CN103377473B - An image duplication apparatus and method - Google Patents

An image duplication apparatus and method Download PDF

Info

Publication number
CN103377473B
CN103377473B CN 201210115756 CN201210115756A CN103377473B CN 103377473 B CN103377473 B CN 103377473B CN 201210115756 CN201210115756 CN 201210115756 CN 201210115756 A CN201210115756 A CN 201210115756A CN 103377473 B CN103377473 B CN 103377473B
Authority
CN
Grant status
Grant
Patent type
Application number
CN 201210115756
Other languages
Chinese (zh)
Other versions
CN103377473A (en )
Inventor
张永华
关涛
黄斌强
Original Assignee
深圳市世纪光速信息技术有限公司
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
Grant date

Links

Abstract

本发明公开了一种图像排重的方法和装置,该方法,包括:获取目标图片和至少一个基准图片的特征值;根据获取的所述特征值,判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值;如果是,则确定所述目标图片不与所述基准图片重复,并保留所述目标图片;否则,确定所述目标图片与所述基准图片重复,并丢弃所述目标图片。 The present invention discloses an image heavy discharge method and apparatus, the method comprising: obtaining a target image and the reference value of the at least one feature image; obtained according to the characteristic value, it is determined whether the target picture and the reference pictures of each feature value of the difference is greater than a predetermined threshold value; if yes, determining that the images do not overlap with the target reference picture, and retain the target picture; otherwise, determining the target picture and the reference picture is repeated, and discarded the target picture. 采用本发明提供的方法,可以提高系统资源的利用效率。 The method provided by the invention, can improve the efficiency of system resources.

Description

一种图像排重方法和装置 An image duplication apparatus and method

技术领域 FIELD

[0001] 本发明涉及网络技术领域,特别涉及一种图像排重方法和装置。 [0001] The present invention relates to network technologies, and particularly to a method and apparatus for image duplication.

背景技术 Background technique

[0002] 随着互联网技术的发展,网络搜索已经成为了人们浏览网络、获取信息最常用的手段之一,而网络图片的搜索又是网络搜索中一个非常重要的组成部分。 [0002] With the development of Internet technology, Web search has become the people to browse the web, access to one of the most commonly used means of information, images and web search web search is a very important part.

[0003] 用户终端在进行图片的搜索时,服务器会根据用户的搜索关键词搜索相关的图片文件,并将搜索结果提供给用户终端。 When the [0003] user terminal during the search image, the server will search for relevant picture files based on the user's search keywords and search results to the user terminal. 然而,往往在搜索结果中会有大量的重复性的图片。 However, there is often a lot of repetitive pictures in the search results. 图像排重就是排除图片搜索结果中的重复图片。 The image duplication image search results that exclude duplicate images. 现有技术采用的图像排重方法,一般是先获取图片文件的数据代码,然后进行比较,对于数据代码相同的图片文件,只保留其中的一个图片文件,而将其他图片文件排除。 The image duplication prior art method used, typically by obtaining a data file of the code image, and then compares the code data for the same image file, retaining only one of the image file, and to exclude the other picture.

[0004] 然而,发明人发现,现有技术至少存在如下问题:在进行图片文件搜索时,搜索结果中经常会出现内容非常接近,可是又不完全相同的搜索结果,根据现有技术的方法,排除数据代码相同的搜索结果,并不能将这些内容接近的搜索结果排除掉,这使得搜索结果仍然有大量重复的无效信息,浪费了大量的系统资源。 [0004] However, the inventor found that the prior art has at least the following questions: During the image file search, search results often appear content is very close, but not identical search results, according to the prior art method, exclude data code is the same search results, it does not bring them closer to the content of search results ruled out, which makes the search results are still a large number of repeat invalid information, wasting a lot of system resources.

发明内容 SUMMARY

[0005] 本发明的目的在于提供一种图像排重方法和装置,以提高系统资源的利用效率, 为此,本发明实施例采用如下技术方案: [0005] The object of the present invention is to provide a method and an image duplication apparatus, in order to improve the efficiency of system resources, end, embodiments of the present invention uses the following technical solutions:

[0006] 一种图像排重的方法,包括: [0006] An image-duplication method, comprising:

[0007] 获取目标图片和至少一个基准图片的特征值; [0007] obtaining at least one target picture and the reference picture feature value;

[0008] 根据获取的所述特征值,判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值; [0008] The acquisition of the characteristic value, determining whether the target image and each reference image of the characteristic values ​​are greater than a preset difference threshold;

[0009] 如果是,则确定所述目标图片不与所述基准图片重复,并保留所述目标图片;否贝1J,确定所述目标图片与所述基准图片重复,并丢弃所述目标图片。 [0009] If yes, it is determined that the images do not overlap with the target reference picture, and retain the target picture; No shell 1J, determining the target picture and the reference picture is repeated, and discards the target image.

[0010] 一种图像排重的装置,包括: [0010] An image duplication apparatus, comprising:

[0011] 获取模块,用于获取目标图片和至少一个基准图片的特征值; [0011] obtaining module, configured to acquire the target picture and the reference value of the at least one characteristic of the image;

[0012] 排重模块,用于根据获取的所述特征值,判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值; [0012] duplication module, according to the characteristic value acquired, it is determined whether the target picture with the reference picture feature difference values ​​are greater than a predetermined threshold level;

[0013] 如果是,则确定所述目标图片不与所述基准图片重复,并保留所述目标图片;否贝1J,确定所述目标图片与所述基准图片重复,并丢弃所述目标图片。 [0013] If yes, it is determined that the images do not overlap with the target reference picture, and retain the target picture; No shell 1J, determining the target picture and the reference picture is repeated, and discards the target image.

[0014] 本发明的上述实施例,获取目标图片和至少一个基准图片的特征值,根据获取的所述特征值,判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值;如果是,则确定所述目标图片不与所述基准图片重复,并保留所述目标图片;否则,确定所述目标图片与所述基准图片重复,并丢弃所述目标图片。 [0014] The embodiments of the present invention, the obtaining target picture and reference picture least one characteristic value of the characteristic value according to the acquired target image for determining whether the feature values ​​of the respective differences are larger than the reference image a preset threshold ; if yes, determining that the images do not overlap with the target reference picture, and retain the target picture; otherwise, determining the target picture and the reference picture is repeated, and discards the target image. 从而,可以减小重复图片出现的可能, 提高系统资源的利用效率。 Thus, can reduce the possibility of duplicate images appear to improve the utilization efficiency of system resources.

附图说明 BRIEF DESCRIPTION

[0015] 图1为本发明实施例提供的图像排重方法的流程示意图之一; [0015] FIG image duplication process flow according to an embodiment of the present invention, a schematic view of one;

[0016] 图2为本发明实施例提供的图像排重方法的应用实例的示意图之一; [0016] FIG 2, one embodiment provides an application example of an image duplication process schematic of the present invention;

[0017] 图3为本发明实施例提供的图像排重方法的流程示意图之二; [0017] FIG 3 image duplication process flow according to an embodiment of the present invention, a schematic diagram of the two;

[0018] 图4为本发明实施例提供的图像排重方法的应用实例的示意图之二; [0018] FIG. 4 embodiment provides a schematic bis application example of an image duplication process of the present invention;

[0019] 图5为本发明实施例提供的图像排重装置的结构示意图。 [0019] FIG. 5 is a schematic structure of an image duplication apparatus according to an embodiment of the present invention.

具体实施方式 detailed description

[0020] 下面将结合本发明中的附图,对本发明中的技术方案进行清楚、完整的描述,显然,所描述的实施例是本发明的一部分实施例,而不是全部的实施例。 [0020] Next, the present invention in conjunction with the accompanying drawings, the present invention of clearly and completely described, obviously, the described embodiments are part of the embodiments of the present invention rather than all embodiments. 基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动的前提下所获得的所有其它实施例,都属于本发明保护的范围。 Based on the embodiments of the present invention, all other embodiments of ordinary skill in the art without creative efforts shall fall within the scope of the present invention.

[0021] 如图1所示,为本发明实施例提供的图像排重的方法的流程,具体包括以下步骤: [0021] As shown in FIG 1, the present image-duplication process according to an embodiment of the method of the invention, includes the following steps:

[0022] 步骤101,服务器获取目标图片和至少一个基准图片的特征值。 [0022] Step 101, the server acquires the target image and the reference value of at least one characteristic of the image.

[0023] 当用户在用户终端输入搜索关键词,并点击开始搜索按钮时,终端将相应的搜索关键词发送给服务器,服务器则根据搜索关键词进行相应的图片搜索。 [0023] When the user inputs a search keyword in the user terminal, and click on the search start button, the corresponding terminal transmits a search keyword to the server, the corresponding image search according to the search keyword. 当搜索到第二张图片时,便将第二张图片作为目标图片进行图像排重处理,此时的基准图片是搜索到的第一张图片。 When searching for the second picture, put the second image to image as the target image duplication process, when the reference image is searched first picture. 当搜索到第三张图片时,将第三张图片作为目标图片进行图像排重处理,此时的基准图片可以是第一张图片和第二张图片(如果第二张图片没有被排除)。 When searching for the third image, the third image to image as the target image duplication process, at this time the reference image may be the first picture and the second picture (second picture, if not excluded). 按照此方式,每搜索到一张图片,则将此图片作为目标图片,并将之前经过排重处理保留下来的图片作为基准图片(可以是之前保留的全部图片,也可以根据具体情况选择部分图片),对新搜索到的图片进行图像排重处理。 In this manner, each search picture, this picture as the target image, and after the duplication process before the retained picture as a reference picture (entire image may be previously reserved, some images may be selected depending on the circumstances ), to search for new images for image duplication process.

[0024] 具体的,服务器可以根据图片(包括目标图片和基准图片)中像素点在预设颜色空间中不同分量的强度,确定图片的特征值。 [0024] Specifically, the server may be based on image feature values ​​(including a reference picture and the target picture) is preset in the pixel intensities of the different components of the color space, the image is determined. 颜色空间由多个通道(一般为三个或者四个)组成,每个通道对应一个用于描述像素点显示特征的参数,对于图片中的像素点,通过对每个通道对应的参数进行相应的取值,可以确定像素点的显示特性,各通道对应的参数可称为显示该像素点的分量(或显示分量),相应的取值可称为该分量的强度(一般是0-255之间的一个数值)。 Color space defined by a plurality of channels (usually three or four), each channel corresponds to a parameter used to describe the characteristics of a display pixel, for pixels in the image, parameters corresponding to each channel by the corresponding value, may determine the display characteristic of the pixel, the parameters corresponding to each channel may be referred to as a component of the pixel of the display (or the display component), it can be referred to the corresponding values ​​of the intensity components (typically between 0-255 a value). 每张图片都可以按照不同的颜色空间将其中的像素点颜色划分为多个分量。 Each image can be in a different color space in which the pixels are divided into a plurality of color components. 例如,最常见的RGB (Red Green Blue,红绿蓝)颜色空间,将颜色空间划分为R (红)、G (绿)、B (蓝)三个通道,按照RGB颜色空间定义的图片中的像素点由R、G、B三个分量按照不同的强度组成。 For example, the most common RGB (Red Green Blue, Red Green Blue) color space, the color space is divided into R (red), G (green), B (blue) three channels, according to the image defined in the RGB color space the pixels R, G, B components of the composition according to three different intensities. 又例如,HSV (Hue Saturation Value,色彩、饱和度、明度)颜色空间,将颜色空间划分为H (色彩)、S (饱和度)、V (明度)三个通道,按照HSV颜色空间定义的图片中的像素点由H、S、 V三个分量按照不同的强度组成。 As another example, HSV (Hue Saturation Value, the color, saturation, lightness) color space, the color space is divided by H (hue), S (Saturation), V (brightness) of three channels, HSV color space image in accordance with the defined the pixels in the H, S, V components of the composition according to three different intensities.

[0025] 需要指出的是,预设颜色空间可以根据具体情况任意选择,目标图片和基准图片要选择相同的预设颜色空间,以根据像素点在该预设颜色空间中不同分量的强度,确定图片的特征值。 [0025] It should be noted that the predetermined color space may be arbitrarily selected depending on the circumstances, the target pictures and reference pictures to be selected the same predetermined color space, according to the intensity of pixels in the predetermined color space different components, determined characteristic picture of value.

[0026] 优选的,可以按照如下步骤确定图片的特征值。 [0026] Preferably, the image characteristic value can be determined according to the following steps.

[0027] 步骤一,服务器可以针对预设颜色空间的不同通道,按预设规则在图片上分别划分出至少一个区域。 [0027] Step a, the server may be preset for different channels of color space, according to a preset rule on the at least one image each divided region. 其中,区域的形状可以为方形、圆形或其他形状,区域的数量可以任意, 区域之间可以存在间隙,也可以重叠。 Wherein the shape of the region may be any number of square, circular or other shape, the area, there may be a gap between the regions may overlap.

[0028] 例如,如图2所示,可以在G通道上将图片划分出4X4个小方块,在R通道上将图片划分出3 X 3个小方块,在B通道上将图片划分出2 X 2个小方块。 [0028] For example, as shown in FIG. 2, may be divided on the G channel image illustrating 4X4 small blocks, on the R channel in the 3 X 3 image divided into small blocks, the B channel will be divided into 2 X Image 2 small squares.

[0029] 在进行图像排重的过程中,目标图片与各基准图片使用相同的预设规则划分区域。 [0029] The image reconstruction process is performed in the row, each of the reference picture and the target picture using the same preset rule divided areas. 具体的,可以在目标图片与基准图片中划分数量相同的区域,各对应区域的尺寸、中心位置相同,这样划分,对于尺寸(分辨率)不同的图片,即使画面内容完全相同,也会被识别为不重复的图片。 Specifically, the target image may be a reference picture with the same number of divided regions, corresponding to the size of each area, the same as the center position, this division, for the size (resolution) of different pictures, even if the picture content is identical, it will be identified not to duplicate images. 另外,也可以在对尺寸不同的图片划分区域时,在划分相同数量的区域的同时,根据两张图片尺寸的比例,将对应区域的尺寸进行等比例缩放,这样划分,对于尺寸不同而画面内容很接近的图片,将会被识别为重复的图片。 Further, when the size may be different pictures divided areas, while the same number of divided regions according to the size ratio of two pictures, the size of a corresponding region will be scaled and the like, this division, different size picture content very close to the picture, will be recognized as duplicate images. 例如,目标图片尺寸为20X20个像素点,基准图片尺寸为40X40,目标图片中某区域为中心在坐标(5,5)处,半径为5的圆形区域,基准图片中对应的区域应为中心在坐标(10,10)处,半径为10的圆形区域。 For example, the target image is a 20X20 pixel points, the reference image size is 40X40, a target picture area centered at the coordinates (5,5), the radius of circular region 5, the region corresponding to the reference picture shall be centered the coordinates (10, 10), the radius of the circular region 10.

[0030] 当不同通道分别为不同的颜色通道(对应颜色参数的通道,例如,RGB颜色空间中R、G、B通道分别对应不同的颜色参数,所以可以称作颜色通道)时,对于视觉敏感度越高的颜色对应的颜色通道,在图片上划分的区域数量越多。 [0030] When the different channels are different color channel (color parameter corresponding to the channel, eg, RGB color space R, G, B color channels corresponding to different parameters, can be referred to as color channel), for visually sensitive the higher the degree of color channels corresponding to the color, the more the number of the divided region of the image. 例如,绿色的视觉敏感度高于红色的视觉敏感度,红色的视觉敏感度高于蓝色的视觉敏感度,那么可以在G通道上划分区域的数量最多,在R通道上划分区域的数量次之,B通道再次之。 For example, the visual sensitivity of green than red visual sensitivity, visual sensitivity of red than blue visual sensitivity, then the number may be divided up in the region of the G channel, the channel divided areas on the number of times R the, B of the channel again.

[0031] 步骤二,服务器根据不同通道上各区域中相应分量的强度,确定所述图片的特征值。 [0031] Step two, the server according to the intensity of each channel in different regions of the respective component, determining a characteristic value of the picture. 其中,区域中某分量的强度,可以是区域内各像素点该分量的强度的总和或者平均值。 Wherein the intensity of a component of the region, or may be the average of the sum of the components of each pixel intensity in the region.

[0032] 具体的,图片的特征值确定过程可以如下面流程,下面将结合RGB颜色空间的实例进行详细阐述。 [0032] Specifically, the image characteristic value determination process may be the process as described below, in conjunction with the following examples RGB color space will be described in detail.

[0033] 步骤a,对于每个通道,根据各区域中相应分量的强度,确定强度平均值。 [0033] Step a, for each channel, according to the strength of each corresponding component regions, determining the average intensity.

[0034] 例如,在G通道上,图片中划分出Ngf小方格,对于每个小方格,分别计算其范围内所有像素的绿色分量的强度总和,计算结果分别记为&,62,63,-_,6_;在1?通道上,图片中划分出Nr个方格,对于每个小方格,分别计算其范围内所有像素的红色分量的强度总和,计算结果分别记为Rl,R2,R3,…,RNr ;在R通道上,图片中划分出Nb个方格,对于每个小方格,分别计算其范围内所有像素的蓝色分量的强度总和,计算结果分别记为 [0034] For example, on the G channel, the picture divided Ngf small squares, for each small square, respectively, calculates the sum of intensity of the green component of all the pixels within its scope, the results were recorded as amp &;, 62, 63, -_, 6_;? 1 in the channel, the picture divided into Nr squares, for each small square, respectively, calculates the sum of intensity of the red component of all the pixels within its scope, results are denoted as Rl, R2, R3, ..., RNr; on R channel, Nb picture divided squares, for each small square, respectively, calculates the sum of intensity of the blue components of all the pixels within its scope, the results were recorded as

[0035] 强度平均值即为该通道上,各区域中相应分量强度的平均值,各通道上的强度平均值,即为该通道上所有区域中相应分量强度的总和,除以划分区域的个数,R、G、B通道上的强度平均值可以分别为: [0035] The average strength of the channel is the average value of the corresponding component intensity in each region, the average value of the intensities of the respective channels, on the channel that is the sum of all areas of the respective component intensity, divided by the divided region number, R, G, B channels on the average strength may be respectively:

Figure CN103377473BD00071

[0039] 步骤b,根据各通道上各区域中相应分量的强度,确定强度总平均值。 [0039] Step B, in accordance with each region corresponding to the intensity components of the respective channels, the total intensity average value is determined.

[0040]强度总平均值可以为所有通道上所有区域中相应分量的强度的总和,除以所有通道上划分的区域数的总和,具体可以为: [0040] The total sum of the intensity average intensity of the respective components may be on all channels in all regions, divided by the total number of regions divided on all channels, may specifically be:

Figure CN103377473BD00081

[0042] 以下c、d步骤,是将上面获得的强度平均值进行二值化处理的过程。 [0042] Here c, d step, the average intensity obtained above was subjected to binarization processing procedure.

[0043] 步骤c,对于每个通道,将各区域中相应分量的强度与强度平均值比较,如果小于强度平均值,则确定该区域的特征值为〇,否则,确定该区域的特征值为1。 [0043] Step C, for each channel, comparing the intensity of each component in the region corresponding to the average intensity, the intensity is less than the average value, it is determined that the feature region is square, otherwise, determining a characteristic value of the area 1. 各区域的特征值可以按照如下函数获得: Characteristic value of each region can be obtained according to the following function:

Figure CN103377473BD00082

[0047]步骤d,将各通道的强度平均值与强度总平均值比较,如果小于强度总平均值,则确定该通道的特征值为〇,否则,确定该通道的特征值为1。 [0047] Step d, the average strength of the overall average intensity of each channel, and if the total is less than the average intensity, it is determined that the channel characteristic value of square, otherwise, determining a characteristic value of the channel 1. 各通道的特征值可以按照如下函数获得: Feature value for each channel can be obtained according to the following function:

Figure CN103377473BD00083

[0051]步骤e,在各通道上各区域的特征值和各通道的特征值中,获取预设数目的(假设为P个)特征值,并将所述预设数目的特征值按照预设顺序顺次连接得到所述图片的特征值。 [0051] Step E, wherein each channel on the eigenvalues ​​of each region and each channel, the acquisition (assumed to be of P) of a predetermined number of feature values ​​and feature values ​​of the preset number of preset and sequentially connecting the obtained image characteristic value. 例如,获取特征值fhft,…,fP,然后根据下面公式将其顺次合并成一个P位的二进制数F0 For example, the feature value acquiring fhft, ..., fP, and then sequentially combined according to the following formulas into which a P-bit binary number F0

Figure CN103377473BD00084

[0053] 上述预设数目可以根据具体情况任意设置,不过对于目标图片和基准图片要设置相同的预设数目,即目标图片和基准图片的特征值位数相同。 [0053] The predetermined number may be arbitrarily set depending on the circumstances, but for the target picture and a reference picture to be set the same preset number, i.e. the features of the target image and the reference image the same value bits. 上述预设顺序可以根据具体情况任意设置,不过对于目标图片和基准图片要设置相同的预设顺序。 The preset order may be arbitrarily set depending on the circumstances, but for the target picture and a reference picture to be set the same preset order.

[0054] 上述强度平均值和强度总平均值的计算,仅是以算数平均值为例,还可以采用几何平均值、中位数、众数等计算方式获取。 [0054] The calculation of the total intensity above the average and the average value, arithmetic average is only an example, may also be employed geometric mean, median, and other public acquisition calculated.

[0055] 步骤102,服务器根据获取的特征值,判断目标图片是否与各基准图片的特征值差异度都大于预设阈值,如果是,则执行步骤103,否则,执行步骤104。 [0055] Step 102, the feature value obtaining server, it is determined whether the target image with the reference image of the characteristic values ​​are greater than a preset difference threshold, if yes, step 103 is performed, otherwise, step 104 is performed. 其中,目标图片与基准图片的特征值差异度,可以是目标图片的特征值与基准图片的特征值的海明距离。 Wherein the image feature of the target image and the reference value of the difference, the feature value may be a Hamming distance of the target image characteristic value and the reference picture.

[0056] 优选的,为了提高处理效率,可以通过建立哈希桶的方式来进行上述判断过程,如图3所示,具体处理流程如下: [0056] Preferably, in order to improve processing efficiency, the determination process may be performed by way of establishing a hash bucket, shown in Figure 3, the specific process is as follows:

[0057] 步骤1021,将图片(包括目标图片和各基准图片)的特征值划分为预设组数目的分组。 [0057] Step 1021, the image (including the respective target picture and a reference picture) is divided into a number of characteristic values ​​of a preset set of packets. 其中,预设组数目大于预设阈值。 Wherein the number of groups is greater than the preset threshold value preset. 各分组的长度可以根据具体情况任意选择;分组的数目也可以根据具体情况任意选择,只要大于预设阈值,且不超过图片特征值的位数。 The length of each packet may be arbitrarily selected according to the circumstances; number of packets may be arbitrarily selected depending on the case, greater than a predetermined threshold value, does not exceed the number of bits of image feature values. 目标图片和各基准图片要按照相同的方式分组。 Each target picture and reference pictures to be grouped in the same way.

[0058] 优选的,为了提高处理效率,预设组数目可以为预设阈值加1,位数相同的分组的数目至少为预设组数目减1。 [0058] Preferably, in order to improve the processing efficiency, a predetermined number of groups can be added to a predetermined threshold, the number of bits of the same packet group is at least a preset number minus 1. 例如,设特征值F的位数为P,预设阈值为q,将特征值F划分为q+ 1个分组,且前q个分组的位数为w,w可以由p/(q+l)取整加1得到,最后一个分组的位数为P-wq,一种特殊情况下,P-wq可能等于w,这时所有分组的位数都为w。 For example, median feature value F set as P, q preset threshold value, the feature value F is divided into q + 1 packets, and the number of bits of packets before the q w, w may be formed of p / (q + l) adding a rounding obtained, the number of bits for the last packet P-wq, a special case, P-wq may be equal to w, the median time of all packets are w.

[0059] 步骤1022,如图4所示,对于每个分组,将分组中的特征值片段作为key,并将除去该特征值片段后的特征值作为value,建立哈希桶,并将每个分组对应的哈希桶组成哈希桶集合。 [0059] Step 1022, shown in Figure 4, for each packet, the packet fragments characteristic values ​​as key, and the feature value is removed after the feature value as the value segment, establishing hash bucket, and each packets corresponding hash bucket hash bucket set composition. q+Ι个分组的哈希桶集合可以记作H= {Hi,H2,H3,'",Hq+i}。 Hash bucket set of q + Ι packets may be denoted as H = {Hi, H2, H3, ' ", Hq + i}.

[0060] 步骤1023,将目标图片的各分组中的key与各基准图片的对应分组中的key进行比较,并判断是否存在相同的key,如果存在,则执行步骤1024,否则执行步骤1026。 [0060] Step 1023, a packet for each packet corresponding to the target image in the reference image for each key in the key are compared, and determines whether there is the same key, if present, step 1024 is executed, otherwise step 1026.

[0061 ] 步骤1024,判断相同的key对应的va Iue的海明距离是否小于或等于预设阈值,如果是,则执行步骤1025,如果所有相同的key对应的value的海明距离都大于预设阈值,则执行步骤1026。 [0061] Step 1024, whether the Hamming distance corresponding to the same key determination va Iue is less than or equal to a preset threshold, and if yes, execute step 1025, if all the same key corresponding Hamming distances are greater than a predetermined value threshold, step 1026.

[0062] 步骤1025,确定目标图片与相应的基准图片的特征值差异度不大于预设阈值。 [0062] Step 1025, the determined target picture and a reference picture corresponding to the feature value difference is not greater than a predetermined threshold value.

[0063] 步骤1026,确定目标图片与各基准图片的特征值差异度都大于预设阈值。 [0063] Step 1026, the target picture is determined with the difference value of the reference image characteristics are greater than a preset threshold value.

[0064] 例如,可以先将目标图片特征值第一分组中的key与各基准图片特征值第一分组中的key比较,如果没有相同的key,则将第二分组中的key进行比较,以此类推,如果找到相同的key,则确定两个相同key对应的V a I ue的海明距离,如果小于或等于阈值q,则说明目标图片与该key对应的基准图片的海明距离小于阈值q,如果大于阈值q,则继续查找相同的key,如果查找到的所有对相同的key对应的V a I ue的海明距离都大于q,或没有找到相同的key,则说明目标图片与各基准图片的海明距离者大于阈值q。 [0064] For example, the target image may first compare the feature value of the first packet of each key image characteristic value of the first reference packet key, if the key is not the same, then the second key packet is compared to such pushing the same key, if found, it is determined that the same key two Hamming distance corresponding V a I ue is, is less than or equal to the threshold value q, then the Hamming distance of the target image and the reference image corresponding to the key is less than the threshold value q, if it exceeds the threshold value q, then continue to look the same key, if found all the same key corresponding to V a I ue of Hamming distances greater than q, or is not found the same key, then the target picture with Hamming distance reference picture signal exceeds the threshold value q.

[0065] 步骤103,服务器确定目标图片不与基准图片重复,并保留此目标图片。 [0065] Step 103, the server determines the target image and the reference image will not be repeated, and retain this target image. 目标图片与各基准图片的特征值差异度足够高说明目标图片没有与基准图片重复,所以可以将目标图片保留到搜索结果中,向终端发送。 Feature value of each difference of the reference image and the target image is sufficiently high and no duplicate description reference image for the target image, the target image may be retained to the search results to the terminal. 而且,服务器可以将此目标图片增加到基准图片中, 用于下一张图片的排重判断中。 Also, the server can be added to this target picture in the reference picture, for re-ranked to determine the next image in. 服务器可以将经过排重处理的所述图片作为基准图片,也可以从中选择部分图片作为基准图片 The server may be through the image duplication process as a reference picture, which may be part of the picture as a reference picture selected

[0066] 步骤104,服务器确定所述目标图片与所述基准图片重复,并丢弃所述目标图片, 进行下一个目标图片的排重处理。 [0066] Step 104, the server determines the target picture and the reference picture is repeated, and discards the target picture, for re-processing of the next row of the target image. 目标图片与基准图片的特征值差异度不够高,说明与该基准图片十分相似,目标图片被认为是重复图片,可以将其丢弃,不放入搜索结果中。 Target picture and the reference picture of the feature value difference is not high enough, indicating that the reference picture is very similar to the target picture is considered to be duplicate images, may be discarded, not placed in the search results.

[0067] 本发明的实施例中,服务器获取目标图片和至少一个基准图片的特征值,根据获取的所述特征值,并判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值;如果是,则确定所述目标图片不与所述基准图片重复,并保留所述目标图片;否则,确定所述目标图片与所述基准图片重复,并丢弃所述目标图片。 Example [0067] the present invention, the server obtains the target picture and reference picture least one characteristic value of the characteristic value obtained in accordance with, and determines whether or not the target picture and the reference picture for each characteristic value of the difference is greater than a pre- thresholding; if yes, determining that the images do not overlap with the target reference picture, and retain the target picture; otherwise, determining the target picture and the reference picture is repeated, and discards the target image. 从而,可以减小重复图片出现的可能,提高系统资源的利用效率。 Thus, can reduce the possibility of duplicate images appear to improve the utilization efficiency of system resources.

[0068] 基于相同的技术构思,本发明实施例还提供了一种图像排重装置,如图5所示,包括: [0068] Based on the same technical concept, embodiments of the present invention further provides an image duplication apparatus shown in Figure 5, comprising:

[0069] 获取模块510,用于获取目标图片和至少一个基准图片的特征值; [0069] The obtaining module 510, configured to acquire the target picture and the reference picture of the at least one feature value;

[0070] 排重模块520,用于根据获取的所述特征值,判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值; [0070] duplication module 520, according to the characteristic value acquired, it is determined whether the target picture with the reference picture feature difference values ​​are greater than a predetermined threshold level;

[0071] 如果是,则确定所述目标图片不与所述基准图片重复,并保留所述目标图片;否贝1J,确定所述目标图片与所述基准图片重复,并丢弃所述目标图片。 [0071] If yes, it is determined that the images do not overlap with the target reference picture, and retain the target picture; No shell 1J, determining the target picture and the reference picture is repeated, and discards the target image.

[0072] 优选的,所述获取模块510,具体用于: [0072] Preferably, the obtaining module 510, configured to:

[0073] 根据图片中像素点在预设颜色空间中不同分量的强度,确定图片的特征值;其中, 所述图片包括所述目标图片和各基准图片。 [0073] The intensities of the different components in a predetermined color space, determining the characteristic value based on the picture image pixel points; wherein, the picture of the target picture and each comprising a reference image.

[0074] 优选的,所述获取模块510,具体用于: [0074] Preferably, the obtaining module 510, configured to:

[0075] 针对所述预设颜色空间的不同通道,按预设规则在图片上分别划分出至少一个区域; [0075] for the preset color space different channels, according to a preset rule on the at least one image each divided region;

[0076] 根据不同通道上各区域中相应分量的强度,确定所述图片的特征值。 [0076] The intensity of each area on different channels of the respective components, determining a characteristic value of the picture.

[0077] 优选的,所述通道具体为颜色通道; [0077] Preferably, said channel in particular color channel;

[0078] 所述获取模块510,具体用于对于视觉敏感度越高的颜色对应的颜色通道,在图片上划分的区域数量越多。 [0078] The obtaining module 510, the higher the sensitivity for particular visual color channels corresponding to the color, the more the number of the divided region of the image.

[0079] 优选的,所述获取模块510,具体用于: [0079] Preferably, the obtaining module 510, configured to:

[0080] 对于每个通道,根据各区域中相应分量的强度,确定强度平均值;并根据各通道上各区域中相应分量的强度,确定强度总平均值; [0080] For each channel, according to the strength of each corresponding component regions, determining the average intensity; and the respective components in accordance with the intensity in each region on each channel, determining a total average intensity;

[0081] 对于每个通道,将各区域中相应分量的强度与强度平均值比较,如果小于强度平均值,则确定该区域的特征值为〇,否则,确定该区域的特征值为1; [0081] For each channel, the intensity and the intensity of each component of the average area of ​​the respective comparison, is less than the average intensity, it is determined that the feature region is square, otherwise, determining a characteristic of the region is 1;

[0082] 将各通道的强度平均值与所述强度总平均值比较,如果小于强度总平均值,则确定该通道的特征值为〇,否则,确定该通道的特征值为1; [0082] The average intensity of each channel is compared with the overall average of the intensity, the intensity is less than the overall average, it is determined that the channel characteristic value of square, otherwise, determining a characteristic of the channel is 1;

[0083] 在各通道上各区域的特征值和各通道的特征值中,获取预设数目的特征值,并将所述预设数目的特征值按照预设顺序顺次连接得到所述图片的特征值,该特征值为二进制数。 [0083] on each channel characteristic value and the characteristic value for each channel in each region, a predetermined number of feature values ​​acquired, and the predetermined number of feature values ​​obtained are sequentially connected in preset order of the picture feature value, the feature value is a binary number.

[0084] 优选的,所述目标图片与所述基准图片的特征值差异度,具体为所述目标图片的特征值与所述基准图片的特征值的海明距离。 Hamming distance values ​​eigenvalue of the reference picture [0084] Preferably, the characteristic values ​​of the difference of the target image and the reference image, the particular target image.

[0085] 优选的,所述排重模块520,具体用于: [0085] Preferably, the duplication module 520, configured to:

[0086] 将图片的特征值划分为预设组数目的分组,其中,所述预设组数目大于所述预设阈值,所述图片包括所述目标图片和各基准图片; [0086] The image characteristic values ​​are divided into a predetermined number of groups of packets, wherein the predetermined number of groups greater than the preset threshold value, the picture image including the target image and each of reference;

[0087] 对于每个分组,将分组中的特征值片段作为key,并将除去该特征值片段后的特征值作为value,建立哈希桶,并将每个分组对应的哈希桶组成哈希桶集合; [0087] For each packet, the packet fragments characteristic values ​​as key, and the feature value is removed after the feature value as the value segment, establishing hash bucket, and each packet corresponding to the hash hash bucket composition bucket collection;

[0088] 将目标图片的各分组中的key与各基准图片的对应分组中的key进行比较;如果存在相同的key,则判断所述相同的key对应的value的海明距离是否小于或等于所述预设阈值;如果是,则确定所述目标图片与相应的基准图片的特征值差异度不大于所述预设阈值; 如果不存在相同的key,或者,所有相同的key对应的value的海明距离都大于所述预设阈值,则确定所述目标图片与各基准图片的特征值差异度都大于所述预设阈值。 [0088] The packets each packet corresponding to the target image in the reference image for each key in the key is compared; if the same key is present, it is determined whether the Hamming distance value corresponding to the same key is less than or equal said predetermined threshold; if yes, determining the target image and the reference image corresponding to the difference of the characteristic value is not greater than the preset threshold value; if the same key does not exist, or, all corresponding to the same key value sea Hamming distance is greater than the preset threshold, it is determined that the target picture and the reference picture for each feature value of the difference is greater than the preset threshold.

[0089] 优选的, [0089] Preferably,

[0090] 所述预设组数目具体为所述预设阈值加1; [0090] The preset number of the particular set plus a preset threshold value;

[0091] 位数相同的分组的数目至少为所述预设组数目减1。 [0091] The number of packets of the same number of bits is at least the predetermined number of groups minus 1.

[0092] 本发明的实施例中,服务器获取目标图片和至少一个基准图片的特征值,根据获取的所述特征值,并判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值;如果是,则确定所述目标图片不与所述基准图片重复,并保留所述目标图片;否则,确定所述目标图片与所述基准图片重复,并丢弃所述目标图片。 Example [0092] the present invention, the server obtains the target picture and reference picture least one characteristic value of the characteristic value obtained in accordance with, and determines whether or not the target picture and the reference picture for each characteristic value of the difference is greater than a pre- thresholding; if yes, determining that the images do not overlap with the target reference picture, and retain the target picture; otherwise, determining the target picture and the reference picture is repeated, and discards the target image. 从而,可以减小重复图片出现的可能,提高系统资源的利用效率。 Thus, can reduce the possibility of duplicate images appear to improve the utilization efficiency of system resources.

[0093] 本领域技术人员可以理解实施例中的装置中的模块可以按照实施例描述进行分布于实施例的装置中,也可以进行相应变化位于不同于本实施例的一个或多个装置中。 [0093] Those skilled in the art will be appreciated apparatus embodiment that the modules can be distributed in accordance with an embodiment of the apparatus of the embodiment may be performed according to the present embodiment which are different from one case or more devices. 上述实施例的模块可以合并为一个模块,也可以进一步拆分成多个子模块。 Modules of the embodiments may be combined into one module, or split into multiple submodules.

[0094] 上述本发明实施例序号仅仅为了描述,不代表实施例的优劣。 Embodiment [0094] The present invention No. merely for description, the embodiments do not represent the merits embodiment.

[0095] 通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到本发明可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。 [0095] By the above described embodiments, those skilled in the art may clearly understand that the present invention may be implemented by software plus a necessary universal hardware platform, also be implemented by hardware, but the former is in many cases more good embodiments. 基于这样的理解,本发明的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质中,包括若干指令用以使得一台终端设备(可以是手机,个人计算机,服务器,或者网络设备等)执行本发明各个实施例所述的方法。 Based on such understanding, the technical solutions of the present invention in essence or the part contributing to the prior art may be embodied in a software product, which computer software product is stored in a storage medium and includes several instructions to enable a terminal devices (which may be a mobile phone, a personal computer, a server, or network device) to execute the methods according to embodiments of the present invention.

[0096] 以上所述仅是本发明的优选实施方式,应当指出,对于本技术领域的普通技术人员来说,在不脱离本发明原理的前提下,还可以做出若干改进和润饰,这些改进和润饰也应视本发明的保护范围。 [0096] The above are only preferred embodiments of the present invention, it should be noted that those of ordinary skill in the art, in the present invention without departing from the principles of the premise, can make various improvements and modifications, such modifications and modifications should also depend on the scope of the present invention.

Claims (14)

  1. 1. 一种图像排重的方法,其特征在于,包括: 获取目标图片和至少一个基准图片的特征值;其中,针对颜色空间的不同分量,将图片划分出至少一个区域,所述特征值根据图片中像素点在不同分量上各区域中相应分量的强度确定得到,包括: 对于每个通道,根据各区域中相应分量的强度,确定强度平均值;并根据各通道上各区域中相应分量的强度,确定强度总平均值; 对于每个通道,将各区域中相应分量的强度与强度平均值比较,如果小于强度平均值, 则确定该区域的特征值为O,否则,确定该区域的特征值为1; 将各通道的强度平均值与所述强度总平均值比较,如果小于强度总平均值,则确定该通道的特征值为O,否则,确定该通道的特征值为1; 在各通道上各区域的特征值和各通道的特征值中,获取预设数目的特征值,并将所述预设数目的特征值按 An image heavy discharge method comprising: obtaining a target image and reference images at least one characteristic value; wherein, for the different components of the color space, the image divided into at least one region, according to the characteristic value the picture corresponding to the intensity components of the pixels in each region is determined to obtain the different components, comprising: for each channel, according to the strength of each corresponding component regions, determining the average intensity; and the corresponding components of the respective channels in accordance with respective regions intensity, the average total intensity is determined; for each channel, the intensity and the intensity of each component of the average area of ​​the respective comparison, is less than the average intensity, it is determined that the feature region is O, otherwise, determining a characteristic of the region is 1; comparing the total intensity of the average intensity of each channel with the average value, the average value is less than the total intensity, it is determined that the channel characteristic value of O, otherwise, determining a characteristic of the channel is 1; in the eigenvalues ​​value of each region and each channel on the channel, the acquired predetermined number of feature values ​​and the feature values ​​by the predetermined number of 预设顺序顺次连接得到所述图片的特征值,该特征值为二进制数; 根据获取的所述特征值,判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值; 如果是,则确定所述目标图片不与所述基准图片重复,并保留所述目标图片;否则,确定所述目标图片与所述基准图片重复,并丢弃所述目标图片。 Connecting predetermined sequential order to obtain the image feature value, the feature value is a binary number; obtained according to the characteristic value, determines whether or not the target picture and the reference picture for each characteristic value of the difference is greater than a predetermined threshold value; If so, it is determined that the images do not overlap with the target reference picture, and retain the target picture; otherwise, determining the target picture and the reference picture is repeated, and discards the target image.
  2. 2. 如权利要求1所述的方法,其特征在于,获取图片的特征值的方法,具体为: 根据图片中像素点在预设颜色空间中不同分量的强度,确定图片的特征值;其中,所述图片包括所述目标图片和各基准图片。 2. The method according to claim 1, wherein the method of obtaining the feature value of the images, specifically as follows: The pixel points in the preset image intensities of the different components of the color space, the image characteristic value determined; wherein, the picture includes a picture of the target and each reference pictures.
  3. 3. 如权利要求2所述的方法,其特征在于, 针对所述预设颜色空间的不同通道,按预设规则在图片上分别划分出至少一个区域; 根据不同通道上各区域中相应分量的强度,确定所述图片的特征值。 3. The method according to claim 2, wherein the predetermined different channels for color space, according to a preset rule on the at least one image each divided region; respective components according to the different channels in each region strength, determining a characteristic value of the picture.
  4. 4. 如权利要求3所述的方法,其特征在于, 所述通道具体为颜色通道; 对于视觉敏感度越高的颜色对应的颜色通道,在图片上划分的区域数量越多。 4. The method according to claim 3, wherein the color channel is a channel specifically; the higher the sensitivity of the visual color corresponding to the color channel, the more the number of the divided region of the image.
  5. 5. 如权利要求1所述的方法,其特征在于,所述目标图片与所述基准图片的特征值差异度,具体为所述目标图片的特征值与所述基准图片的特征值的海明距离。 Hamming 5. The method of claim 1 wherein the value of the reference image characteristic value, characterized in that, wherein the target image and the reference image of the difference value, in particular the target picture distance.
  6. 6. 如权利要求5所述的方法,其特征在于,所述根据获取的所述特征值,判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值,具体为: 将图片的特征值划分为预设组数目的分组,其中,所述预设组数目大于所述预设阈值, 所述图片包括所述目标图片和各基准图片; 对于每个分组,将分组中的特征值片段作为key,并将除去该特征值片段后的特征值作为value,建立哈希桶,并将每个分组对应的哈希桶组成哈希桶集合; 将目标图片的各分组中的key与各基准图片的对应分组中的key进行比较;如果存在相同的key,则判断所述相同的key对应的value的海明距离是否小于或等于所述预设阈值;如果是,则确定所述目标图片与相应的基准图片的特征值差异度不大于所述预设阈值;如果不存在相同的key,或者,所有相同的key对应的value的海明距离都大于所 6. The method according to claim 5, characterized in that the characteristic value according to the acquired target image for determining whether the feature value of the difference of each picture are greater than a preset reference threshold, specifically to: image characteristic values ​​are divided into a predetermined number of groups of packets, wherein the predetermined number of groups greater than the preset threshold value, the picture image including the target image and each of reference; for each packet, the packet fragment as a key feature value and feature value is removed after the feature value as the value segment, establishing hash bucket, and the composition of each packet corresponding to a set of hash bucket hash bucket; each of the key images in the target packet is compared with each reference packet corresponding to the key picture; if present, the same key, it is determined whether the Hamming distance corresponding to the same key value is equal to or less than the preset threshold; if yes, determining the feature value corresponding to the difference of the target image and the reference image is not greater than the predetermined threshold; if not the same key, or the key corresponding to all of the same Hamming distance is greater than the value of 预设阈值,则确定所述目标图片与各基准图片的特征值差异度都大于所述预设阈值。 Predetermined threshold value, it is determined that the target image with the reference image feature value of the difference is greater than the preset threshold.
  7. 7. 如权利要求6所述的方法,其特征在于, 所述预设组数目具体为所述预设阈值加I; 位数相同的分组的数目至少为所述预设组数目减1。 7. The method according to claim 6, wherein said predetermined number is specifically set the preset threshold plus I; the same number of bits as the number of packets of at least a predetermined number of groups minus 1.
  8. 8. —种图像排重的装置,其特征在于,包括: 获取模块,用于获取目标图片和至少一个基准图片的特征值;其中,针对颜色空间的不同分量,将图片划分出至少一个区域,所述特征值根据图片中像素点在不同分量上各区域中相应分量的强度确定得到; 所述获取模块具体用于:对于每个通道,根据各区域中相应分量的强度,确定强度平均值;并根据各通道上各区域中相应分量的强度,确定强度总平均值;对于每个通道,将各区域中相应分量的强度与强度平均值比较,如果小于强度平均值,则确定该区域的特征值为〇,否则,确定该区域的特征值为1;将各通道的强度平均值与所述强度总平均值比较,如果小于强度总平均值,则确定该通道的特征值为0,否则,确定该通道的特征值为1;在各通道上各区域的特征值和各通道的特征值中,获取预设数目的特征值,并 8. - kinds of image duplication apparatus, characterized by comprising: an obtaining module, configured to acquire the target picture and the at least one characteristic value of the reference picture; wherein, for the different components of the color space, the at least one image divided region, the characteristic value according to the intensity of the corresponding pixel in the image component obtained in determining the different components in each region; the obtaining module is configured to: for each channel, according to the strength of each corresponding component regions, determining the average intensity; the intensity of each region and each channel corresponding components, determine a total average strength; for each channel, comparing the intensity of each component in the region corresponding to the average intensity, the intensity is less than the average value, characterized in that the area is determined square value, otherwise, determining a characteristic of the region is 1; the average intensity of each channel is compared with the overall average intensity is less than the average total intensity, it is determined that the channel characteristic value 0 otherwise, determining eigenvalues ​​of the channel 1; on each channel characteristic value and the characteristic value for each channel in each region, a predetermined number of feature values ​​acquired, and 所述预设数目的特征值按照预设顺序顺次连接得到所述图片的特征值,该特征值为二进制数; 排重模块,用于根据获取的所述特征值,判断所述目标图片是否与各基准图片的特征值差异度都大于预设阈值; 如果是,则确定所述目标图片不与所述基准图片重复,并保留所述目标图片;否则,确定所述目标图片与所述基准图片重复,并丢弃所述目标图片。 The predetermined number of feature values ​​in preset order of sequentially connecting the images obtained characteristic value, the characteristic value is a binary number; duplication module, according to the acquired feature value, determining whether the target picture and each of the reference image characteristic values ​​are greater than a preset difference degree threshold; if yes, determining that the images do not overlap with the target reference picture, and retain the target picture; otherwise, determining the target picture and the reference duplicate images and discard the target picture.
  9. 9. 如权利要求8所述的装置,其特征在于,所述获取模块,具体用于: 根据图片中像素点在预设颜色空间中不同分量的强度,确定图片的特征值;其中,所述图片包括所述目标图片和各基准图片。 9. The apparatus according to claim 8, wherein the obtaining module is configured to: according to a preset image intensity of pixel points in the color space of the different components, the determined image characteristic value; wherein, said images include pictures of the target and each reference pictures.
  10. 10. 如权利要求9所述的装置,其特征在于,所述获取模块,具体用于: 针对所述预设颜色空间的不同通道,按预设规则在图片上分别划分出至少一个区域; 根据不同通道上各区域中相应分量的强度,确定所述图片的特征值。 10. The apparatus according to claim 9, wherein the obtaining module is configured to: for the preset color space different channels, according to a preset rule on the at least one image each divided region; The corresponding to the intensity of the components of the different channels in each region, determining a characteristic value of the picture.
  11. 11. 如权利要求10所述的装置,其特征在于,所述通道具体为颜色通道; 所述获取模块,具体用于对于视觉敏感度越高的颜色对应的颜色通道,在图片上划分的区域数量越多。 11. The apparatus according to claim 10, wherein the color channel is a channel specifically; the obtaining module is used for a higher sensitivity of the visual color corresponding to the color channels in the image divided regions the greater the number.
  12. 12. 如权利要求8所述的装置,其特征在于,所述目标图片与所述基准图片的特征值差异度,具体为所述目标图片的特征值与所述基准图片的特征值的海明距离。 Hamming eigenvalue 12. The apparatus according to claim 8, wherein the target image characteristic and the reference value of the degree of difference images, particularly images of the target characteristic value and the reference picture distance.
  13. 13. 如权利要求12所述的装置,其特征在于,所述排重模块,具体用于: 将图片的特征值划分为预设组数目的分组,其中,所述预设组数目大于所述预设阈值, 所述图片包括所述目标图片和各基准图片; 对于每个分组,将分组中的特征值片段作为key,并将除去该特征值片段后的特征值作为value,建立哈希桶,并将每个分组对应的哈希桶组成哈希桶集合; 将目标图片的各分组中的key与各基准图片的对应分组中的key进行比较;如果存在相同的key,则判断所述相同的key对应的value的海明距离是否小于或等于所述预设阈值;如果是,则确定所述目标图片与相应的基准图片的特征值差异度不大于所述预设阈值;如果不存在相同的key,或者,所有相同的key对应的value的海明距离都大于所述预设阈值,则确定所述目标图片与各基准图片的特征值差异度都大于所述预设阈 13. The apparatus of claim 12, wherein said duplication module is configured to: divide the image feature values ​​of the number of preset set of packets, wherein said predetermined number is greater than the set predetermined threshold value, the picture image including the target image and each of reference; for each packet, the packet fragments characteristic values ​​as key, and after the removal of the feature value as a feature value fragment value, establishing hash bucket and each packet corresponding to a set of hash bucket hash bucket composition; packets each packet corresponding to the target image in the reference image for each key in the key is compared; if present, the same key, it is determined that the same whether the Hamming distance corresponding to the key value is equal to or less than the preset threshold; if yes, determining a characteristic of the target image and the reference image corresponding to the difference value is not greater than the preset threshold; the same if there is no the key, or all the same key value corresponding Hamming distances are greater than the preset threshold value, it is determined that the target picture and the reference pictures of each of the difference values ​​are greater than the preset threshold characterized in 值。 value.
  14. 14. 如权利要求13所述的装置,其特征在于, 所述预设组数目具体为所述预设阈值加I; 位数相同的分组的数目至少为所述预设组数目减1。 14. The apparatus according to claim 13, wherein said predetermined number is specifically set the preset threshold plus I; the same number of bits as the number of packets of at least a predetermined number of groups minus 1.
CN 201210115756 2012-04-19 2012-04-19 An image duplication apparatus and method CN103377473B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201210115756 CN103377473B (en) 2012-04-19 2012-04-19 An image duplication apparatus and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201210115756 CN103377473B (en) 2012-04-19 2012-04-19 An image duplication apparatus and method

Publications (2)

Publication Number Publication Date
CN103377473A true CN103377473A (en) 2013-10-30
CN103377473B true CN103377473B (en) 2017-10-24

Family

ID=49462541

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201210115756 CN103377473B (en) 2012-04-19 2012-04-19 An image duplication apparatus and method

Country Status (1)

Country Link
CN (1) CN103377473B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1385329A2 (en) * 2002-07-27 2004-01-28 Samsung Electronics Co., Ltd. Document content classification
CN101403703A (en) * 2008-11-07 2009-04-08 清华大学 Real-time detection method for foreign fiber in lint
CN101441771A (en) * 2008-12-19 2009-05-27 中国科学技术大学 Video fire hazard smoke detecting method based on color saturation degree and movement mode
CN101989302A (en) * 2010-10-22 2011-03-23 西安交通大学 Multilayer bitmap color feature-based image retrieval method
EP2390838A1 (en) * 2009-01-23 2011-11-30 Nec Corporation Image identifier extracting apparatus
CN102349091A (en) * 2009-03-13 2012-02-08 日本电气株式会社 Image identifier matching device

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101576896A (en) * 2008-05-09 2009-11-11 鸿富锦精密工业(深圳)有限公司;鸿海精密工业股份有限公司 Retrieval system and retrieval method for similar pictures
CN101290634B (en) * 2008-06-03 2010-07-28 北京搜狗科技发展有限公司 Method for recognizing repeated miniature, device and its uses in search engine
CN101539930B (en) * 2009-04-21 2011-06-15 武汉大学 Search method of related feedback images
JP5521436B2 (en) * 2009-08-19 2014-06-11 ソニー株式会社 Moving picture recording apparatus, a moving image recording method and program
CN101706793B (en) * 2009-11-16 2012-09-26 中兴通讯股份有限公司 Method and device for searching picture
CN102033937A (en) * 2010-12-20 2011-04-27 百度在线网络技术(北京)有限公司 Method and system for displaying video search result
CN102306287B (en) * 2011-08-24 2017-10-10 百度在线网络技术(北京)有限公司 Method and apparatus for image recognition sensitive

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1385329A2 (en) * 2002-07-27 2004-01-28 Samsung Electronics Co., Ltd. Document content classification
CN101403703A (en) * 2008-11-07 2009-04-08 清华大学 Real-time detection method for foreign fiber in lint
CN101441771A (en) * 2008-12-19 2009-05-27 中国科学技术大学 Video fire hazard smoke detecting method based on color saturation degree and movement mode
EP2390838A1 (en) * 2009-01-23 2011-11-30 Nec Corporation Image identifier extracting apparatus
CN102349091A (en) * 2009-03-13 2012-02-08 日本电气株式会社 Image identifier matching device
CN101989302A (en) * 2010-10-22 2011-03-23 西安交通大学 Multilayer bitmap color feature-based image retrieval method

Also Published As

Publication number Publication date Type
CN103377473A (en) 2013-10-30 application

Similar Documents

Publication Publication Date Title
US8494234B1 (en) Video hashing system and method
US20120008821A1 (en) Video visual and audio query
CN1761204A (en) System for blocking off erotic images and unhealthy information in internet
Basilio et al. Explicit image detection using YCbCr space color model as skin detection
CN101714257A (en) Method for main color feature extraction and structuring description of images
Tsai et al. Location coding for mobile image retrieval
JP2002216129A (en) Face area detector, its method and computer readable recording medium
US20050196044A1 (en) Method of extracting candidate human region within image, system for extracting candidate human region, program for extracting candidate human region, method of discerning top and bottom of human image, system for discerning top and bottom, and program for discerning top and bottom
Zhao et al. Detecting digital image splicing in chroma spaces
US20120300991A1 (en) Image preprocessing
Kaur et al. Comparison between YCbCr color space and CIELab color space for skin color segmentation
JP2009086926A (en) Image recognition method and device
US20090263014A1 (en) Content fingerprinting for video and/or image
CN103176761A (en) Screen background setting method and movable terminal
CN102254144A (en) Robust method for extracting two-dimensional code area in image
JPH09229646A (en) Object recognition method
US20100061587A1 (en) System, method, and apparatus for video fingerprinting
Thai et al. Statistical model of quantized DCT coefficients: Application in the steganalysis of Jsteg algorithm
Teng et al. Image indexing and retrieval based on vector quantization
US20140270344A1 (en) Reducing object detection time by utilizing space localization of features
Singh et al. Fast and efficient region duplication detection in digital images using sub-blocking method
US20110182509A1 (en) Image Preprocessing
CN101201939A (en) Method and system for generating picture identifying code
Agarwal et al. Content based image retrieval using color edge detection and discrete wavelet transform
Tang et al. Robust image hash function using local color features

Legal Events

Date Code Title Description
C06 Publication
C41 Transfer of patent application or patent right or utility model
COR Change of bibliographic data

Free format text: CORRECT: ADDRESS; FROM: 518057 SHENZHEN, GUANGDONG PROVINCE TO: 518044 SHENZHEN, GUANGDONG PROVINCE

ASS Succession or assignment of patent right

Owner name: SHENZHEN SHIJI LIGHT SPEED INFORMATION TECHNOLOGY

Free format text: FORMER OWNER: TENGXUN SCI-TECH (SHENZHEN) CO., LTD.

Effective date: 20131016

C10 Entry into substantive examination
GR01