要求3所述的视频缩略图智能选取方法,其特征在于:所述步骤(29)中输入缩略图,设彩色RGB图像的像素值O^gpbi), i = 1,...,1父!1,其中1,!1分别为缩略图的宽和高,base_score为基准分数,则计算彩色RGB块之间的离散距离值sCOre_rgb_dist 的计算公式为: 12. The video thumbnails according to claim 3 intelligent selection method, wherein: said step (29) input thumbnail image pixel RGB color values ​​provided O ^ gpbi), i = 1, ... 1 parent 1, wherein the 1, 1, respectively, the width and height of the thumbnail, base_score for the benchmark score, calculated discrete distance between the RGB color value is calculated for the block sCOre_rgb_dist!!:
  13. 13.如权利要求2所述的视频缩略图智能选取方法,其特征在于:所述步骤(4)中最终的分值final_SCOre由下列公式计算得到:f inal_score = W1X score_face_frontal+w2 X score_face_prof i 1 e+w3 X score_ ful l_body+w4 X score_upper_body+w5X score_lower_body+w6 X score_hi st_ distrib+w7X score_gray_hist_std+w8X score_RGB_hist_std+w9X score_rgb2gray_ diff+w10X score—gray—std+wnX score—RGB—std ;其中,W1 = 1. 0,w2 = 0. 5,W3 = 0. 5,W4 = 0. 5,w5 = 0. 5,w6 = 0. 5,w7 = 0. 5,w8 = 1. 2, W9 = 1. 2,W10 = 0. 5,W11 = 1. 2。 13. The video thumbnails according to claim 2 smart selection method, wherein: said step (4) the final score is calculated by the following equation final_SCOre obtained: f inal_score = W1X score_face_frontal + w2 X score_face_prof i 1 e + w3 X score_ ful l_body + w4 X score_upper_body + w5X score_lower_body + w6 X score_hi st_ distrib + w7X score_gray_hist_std + w8X score_RGB_hist_std + w9X score_rgb2gray_ diff + w10X score-gray-std + wnX score-RGB-std; wherein, W1 = 1. 0, w2 = 0. 5, W3 = 0. 5, W4 = 0. 5, w5 = 0. 5, w6 = 0. 5, w7 = 0. 5, w8 = 1. 2, W9 = 1. 2, W10 = 0. 5, W11 = 1. 2.
CN 201010180153 2010-05-20 2010-05-20 Intelligent video thumbnail selection method CN101853286B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201010180153 CN101853286B (en) 2010-05-20 2010-05-20 Intelligent video thumbnail selection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201010180153 CN101853286B (en) 2010-05-20 2010-05-20 Intelligent video thumbnail selection method

Publications (2)

Publication Number Publication Date
CN101853286A true true CN101853286A (en) 2010-10-06
CN101853286B CN101853286B (en) 2016-08-10

Family

ID=42804778

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201010180153 CN101853286B (en) 2010-05-20 2010-05-20 Intelligent video thumbnail selection method

Country Status (1)

Country Link
CN (1) CN101853286B (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104063844A (en) * 2013-03-21 2014-09-24 腾讯科技(深圳)有限公司 Reduced graph generation method and system
CN104506968A (en) * 2014-12-31 2015-04-08 北京奇艺世纪科技有限公司 Method and device for determining video abstract figure
CN104581379A (en) * 2014-12-31 2015-04-29 乐视网信息技术(北京)股份有限公司 Video preview image selecting method and device
CN104598921A (en) * 2014-12-31 2015-05-06 乐视网信息技术(北京)股份有限公司 Video preview selecting method and device
CN105704559A (en) * 2016-01-12 2016-06-22 深圳市茁壮网络股份有限公司 Poster generation method and apparatus thereof
CN106101868A (en) * 2016-07-18 2016-11-09 乐视控股(北京)有限公司 Thumbnail generation method and generation device

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2003333523A (en) * 2002-05-17 2003-11-21 Newsoft Technology Corp Character thumbnail sequence generating system and method thereof
CN1561513A (en) * 2001-10-01 2005-01-05 汤姆森特许公司 System and method for identifying data on a storage medium
US20070294641A1 (en) * 2000-05-31 2007-12-20 Rashkovskiy Oleg B Automatically preparing streaming video programming guides
US20080046406A1 (en) * 2006-08-15 2008-02-21 Microsoft Corporation Audio and video thumbnails
WO2009044351A1 (en) * 2007-10-04 2009-04-09 Koninklijke Philips Electronics N.V. Generation of image data summarizing a sequence of video frames
US20100107126A1 (en) * 2008-10-28 2010-04-29 Hulu Llc Method and apparatus for thumbnail selection and editing
JP4612826B2 (en) * 2004-10-01 2011-01-12 キヤノン株式会社 Storage device, storage method, a computer readable storage medium, and program

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070294641A1 (en) * 2000-05-31 2007-12-20 Rashkovskiy Oleg B Automatically preparing streaming video programming guides
CN1561513A (en) * 2001-10-01 2005-01-05 汤姆森特许公司 System and method for identifying data on a storage medium
JP2003333523A (en) * 2002-05-17 2003-11-21 Newsoft Technology Corp Character thumbnail sequence generating system and method thereof
JP4612826B2 (en) * 2004-10-01 2011-01-12 キヤノン株式会社 Storage device, storage method, a computer readable storage medium, and program
US20080046406A1 (en) * 2006-08-15 2008-02-21 Microsoft Corporation Audio and video thumbnails
WO2009044351A1 (en) * 2007-10-04 2009-04-09 Koninklijke Philips Electronics N.V. Generation of image data summarizing a sequence of video frames
US20100107126A1 (en) * 2008-10-28 2010-04-29 Hulu Llc Method and apparatus for thumbnail selection and editing

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104063844A (en) * 2013-03-21 2014-09-24 腾讯科技(深圳)有限公司 Reduced graph generation method and system
CN104506968A (en) * 2014-12-31 2015-04-08 北京奇艺世纪科技有限公司 Method and device for determining video abstract figure
CN104581379A (en) * 2014-12-31 2015-04-29 乐视网信息技术(北京)股份有限公司 Video preview image selecting method and device
CN104598921A (en) * 2014-12-31 2015-05-06 乐视网信息技术(北京)股份有限公司 Video preview selecting method and device
CN104506968B (en) * 2014-12-31 2018-08-07 北京奇艺世纪科技有限公司 FIG kinds of video summary determination method and apparatus
CN105704559A (en) * 2016-01-12 2016-06-22 深圳市茁壮网络股份有限公司 Poster generation method and apparatus thereof
CN106101868A (en) * 2016-07-18 2016-11-09 乐视控股(北京)有限公司 Thumbnail generation method and generation device

Also Published As

Publication number Publication date Type
CN101853286B (en) 2016-08-10 grant

Similar Documents

Publication Publication Date Title
Cheng et al. A novel fuzzy logic approach to contrast enhancement
Singh et al. Image enhancement using exposure based sub image histogram equalization
Wang et al. Face recognition under varying lighting conditions using self quotient image
Han et al. A comparative study on illumination preprocessing in face recognition
US20060126941A1 (en) Face region estimating device, face region estimating method, and face region estimating program
Celik Two-dimensional histogram equalization and contrast enhancement
US20050254727A1 (en) Method, apparatus and computer program product for determining image quality
Cheng et al. Contrast enhancement based on a novel homogeneity measurement
Fan et al. Homomorphic filtering based illumination normalization method for face recognition
CN101211356A (en) Image inquiry method based on marking area
CN103268475A (en) Skin beautifying method based on face and skin color detection
JP3465226B2 (en) Image density conversion processing method
JP2010093466A (en) Information extraction device, information extracting method, program, and recording medium
Huang et al. Image contrast enhancement for preserving mean brightness without losing image features
Bai et al. Edge preserved image fusion based on multiscale toggle contrast operator
CN101493892A (en) Image characteristic extracting method and device
CN102324019A (en) Method and system for automatically extracting gesture candidate region in video sequence
JP2004062605A (en) Scene identification method and device, and program
JP2006013722A (en) Unit and method for processing image
CN103002225A (en) Multiple exposure high dynamic range image capture
CN101383008A (en) Image classification method based on visual attention model
CN102129693A (en) Image vision significance calculation method based on color histogram and global contrast
CN101840573A (en) Method for estimating pixel-level image fusion quality
Lai et al. Robust facial feature point detection under nonlinear illuminations
US20110182509A1 (en) Image Preprocessing

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C14 Grant of patent or utility model