WO2015120771A1 - 一种数字图像质量分级的方法和装置 - Google Patents

一种数字图像质量分级的方法和装置 Download PDF

Info

Publication number
WO2015120771A1
WO2015120771A1 PCT/CN2015/072114 CN2015072114W WO2015120771A1 WO 2015120771 A1 WO2015120771 A1 WO 2015120771A1 CN 2015072114 W CN2015072114 W CN 2015072114W WO 2015120771 A1 WO2015120771 A1 WO 2015120771A1
Authority
WO
WIPO (PCT)
Prior art keywords
digital image
value
quality
normalized
quality score
Prior art date
Application number
PCT/CN2015/072114
Other languages
English (en)
French (fr)
Inventor
郑琪
王永攀
Original Assignee
阿里巴巴集团控股有限公司
郑琪
王永攀
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
Application filed by 阿里巴巴集团控股有限公司, 郑琪, 王永攀 filed Critical 阿里巴巴集团控股有限公司
Priority to US15/116,828 priority Critical patent/US10026184B2/en
Publication of WO2015120771A1 publication Critical patent/WO2015120771A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/10Segmentation; Edge detection
    • G06T7/11Region-based segmentation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/0002Inspection of images, e.g. flaw detection
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20021Dividing image into blocks, subimages or windows
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30168Image quality inspection

Definitions

  • the present application relates to the field of computer communication technologies, and in particular, to a method and apparatus for digital image quality grading.
  • the existing digital image quality grading method is as follows: acquiring a digital image, and extracting a body region block in the digital image; calculating, according to the extracted body region block, the number, size, and position of the body included in the digital image; The number, size, and position, as well as the preset number threshold, size threshold, and bit threshold, are used to grade the digital image.
  • the existing digital image quality grading method needs to classify the digital image according to the number, size and position of the main body, and the preset number threshold, the size threshold and the bit threshold, and the number of elements required is relatively large, and the number is When the image is graded, the difficulty of setting the classification rule is relatively large, and the implementation process is cumbersome.
  • the technical problem to be solved by the present application is to provide a method and apparatus for digital image quality grading by calculating a first ratio value of the area of each body area block and the total area of the digital image, the area of the background area block, and the digital image.
  • the second ratio value of the total area, the normalized distance value of all the pixel points in each body region block from the central pixel point of the digital image, and the pre-predetermined distance value according to the first scale value, the second scale value, and the normalized distance value Set the digital image quality score conversion relationship, calculate the quality score value of the digital image, according to the quality score value of the digital image and the preset digital image quality
  • the threshold is used to classify the quality of the digital image, and the digital image quality threshold is easy to configure, which is simple and efficient.
  • the present application discloses a method for digital image quality grading, the method comprising:
  • each of the body area blocks a normalized distance value of all pixel points in a distance from a central pixel of the digital image; wherein the background area block is an area block remaining after n pieces of the body area block are extracted from the digital image;
  • the preset digital image quality score conversion relationship S is:
  • the S fi represents a first ratio value of an area of the i-th body block and a total area of the digital image
  • the S b represents an area of the background area block and a total of the digital image.
  • the n represents the number of the body region blocks extracted from the digital image
  • the R i represents all pixel points in the i-th body region block from the number The normalized distance value of the center pixel of the image.
  • calculating a normalized distance value of all pixel points in each of the body region blocks from a central pixel point of the digital image, including:
  • the r i represents a total value of the square root distance of all the pixels in the i-th body region block, Represents a selected normalized parameter value, the w representing the width of the digital image, and h representing the height of the digital image.
  • the method further includes:
  • classifying the quality of the digital image according to the quality score value of the digital image and the preset digital image quality threshold value including:
  • the quality of the digital image is ranked according to a normalized quality score value of the digital image and a preset normalized digital image quality sub-threshold.
  • classifying the quality of the digital image according to the quality score value of the digital image and the preset digital image quality threshold value including:
  • the level of the digital image is set to a level corresponding to a threshold value greater than a preset digital image quality
  • the level of the digital image is set to a level corresponding to a threshold value less than or equal to a preset digital image quality.
  • the present application also discloses an apparatus for digital image quality grading, the apparatus comprising:
  • An acquiring module configured to acquire a digital image, and extract n pieces of body regions from the digital image; wherein the n is a natural number;
  • a first calculation module configured to calculate a first ratio value of an area of each of the body area blocks and a total area of the digital image, a second ratio of an area of the background area block to a total area of the digital image, a normalized distance value of all pixel points in each of the body region blocks from a central pixel point of the digital image; wherein the background region block is obtained by extracting n pieces of the body region block in the digital image The remaining area blocks;
  • a second calculating module configured to calculate, according to the first proportional value, the second proportional value, and the normalized distance value, a quality score of the digital image by using a preset digital image quality score conversion relationship a value; wherein the preset digital image quality score conversion relationship is a function proportional to a sum of a first ratio value of all body region blocks, a second ratio value, and a normalized distance value of all body region blocks ;
  • a grading module configured to classify the quality of the digital image according to a quality score value of the digital image and a preset digital image quality sub-threshold.
  • the preset digital image quality score conversion relationship S is:
  • the S fi represents a first ratio value of an area of the i-th body block and a total area of the digital image
  • the S b represents an area of the background area block and a total of the digital image.
  • the n represents the number of the body region blocks extracted from the digital image
  • the R i represents all pixel points in the i-th body region block from the number The normalized distance value of the center pixel of the image.
  • the first calculation module includes:
  • a first calculating unit configured to calculate each pixel in each of the body region blocks and the number The square root distance value between the central pixels of the image
  • a summation unit configured to sum the square root distance values of all the pixels in each of the body region blocks to obtain a total square root distance value of all the pixel points in each of the body region blocks;
  • a second calculating unit configured to calculate a distance of all pixel points in each of the body region blocks by using a preset distance normalized conversion relationship according to a total value of square root distances of all pixel points in each of the body region blocks a normalized distance value of a central pixel of the digital image, wherein the preset distance normalized conversion relationship is a square root distance of all pixels in each of the body region blocks according to the selected normalized parameter value The total value is normalized.
  • the r i represents a total value of the square root distance of all the pixels in the i-th body region block, Represents a selected normalized parameter value, the w representing the width of the digital image, and h representing the height of the digital image.
  • the second calculation module includes:
  • a processing unit configured to use the calculated quotient of the quality score value of the digital image and the optimal theoretical quality score value as a normalized quality score value of the digital image;
  • the grading module comprises:
  • a grading unit configured to classify the quality of the digital image according to a normalized quality score value of the digital image and a preset normalized digital image quality sub-threshold.
  • the obtaining unit includes:
  • Selecting a subunit configured to determine a relationship between the first ratio value, the second ratio value, and the normalized distance value according to a quality relationship between the number of the body region blocks and the digital image, Selecting the values of the n, the S fi , the S b , and the R i ;
  • a processing subunit configured to substitute the selected value of the n, the S fi , the S b , and the R i into a preset digital image quality score conversion relationship S, and calculate the maximum value as a The optimal theoretical mass score.
  • the grading module includes:
  • a comparison unit configured to divide the quality score of the digital image and the preset digital image quality Threshold comparison
  • a first grading unit configured to set a level of the digital image to a level corresponding to a preset digital image quality sub-threshold if the quality score value of the digital image is greater than a preset digital image quality sub-threshold;
  • a second grading unit configured to set a level of the digital image to be equal to or less than a preset digital image quality threshold if the quality score of the digital image is less than or equal to a preset digital image quality threshold level.
  • the quality score value of the digital image is calculated, according to the number
  • the image quality score value and the preset digital image quality sub-threshold value the quality of the digital image is graded, and the digital image quality sub-threshold value is easy to configure, and the implementation is simple and efficient.
  • FIG. 1 is a flow chart of a method for digital image quality grading according to an embodiment of the present application
  • FIG. 2 is a schematic diagram of a digital image quality grading according to an embodiment of the present application.
  • FIG. 3 is a schematic structural diagram of an apparatus for digital image quality grading according to an embodiment of the present application.
  • a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
  • processors CPUs
  • input/output interfaces network interfaces
  • memory volatile and non-volatile memory
  • the memory may include non-persistent memory, random access memory (RAM), and/or non-volatile memory in a computer readable medium, such as read only memory (ROM) or flash memory.
  • RAM random access memory
  • ROM read only memory
  • Memory is an example of a computer readable medium.
  • Computer readable media includes both permanent and non-persistent, removable and non-removable media.
  • Information storage can be implemented by any method or technology.
  • the information can be computer readable instructions, data structures, modules of programs, or other data.
  • Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory. (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disk read only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, Magnetic tape cartridges, magnetic tape storage or other magnetic storage devices or any other non-transportable media can be used to store information that can be accessed by a computing device.
  • computer readable media does not include non-transitory computer readable media, such as modulated data signals and carrier waves.
  • FIG. 1 is a flowchart of a method for digital image quality grading according to an embodiment of the present application, where the method includes:
  • S101 Acquire a digital image, and extract n pieces of body regions from the digital image, where n is a natural number.
  • one body region block may be extracted from the digital image, or multiple body region blocks may be extracted.
  • any feasible implementation method may be adopted, for example, a salient region extraction method or the like may be used, which is not limited thereto.
  • S102 Calculate a first ratio of an area of each body area block to a total area of the digital image, A second ratio value of the area of the background area block to the total area of the digital image, and a normalized distance value of the central pixel point of the digital image from all pixel points in each body area block.
  • the background area block is an area block remaining after extracting n body area blocks in the digital image.
  • calculating a normalized distance value of all pixel points in each body region block from a central pixel point of the digital image including:
  • the square root distance values of all the pixels in each body region block are summed to obtain the total square root distance distance of all the pixels in each body region block;
  • the normalized conversion relationship is calculated by using the preset distance, and the normalized distance of all the pixel points in each body region block from the central pixel point of the digital image is calculated.
  • the value, wherein the preset distance normalized conversion relationship is a normalization process for the total square root distance of all pixel points in each body region block according to the selected normalized parameter value.
  • the normalized parameter value when the normalized parameter value is selected, it can be flexibly selected according to the actual application condition, which is not limited.
  • the square root of the sum of the square of the width of the digital image and the square of the height of the digital image is selected as the return.
  • the distance normalized conversion relationship R i preset in this embodiment is:
  • r i represents the total square root distance of all pixels in the i-th body region block
  • w represents the width of the digital image
  • h represents the height of the digital image
  • the conversion relationship R i is normalized according to the preset distance, and the calculated R i value is the normalized distance value of all the pixel points in each body region block from the central pixel point of the digital image.
  • S103 Calculate a quality score value of the digital image by using a preset digital image quality score conversion relationship according to the first scale value, the second scale value, and the normalized distance value.
  • the preset digital image quality score conversion relationship is proportional to the sum of the first ratio values of all body region blocks, the second ratio value, and the sum of the normalized distance values of all body region blocks.
  • the function The specific expression form of the preset digital image quality score conversion relationship can be flexibly set according to the actual application condition, which is not limited.
  • the preset digital image quality score conversion relationship S is: Wherein, S fi represents a first ratio of the area of the i-th body region block to the total area of the digital image, S b represents a second ratio of the area of the background region block to the total area of the digital image, and n represents the slave digital image.
  • the number of extracted body region blocks, R i represents the normalized distance value of all pixel points in the i-th body region block from the central pixel point of the digital image.
  • the calculated S value is the quality score value of the digital image.
  • the method further includes:
  • S104 Classify the quality of the digital image according to the quality score value of the digital image and the preset digital image quality sub-threshold.
  • the quality score value of the digital image is greater than a preset digital image quality sub-threshold, setting the level of the digital image to a level corresponding to a threshold value greater than a preset digital image quality;
  • the level of the digital image is set to a level corresponding to a threshold value less than or equal to the preset digital image quality.
  • the digital image quality sub-threshold value may be set according to manual experience, or may be manually labeled with different levels of digital images, and obtained by a machine learning method.
  • the quality of the digital image is graded according to the quality score value of the digital image and the preset digital image quality threshold value, including:
  • the quality of the digital image is ranked based on the normalized quality score of the digital image and the preset normalized digital image quality threshold.
  • the first ratio of the area of the first body area block to the total area of the digital image is calculated by the method of the embodiment.
  • S f1 is: 0.3622
  • the second ratio S b of the area of the background area block and the total area of the digital image is: 0.6378
  • the value R 1 is: 0.4451
  • the mass score S of the digital image is calculated as: 97.
  • the quality of the digital image of FIG. 2 is ranked as: a high quality subject based on the normalized quality score value 97 of the digital image and the preset normalized digital image quality sub-threshold.
  • the method for classifying digital image quality by calculating a first ratio value of an area of each body region block and a total area of the digital image, a second ratio of an area of the background region block to a total area of the digital image a normalized distance value of all pixel points in each body region block from a central pixel point of the digital image, using a preset digital image quality score conversion according to the first scale value, the second scale value, and the normalized distance value
  • the relationship is calculated, and the quality score value of the digital image is calculated.
  • the quality score value of the digital image and the preset digital image quality threshold value the quality of the digital image is graded, and the digital image quality threshold is easy to configure, and the implementation is simple and efficient.
  • FIG. 3 it is a device structure diagram of a digital image quality grading according to an embodiment of the present application.
  • the device includes:
  • the obtaining module 201 is configured to acquire a digital image, and extract n pieces of the body region from the digital image; wherein n is a natural number;
  • the first calculating module 202 is configured to calculate a first ratio value of an area of each body area block and a total area of the digital image, a second ratio value of an area of the background area block and a total area of the digital image, and each body area block. a normalized distance value of all the pixels in the distance from the central pixel of the digital image; wherein the background region block is a region block remaining after extracting n body region blocks in the digital image;
  • the second calculating module 203 is configured to calculate a quality score value of the digital image by using a preset digital image quality score conversion relationship according to the first scale value, the second scale value, and the normalized distance value; wherein, the preset The digital image quality score conversion relationship is a function proportional to the sum of the first scale values of all body region blocks, the second scale value, and the sum of the normalized distance values of all body region blocks;
  • the grading module 204 is configured to classify the quality of the digital image according to the quality score value of the digital image and the preset digital image quality sub-threshold.
  • the preset digital image quality score conversion relationship S is:
  • S fi represents a first ratio of the area of the i-th body region block to the total area of the digital image
  • S b represents a second ratio of the area of the background region block to the total area of the digital image
  • n represents the slave digital image.
  • the number of extracted body region blocks, R i represents the normalized distance value of all pixel points in the i-th body region block from the central pixel point of the digital image.
  • the first calculation module 202 includes:
  • a first calculating unit configured to calculate a square root distance value between each pixel point in each body region block and a central pixel point of the digital image
  • a summation unit for summing the square root distance values of all the pixels in each body region block to obtain a total square root distance value of all the pixels in each body region block
  • a second calculating unit configured to calculate a distance from all pixel points in each body region block to the center of the digital image according to a total distance of the square root distance of all the pixel points in each body region block by using a preset distance normalization conversion relationship a normalized distance value of a pixel, wherein the preset distance normalized conversion relationship is a square of all pixel points in each body region block according to the selected normalized parameter value
  • the root distance is normalized to the total value.
  • the preset distance normalized conversion relationship R i is:
  • r i represents the total square root distance of all pixels in the i-th body region block
  • w represents the width of the digital image
  • h represents the height of the digital image
  • the second calculating module 203 comprises:
  • a processing unit configured to calculate a quality score value of the calculated digital image and the optimal theoretical quality score value as a normalized quality score value of the digital image
  • the ranking module 204 includes:
  • the grading unit is configured to classify the quality of the digital image according to the normalized quality score value of the digital image and the preset normalized digital image quality sub-threshold.
  • the obtaining unit comprises:
  • the processing subunit is configured to substitute the value of the selected n, S fi , S b , and R i into the preset digital image quality score conversion relationship S, and calculate the maximum value as the optimal theoretical quality score value.
  • the ranking module 204 comprises:
  • a comparing unit configured to compare a quality score value of the digital image with a preset digital image quality threshold value
  • a first grading unit configured to set a level of the digital image to a level corresponding to a preset digital image quality sub-threshold value if the quality score value of the digital image is greater than a preset digital image quality sub-threshold value;
  • a second grading unit configured to set the level of the digital image to a level corresponding to a threshold value less than or equal to a preset digital image quality threshold if the quality score value of the digital image is less than or equal to a preset digital image quality sub-threshold value.
  • the device corresponds to the foregoing method flow description, and the deficiencies refer to the description of the above method flow, and will not be further described.
  • the apparatus for classifying the digital image quality by calculating a first ratio value of the area of each body area block and the total area of the digital image, the area of the background area block, and the second ratio of the total area of the digital image. a normalized distance value of all pixel points in each body region block from a central pixel point of the digital image, using a preset digital image quality score conversion according to the first scale value, the second scale value, and the normalized distance value The relationship is calculated, and the quality score value of the digital image is calculated. According to the quality score value of the digital image and the preset digital image quality threshold value, the quality of the digital image is graded, and the digital image quality threshold is easy to configure, and the implementation is simple and efficient.

Abstract

本申请公开了一种数字图像质量分级的方法和装置,属于计算机通信技术领域。方法包括:获取数字图像,并从所述数字图像中提取出n个主体区域块;计算每个所述主体区域块的面积与所述数字图像的总面积的第一比例值、背景区域块的面积与所述数字图像的总面积的第二比例值、每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值;根据所述第一比例值、所述第二比例值和所述归一化距离值,利用预设的数字图像质量分数转换关系,计算得到所述数字图像的质量分数值;根据所述数字图像的质量分数值和预设的数字图像质量分阈值,对所述数字图像的质量进行分级。本申请实现简单、高效。

Description

一种数字图像质量分级的方法和装置 技术领域
本申请涉及计算机通信技术领域,具体涉及一种数字图像质量分级的方法和装置。
背景技术
随着数字拍摄设备的普及,现在的个人和企业等管理着越来越多的数字图像。自动删除质量差的数字图像,保留质量好的数字图像,在个人相册管理、商品图片管理等场合都有着重要应用价值。为了可以实现上述功能,现有常常会对数字图像的质量进行分级,根据数字图像质量的分级结果,来确定数字图像的质量,从而进一步确定是删除数字图像还是保留数字图像。因此,如何更好地实现对数字图像的质量进行分级,成为目前研究的热点。
现有数字图像质量分级的方法如下:获取数字图像,并提取数字图像中的主体区域块;根据提取出的主体区域块,计算数字图像中包括的主体的个数、大小和位置;根据主体的个数、大小和位置,以及预设的个数阈值、大小阈值和位阈值置,对数字图像进行质量分级。
现有数字图像质量分级的方法需要根据主体的个数、大小和位置,以及预设的个数阈值、大小阈值和位阈值置,对数字图像进行质量分级,需要的元素数量比较多,对数字图像进行质量分级时,设置分级规则的难度比较大,实现过程比较繁琐。
发明内容
本申请所要解决的技术问题在于提供一种数字图像质量分级的方法和装置,通过计算每个主体区域块的面积与数字图像的总面积的第一比例值、背景区域块的面积与数字图像的总面积的第二比例值、每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值,根据第一比例值、第二比例值和归一化距离值,利用预设的数字图像质量分数转换关系,计算得到数字图像的质量分数值,根据数字图像的质量分数值和预设的数字图像质 量分阈值,对数字图像的质量进行分级,数字图像质量分阈值易于配置,实现简单、高效。
为了解决上述问题,本申请公开了一种数字图像质量分级的方法,所述方法包括:
获取数字图像,并从所述数字图像中提取出n个主体区域块;其中,所述n为自然数;
计算每个所述主体区域块的面积与所述数字图像的总面积的第一比例值、背景区域块的面积与所述数字图像的总面积的第二比例值、每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值;其中,所述背景区域块为所述数字图像中提取出n个所述主体区域块后剩余的区域块;
根据所述第一比例值、所述第二比例值和所述归一化距离值,利用预设的数字图像质量分数转换关系,计算得到所述数字图像的质量分数值;其中,预设的数字图像质量分数转换关系是与所有主体区域块的第一比例值之和、所第二比例值和所有主体区域块的归一化距离值之和成正比例关系的函数;根据所述数字图像的质量分数值和预设的数字图像质量分阈值,对所述数字图像的质量进行分级。
进一步地,预设的数字图像质量分数转换关系S为:
Figure PCTCN2015072114-appb-000001
其中,所述Sfi表示第i个所述主体区域块的面积与所述数字图像的总面积的第一比例值,所述Sb表示所述背景区域块的面积与所述数字图像的总面积的第二比例值,所述n表示从所述数字图像中提取出的所述主体区域块的个数,所述Ri表示第i个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值。
进一步地,计算每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值,包括:
计算每个所述主体区域块中每个像素点与所述数字图像的中心像素点之间的平方根距离值;
将每个所述主体区域块中所有像素点的平方根距离值求和,得到每个所 述主体区域块中所有像素点的平方根距离总值;
根据每个所述主体区域块中所有像素点的平方根距离总值,利用预设的距离归一化转换关系,计算得到每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值,其中,预设的距离归一化转换关系是根据选择的归一化参数值对每个所述主体区域块中所有像素点的平方根距离总值进行归一化处理。
进一步地,预设的距离归一化转换关系Ri为:
Figure PCTCN2015072114-appb-000002
其中,所述ri表示第i个所述主体区域块中所有像素点的平方根距离总值,所述
Figure PCTCN2015072114-appb-000003
表示选择的归一化参数值,所述w表示所述数字图像的宽度,所述h表示所述数字图像的高度。
进一步地,计算得到所述数字图像的质量分数值之后,还包括:
获取最优理论质量分数值;
将计算得到的所述数字图像的质量分数值与所述最优理论质量分数值的商,作为所述数字图像的归一化质量分数值;
相应地,根据所述数字图像的质量分数值和预设的数字图像质量分阈值,对所述数字图像的质量进行分级,包括:
根据所述数字图像的归一化质量分数值和预设的归一化数字图像质量分阈值,对所述数字图像的质量进行分级。
进一步地,获取最优理论质量分数值,包括:
根据所述主体区域块的个数与所述数字图像的质量关系,所述第一比例值、所述第二比例值和所述归一化距离值之间的关系,选择所述n、所述Sfi、所述Sb、所述Ri的取值;
将选择的所述n、所述Sfi、所述Sb、所述Ri的取值代入预设的数字图像质量分数转换关系S,计算得到的最大值,作为所述最优理论质量分数值。
进一步地,根据所述数字图像的质量分数值和预设的数字图像质量分阈值,对所述数字图像的质量进行分级,包括:
将所述数字图像的质量分数值和预设的数字图像质量分阈值进行比较;
如果所述数字图像的质量分数值大于预设的数字图像质量分阈值,则将 所述数字图像的级别设置为与大于预设的数字图像质量分阈值相应的级别;
如果所述数字图像的质量分数值小于等于预设的数字图像质量分阈值,则将所述数字图像的级别设置为与小于等于预设的数字图像质量分阈值相应的级别。
为了解决上述问题,本申请还公开了一种数字图像质量分级的装置,所述装置包括:
获取模块,用于获取数字图像,并从所述数字图像中提取出n个主体区域块;其中,所述n为自然数;
第一计算模块,用于计算每个所述主体区域块的面积与所述数字图像的总面积的第一比例值、背景区域块的面积与所述数字图像的总面积的第二比例值、每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值;其中,所述背景区域块为所述数字图像中提取出n个所述主体区域块后剩余的区域块;
第二计算模块,用于根据所述第一比例值、所述第二比例值和所述归一化距离值,利用预设的数字图像质量分数转换关系,计算得到所述数字图像的质量分数值;其中,预设的数字图像质量分数转换关系是与所有主体区域块的第一比例值之和、所第二比例值和所有主体区域块的归一化距离值之和成正比例关系的函数;
分级模块,用于根据所述数字图像的质量分数值和预设的数字图像质量分阈值,对所述数字图像的质量进行分级。
进一步地,预设的数字图像质量分数转换关系S为:
Figure PCTCN2015072114-appb-000004
其中,所述Sfi表示第i个所述主体区域块的面积与所述数字图像的总面积的第一比例值,所述Sb表示所述背景区域块的面积与所述数字图像的总面积的第二比例值,所述n表示从所述数字图像中提取出的所述主体区域块的个数,所述Ri表示第i个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值。
进一步地,所述第一计算模块包括:
第一计算单元,用于计算每个所述主体区域块中每个像素点与所述数字 图像的中心像素点之间的平方根距离值;
求和单元,用于将每个所述主体区域块中所有像素点的平方根距离值求和,得到每个所述主体区域块中所有像素点的平方根距离总值;
第二计算单元,用于根据每个所述主体区域块中所有像素点的平方根距离总值,利用预设的距离归一化转换关系,计算得到每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值,其中,预设的距离归一化转换关系是根据选择的归一化参数值对每个所述主体区域块中所有像素点的平方根距离总值进行归一化处理。
进一步地,预设的距离归一化转换关系Ri为:
Figure PCTCN2015072114-appb-000005
其中,所述ri表示第i个所述主体区域块中所有像素点的平方根距离总值,所述
Figure PCTCN2015072114-appb-000006
表示选择的归一化参数值,所述w表示所述数字图像的宽度,所述h表示所述数字图像的高度。
进一步地,所述第二计算模块包括:
获取单元,用于获取最优理论质量分数值;
处理单元,用于将计算得到的所述数字图像的质量分数值与所述最优理论质量分数值的商,作为所述数字图像的归一化质量分数值;
相应地,所述分级模块包括:
分级单元,用于根据所述数字图像的归一化质量分数值和预设的归一化数字图像质量分阈值,对所述数字图像的质量进行分级。
进一步地,所述获取单元包括:
选择子单元,用于根据所述主体区域块的个数与所述数字图像的质量关系,所述第一比例值、所述第二比例值和所述归一化距离值之间的关系,选择所述n、所述Sfi、所述Sb、所述Ri的取值;
处理子单元,用于将选择的所述n、所述Sfi、所述Sb、所述Ri的取值代入预设的数字图像质量分数转换关系S,计算得到的最大值,作为所述最优理论质量分数值。
进一步地,所述分级模块包括:
比较单元,用于将所述数字图像的质量分数值和预设的数字图像质量分 阈值进行比较;
第一分级单元,用于如果所述数字图像的质量分数值大于预设的数字图像质量分阈值,则将所述数字图像的级别设置为与大于预设的数字图像质量分阈值相应的级别;
第二分级单元,用于如果所述数字图像的质量分数值小于等于预设的数字图像质量分阈值,则将所述数字图像的级别设置为与小于等于预设的数字图像质量分阈值相应的级别。
与现有技术相比,本申请可以获得包括以下技术效果:
通过计算每个主体区域块的面积与数字图像的总面积的第一比例值、背景区域块的面积与数字图像的总面积的第二比例值、每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值,根据第一比例值、第二比例值和归一化距离值,利用预设的数字图像质量分数转换关系,计算得到数字图像的质量分数值,根据数字图像的质量分数值和预设的数字图像质量分阈值,对数字图像的质量进行分级,数字图像质量分阈值易于配置,实现简单、高效。
当然,实施本申请的任一产品必不一定需要同时达到以上所述的所有技术效果。
附图说明
此处所说明的附图用来提供对本申请的进一步理解,构成本申请的一部分,本申请的示意性实施例及其说明用于解释本申请,并不构成对本申请的不当限定。在附图中:
图1是本申请实施例的一种数字图像质量分级的方法流程图;
图2是本申请实施例的一种数字图像质量分级的示意图;
图3是本申请实施例的一种数字图像质量分级的装置结构示意图。
具体实施方式
以下将配合附图及实施例来详细说明本申请的实施方式,藉此对本申请 如何应用技术手段来解决技术问题并达成技术功效的实现过程能充分理解并据以实施。
在一个典型的配置中,计算设备包括一个或多个处理器(CPU)、输入/输出接口、网络接口和内存。
内存可能包括计算机可读介质中的非永久性存储器,随机存取存储器(RAM)和/或非易失性内存等形式,如只读存储器(ROM)或闪存(flash RAM)。内存是计算机可读介质的示例。
计算机可读介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括,但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带,磁带磁磁盘存储或其他磁性存储设备或任何其他非传输介质,可用于存储可以被计算设备访问的信息。按照本文中的界定,计算机可读介质不包括非暂存电脑可读媒体(transitory media),如调制的数据信号和载波。
实施例描述
下面以一实施例对本申请方法的实现作进一步说明。如图1所示,为本申请实施例的一种数字图像质量分级的方法流程图,该方法包括:
S101:获取数字图像,并从数字图像中提取出n个主体区域块,其中,n为自然数。
具体地,数字图像中的主体,可以有一个,也可以有多个,相应地,从数字图像中可以提取出一个主体区域块,也可以提取出多个主体区域块。并且,在从数字图像中提取主体区域块时,可以采用任何可行的实现方法,比如可以采用显著区域提取方法等,对此不做限定。
S102:计算每个主体区域块的面积与数字图像的总面积的第一比例值、 背景区域块的面积与数字图像的总面积的第二比例值、每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值。
其中,背景区域块为数字图像中提取出n个主体区域块后剩余的区域块。
具体地,计算每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值,包括:
计算每个主体区域块中每个像素点与数字图像的中心像素点之间的平方根距离值(即主体区域块的像素点的坐标与中心像素点的坐标之间作差、平方求和、然后开方);
将每个主体区域块中所有像素点的平方根距离值求和,得到每个主体区域块中所有像素点的平方根距离总值;
根据每个主体区域块中所有像素点的平方根距离总值,利用预设的距离归一化转换关系,计算得到每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值,其中,预设的距离归一化转换关系是根据选择的归一化参数值对每个主体区域块中所有像素点的平方根距离总值进行归一化处理。
具体地,在选择归一化参数值时,可以根据实际应用状况灵活选择,对此不做限定,本实施例中选择数字图像的宽度的平方和数字图像的高度的平方之和的平方根作为归一化参数值,本实施例中预设的距离归一化转换关系Ri为:
Figure PCTCN2015072114-appb-000007
其中,ri表示第i个主体区域块中所有像素点的平方根距离总值,
Figure PCTCN2015072114-appb-000008
表示选择的归一化参数值,w表示数字图像的宽度,h表示数字图像的高度。
根据预设的距离归一化转换关系Ri,计算得到的Ri值即为每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值。
S103:根据第一比例值、第二比例值和归一化距离值,利用预设的数字图像质量分数转换关系,计算得到数字图像的质量分数值。
其中,预设的数字图像质量分数转换关系是与所有主体区域块的第一比例值之和、第二比例值和所有主体区域块的归一化距离值之和成正比例关系 的函数。可以根据实际应用状况,灵活设置预设的数字图像质量分数转换关系的具体表现形式,对此不做限定。
本实施例中,预设的数字图像质量分数转换关系S为:
Figure PCTCN2015072114-appb-000009
其中,Sfi表示第i个主体区域块的面积与数字图像的总面积的第一比例值,Sb表示背景区域块的面积与数字图像的总面积的第二比例值,n表示从数字图像中提取出的主体区域块的个数,Ri表示第i个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值。
根据预设的数字图像质量分数转换关系S,计算得到的S值,即为数字图像的质量分数值。
并且,计算得到数字图像的质量分数值之后,还包括:
获取最优理论质量分数值;
将计算得到的数字图像的质量分数值除以最优理论质量分数值得到的商,作为数字图像的归一化质量分数值。
其中,获取最优理论质量分数值,包括:
根据数字图像中包括的主体区域块个数与数字图像的质量关系,第一比例值、第二比例值和归一化距离值之间的关系,选择n、Sfi、Sb、Ri的取值;
将选择的n、Sfi、Sb、Ri的取值代入预设的数字图像质量分数转换关系S,计算得到的最大值作为最优理论质量分数值。
需要说明的是,当数字图像只有一个主体并居中时,数字图像的质量分数值更高,所以选择n为1,Ri的像素点居中,并且通过选择不同的Sfi、Sb的取值,可以得到当Sfi为37%、Sb为63%时,会得到最优理论质量分数值(约为0.13)。
S104:根据数字图像的质量分数值和预设的数字图像质量分阈值,对数字图像的质量进行分级。
具体地,根据数字图像的质量分数值和预设的数字图像质量分阈值,对数字图像的质量进行分级包括:
将数字图像的质量分数值和预设的数字图像质量分阈值进行比较;
如果数字图像的质量分数值大于预设的数字图像质量分阈值,则将数字图像的级别设置为与大于预设的数字图像质量分阈值相应的级别;
如果数字图像的质量分数值小于等于预设的数字图像质量分阈值,则将数字图像的级别设置为与小于等于预设的数字图像质量分阈值相应的级别。
其中,数字图像质量分阈值可以根据人工经验来设定,也可以人工标注不同级别的数字图像,通过机器学习的方法来获得。
并且,并不限于通过上面的方法对数字图像的质量进行分级,还可以采用其他任何可行的方法实现,对此不作具体限定。
另外,当步骤S103获取的是归一化质量分数值时,相应地,根据数字图像的质量分数值和预设的数字图像质量分阈值,对数字图像的质量进行分级,包括:
根据数字图像的归一化质量分数值和预设的归一化数字图像质量分阈值,对数字图像的质量进行分级。
例如,参见图2,其中的主体区域块(只有一个)为鞋所在的区域块,利用本实施例的方法,计算得到第1个主体区域块的面积与数字图像的总面积的第一比例值Sf1为:0.3622,背景区域块的面积与数字图像的总面积的第二比例值Sb为:0.6378,第1个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值R1为:0.4451,利用预设的数字图像质量分数转换关系S:
Figure PCTCN2015072114-appb-000010
计算得到数字图像的质量分数值S为:97。根据数字图像的归一化质量分数值97和预设的归一化数字图像质量分阈值,将图2的数字图像的质量分级为:高质量主体。
本实施例所述的数字图像质量分级的方法,通过计算每个主体区域块的面积与数字图像的总面积的第一比例值、背景区域块的面积与数字图像的总面积的第二比例值、每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值,根据第一比例值、第二比例值和归一化距离值,利用预设的数字图像质量分数转换关系,计算得到数字图像的质量分数值,根据数字图像的质量分数值和预设的数字图像质量分阈值,对数字图像的质量进行分级,数字图像质量分阈值易于配置,实现简单、高效。
如图3所示,是本申请实施例的一种数字图像质量分级的装置结构图, 该装置包括:
获取模块201,用于获取数字图像,并从数字图像中提取出n个主体区域块;其中,n为自然数;
第一计算模块202,用于计算每个主体区域块的面积与数字图像的总面积的第一比例值、背景区域块的面积与数字图像的总面积的第二比例值、每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值;其中,背景区域块为数字图像中提取出n个主体区域块后剩余的区域块;
第二计算模块203,用于根据第一比例值、第二比例值和归一化距离值,利用预设的数字图像质量分数转换关系,计算得到数字图像的质量分数值;其中,预设的数字图像质量分数转换关系是与所有主体区域块的第一比例值之和、第二比例值和所有主体区域块的归一化距离值之和成正比例关系的函数;
分级模块204,用于根据数字图像的质量分数值和预设的数字图像质量分阈值,对数字图像的质量进行分级。
优选地,预设的数字图像质量分数转换关系S为:
Figure PCTCN2015072114-appb-000011
其中,Sfi表示第i个主体区域块的面积与数字图像的总面积的第一比例值,Sb表示背景区域块的面积与数字图像的总面积的第二比例值,n表示从数字图像中提取出的主体区域块的个数,Ri表示第i个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值。
优选地,第一计算模块202包括:
第一计算单元,用于计算每个主体区域块中每个像素点与数字图像的中心像素点之间的平方根距离值;
求和单元,用于将每个主体区域块中所有像素点的平方根距离值求和,得到每个主体区域块中所有像素点的平方根距离总值;
第二计算单元,用于根据每个主体区域块中所有像素点的平方根距离总值,利用预设的距离归一化转换关系,计算得到每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值,其中,预设的距离归一化转换关系是根据选择的归一化参数值对每个主体区域块中所有像素点的平方 根距离总值进行归一化处理。
优选地,预设的距离归一化转换关系Ri为:
Figure PCTCN2015072114-appb-000012
其中,ri表示第i个主体区域块中所有像素点的平方根距离总值,
Figure PCTCN2015072114-appb-000013
表示选择的归一化参数值,w表示数字图像的宽度,h表示数字图像的高度。
优选地,第二计算模块203包括:
获取单元,用于获取最优理论质量分数值;
处理单元,用于将计算得到的数字图像的质量分数值与所述最优理论质量分数值的商,作为数字图像的归一化质量分数值;
相应地,分级模块204包括:
分级单元,用于根据数字图像的归一化质量分数值和预设的归一化数字图像质量分阈值,对数字图像的质量进行分级。
优选地,获取单元包括:
选择子单元,用于根据主体区域块的个数与数字图像的质量关系,第一比例值、第二比例值和归一化距离值之间的关系,选择n、Sfi、Sb、Ri的取值;
处理子单元,用于将选择的n、Sfi、Sb、Ri的取值代入预设的数字图像质量分数转换关系S,计算得到的最大值,作为最优理论质量分数值。
优选地,分级模块204包括:
比较单元,用于将数字图像的质量分数值和预设的数字图像质量分阈值进行比较;
第一分级单元,用于如果数字图像的质量分数值大于预设的数字图像质量分阈值,则将数字图像的级别设置为与大于预设的数字图像质量分阈值相应的级别;
第二分级单元,用于如果数字图像的质量分数值小于等于预设的数字图像质量分阈值,则将数字图像的级别设置为与小于等于预设的数字图像质量分阈值相应的级别。
所述装置与前述的方法流程描述对应,不足之处参考上述方法流程的叙述,不再一一赘述。
本实施例所述的数字图像质量分级的装置,通过计算每个主体区域块的面积与数字图像的总面积的第一比例值、背景区域块的面积与数字图像的总面积的第二比例值、每个主体区域块中所有像素点距离数字图像的中心像素点的归一化距离值,根据第一比例值、第二比例值和归一化距离值,利用预设的数字图像质量分数转换关系,计算得到数字图像的质量分数值,根据数字图像的质量分数值和预设的数字图像质量分阈值,对数字图像的质量进行分级,数字图像质量分阈值易于配置,实现简单、高效。
上述说明示出并描述了本申请的若干优选实施例,但如前所述,应当理解本申请并非局限于本文所披露的形式,不应看作是对其他实施例的排除,而可用于各种其他组合、修改和环境,并能够在本文所述发明构想范围内,通过上述教导或相关领域的技术或知识进行改动。而本领域人员所进行的改动和变化不脱离本申请的精神和范围,则都应在本申请所附权利要求的保护范围内。

Claims (14)

  1. 一种数字图像质量分级的方法,其特征在于,所述方法包括:
    获取数字图像,并从所述数字图像中提取出n个主体区域块;其中,所述n为自然数;
    计算每个所述主体区域块的面积与所述数字图像的总面积的第一比例值、背景区域块的面积与所述数字图像的总面积的第二比例值、每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值;其中,所述背景区域块为所述数字图像中提取出n个所述主体区域块后剩余的区域块;
    根据所述第一比例值、所述第二比例值和所述归一化距离值,利用预设的数字图像质量分数转换关系,计算得到所述数字图像的质量分数值;其中,预设的数字图像质量分数转换关系是与所有所述主体区域块的第一比例值之和、所述第二比例值和所有所述主体区域块的归一化距离值之和成正比例关系的函数;
    根据所述数字图像的质量分数值和预设的数字图像质量分阈值,对所述数字图像的质量进行分级。
  2. 如权利要求1所述的方法,其特征在于,预设的数字图像质量分数转换关系S为:
    Figure PCTCN2015072114-appb-100001
    其中,所述Sfi表示第i个所述主体区域块的面积与所述数字图像的总面积的第一比例值,所述Sb表示所述背景区域块的面积与所述数字图像的总面积的第二比例值,所述n表示从所述数字图像中提取出的所述主体区域块的个数,所述Ri表示第i个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值。
  3. 如权利要求1所述的方法,其特征在于,计算每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值,包括:
    计算每个所述主体区域块中每个像素点与所述数字图像的中心像素点之间的平方根距离值;
    将每个所述主体区域块中所有像素点的平方根距离值求和,得到每个所 述主体区域块中所有像素点的平方根距离总值;
    根据每个所述主体区域块中所有像素点的平方根距离总值,利用预设的距离归一化转换关系,计算得到每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值,其中,预设的距离归一化转换关系是根据选择的归一化参数值对每个所述主体区域块中所有像素点的平方根距离总值进行归一化处理。
  4. 如权利要求3所述的方法,其特征在于,预设的距离归一化转换关系Ri为:
    Figure PCTCN2015072114-appb-100002
    其中,所述ri表示第i个所述主体区域块中所有像素点的平方根距离总值,所述
    Figure PCTCN2015072114-appb-100003
    表示选择的归一化参数值,所述w表示所述数字图像的宽度,所述h表示所述数字图像的高度。
  5. 如权利要求1所述的方法,其特征在于,计算得到所述数字图像的质量分数值之后,还包括:
    获取最优理论质量分数值;
    将计算得到的所述数字图像的质量分数值与所述最优理论质量分数值的商,作为所述数字图像的归一化质量分数值;
    相应地,根据所述数字图像的质量分数值和预设的数字图像质量分阈值,对所述数字图像的质量进行分级,包括:
    根据所述数字图像的归一化质量分数值和预设的归一化数字图像质量分阈值,对所述数字图像的质量进行分级。
  6. 如权利要求5所述的方法,其特征在于,获取最优理论质量分数值,包括:
    根据所述主体区域块的个数与所述数字图像的质量关系,所述第一比例值、所述第二比例值和所述归一化距离值之间的关系,选择所述n、所述Sfi、所述Sb、所述Ri的取值;
    将选择的所述n、所述Sfi、所述Sb、所述Ri的取值代入预设的数字图像质量分数转换关系S,计算得到的最大值,作为所述最优理论质量分数值。
  7. 如权利要求1-6任一权利要求所述的方法,其特征在于,根据所述数 字图像的质量分数值和预设的数字图像质量分阈值,对所述数字图像的质量进行分级,包括:
    将所述数字图像的质量分数值和预设的数字图像质量分阈值进行比较;
    如果所述数字图像的质量分数值大于预设的数字图像质量分阈值,则将所述数字图像的级别设置为与大于预设的数字图像质量分阈值相应的级别;
    如果所述数字图像的质量分数值小于等于预设的数字图像质量分阈值,则将所述数字图像的级别设置为与小于等于预设的数字图像质量分阈值相应的级别。
  8. 一种数字图像质量分级的装置,其特征在于,所述装置包括:
    获取模块,用于获取数字图像,并从所述数字图像中提取出n个主体区域块;其中,所述n为自然数;
    第一计算模块,用于计算每个所述主体区域块的面积与所述数字图像的总面积的第一比例值、背景区域块的面积与所述数字图像的总面积的第二比例值、每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值;其中,所述背景区域块为所述数字图像中提取出n个所述主体区域块后剩余的区域块;
    第二计算模块,用于根据所述第一比例值、所述第二比例值和所述归一化距离值,利用预设的数字图像质量分数转换关系,计算得到所述数字图像的质量分数值;其中,预设的数字图像质量分数转换关系是与所有所述主体区域块的第一比例值之和、所述第二比例值和所有所述主体区域块的归一化距离值之和成正比例关系的函数;分级模块,用于根据所述数字图像的质量分数值和预设的数字图像质量分阈值,对所述数字图像的质量进行分级。
  9. 如权利要求8所述的装置,其特征在于,预设的数字图像质量分数转换关系S为:
    Figure PCTCN2015072114-appb-100004
    其中,所述Sfi表示第i个所述主体区域块的面积与所述数字图像的总面积的第一比例值,所述Sb表示所述背景区域块的面积与所述数字图像的总面积的第二比例值,所述n表示从所述数字图像中提取出的所述主体区域块的个数,所述Ri表示第i个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值。
  10. 如权利要求8所述的装置,其特征在于,所述第一计算模块包括:
    第一计算单元,用于计算每个所述主体区域块中每个像素点与所述数字图像的中心像素点之间的平方根距离值;
    求和单元,用于将每个所述主体区域块中所有像素点的平方根距离值求和,得到每个所述主体区域块中所有像素点的平方根距离总值;
    第二计算单元,用于根据每个所述主体区域块中所有像素点的平方根距离总值,利用预设的距离归一化转换关系,计算得到每个所述主体区域块中所有像素点距离所述数字图像的中心像素点的归一化距离值,其中,预设的距离归一化转换关系是根据选择的归一化参数值对每个所述主体区域块中所有像素点的平方根距离总值进行归一化处理。
  11. 如权利要求10所述的装置,其特征在于,预设的距离归一化转换关系Ri为:
    Figure PCTCN2015072114-appb-100005
    其中,所述ri表示第i个所述主体区域块中所有像素点的平方根距离总值,所述
    Figure PCTCN2015072114-appb-100006
    表示选择的归一化参数值,所述w表示所述数字图像的宽度,所述h表示所述数字图像的高度。
  12. 如权利要求8所述的装置,其特征在于,所述第二计算模块包括:
    获取单元,用于获取最优理论质量分数值;
    处理单元,用于将计算得到的所述数字图像的质量分数值与所述最优理论质量分数值的商,作为所述数字图像的归一化质量分数值;
    相应地,所述分级模块包括:
    分级单元,用于根据所述数字图像的归一化质量分数值和预设的归一化数字图像质量分阈值,对所述数字图像的质量进行分级。
  13. 如权利要求12所述的装置,其特征在于,所述获取单元包括:
    选择子单元,用于根据所述主体区域块的个数与所述数字图像的质量关系,所述第一比例值、所述第二比例值和所述归一化距离值之间的关系,选择所述n、所述Sfi、所述Sb、所述Ri的取值;
    处理子单元,用于将选择的所述n、所述Sfi、所述Sb、所述Ri的取值代入预设的数字图像质量分数转换关系S,计算得到的最大值,作为所述最优 理论质量分数值。
  14. 如权利要求8-13任一权利要求所述的装置,其特征在于,所述分级模块包括:
    比较单元,用于将所述数字图像的质量分数值和预设的数字图像质量分阈值进行比较;
    第一分级单元,用于如果所述数字图像的质量分数值大于预设的数字图像质量分阈值,则将所述数字图像的级别设置为与大于预设的数字图像质量分阈值相应的级别;
    第二分级单元,用于如果所述数字图像的质量分数值小于等于预设的数字图像质量分阈值,则将所述数字图像的级别设置为与小于等于预设的数字图像质量分阈值相应的级别。
PCT/CN2015/072114 2014-02-11 2015-02-02 一种数字图像质量分级的方法和装置 WO2015120771A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US15/116,828 US10026184B2 (en) 2014-02-11 2015-02-02 Grading method and device for digital image quality

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201410048608.1A CN104837007B (zh) 2014-02-11 2014-02-11 一种数字图像质量分级的方法和装置
CN201410048608.1 2014-02-11

Publications (1)

Publication Number Publication Date
WO2015120771A1 true WO2015120771A1 (zh) 2015-08-20

Family

ID=53799583

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2015/072114 WO2015120771A1 (zh) 2014-02-11 2015-02-02 一种数字图像质量分级的方法和装置

Country Status (4)

Country Link
US (1) US10026184B2 (zh)
CN (1) CN104837007B (zh)
HK (1) HK1208976A1 (zh)
WO (1) WO2015120771A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10031087B2 (en) 2016-09-22 2018-07-24 SSAB Enterprises, LLC Methods and systems for the quantitative measurement of internal defects in as-cast steel products
CN110246110B (zh) * 2018-03-01 2023-08-18 腾讯科技(深圳)有限公司 图像评估方法、装置及存储介质
CN109541583B (zh) * 2018-11-15 2020-05-01 众安信息技术服务有限公司 一种前车距离检测方法及系统
CN113938671B (zh) * 2020-07-14 2023-05-23 北京灵汐科技有限公司 图像内容分析方法、装置、电子设备和存储介质
CN113055549A (zh) * 2020-09-23 2021-06-29 视伴科技(北京)有限公司 一种播放预演视频的方法及装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008158790A (ja) * 2006-12-22 2008-07-10 Matsushita Electric Works Ltd 人体検出装置
CN101452181A (zh) * 2007-12-03 2009-06-10 鸿富锦精密工业(深圳)有限公司 电子装置的自动对焦系统及方法
US20110292234A1 (en) * 2010-05-26 2011-12-01 Canon Kabushiki Kaisha Image processing apparatus and image processing method
CN102270303A (zh) * 2011-07-27 2011-12-07 重庆大学 敏感图像的联合检测方法
CN103024165A (zh) * 2012-12-04 2013-04-03 华为终端有限公司 一种自动设置拍摄模式的方法和装置

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6822675B2 (en) * 2001-07-03 2004-11-23 Koninklijke Philips Electronics N.V. Method of measuring digital video quality
AU2003302539A1 (en) * 2002-12-05 2004-06-23 Koninklijke Philips Electronics N.V. A system management scheme for a signal-processing-based decision support system
US20050219362A1 (en) * 2004-03-30 2005-10-06 Cernium, Inc. Quality analysis in imaging
US7940970B2 (en) * 2006-10-25 2011-05-10 Rcadia Medical Imaging, Ltd Method and system for automatic quality control used in computerized analysis of CT angiography
JP5197084B2 (ja) * 2008-03-25 2013-05-15 キヤノン株式会社 画像検査装置
US8000528B2 (en) * 2009-12-29 2011-08-16 Konica Minolta Systems Laboratory, Inc. Method and apparatus for authenticating printed documents using multi-level image comparison based on document characteristics
CN101945287B (zh) * 2010-10-14 2012-11-21 浙江宇视科技有限公司 一种roi编码方法及其系统
CN103020947B (zh) * 2011-09-23 2016-04-06 阿里巴巴集团控股有限公司 一种图像的质量分析方法及装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008158790A (ja) * 2006-12-22 2008-07-10 Matsushita Electric Works Ltd 人体検出装置
CN101452181A (zh) * 2007-12-03 2009-06-10 鸿富锦精密工业(深圳)有限公司 电子装置的自动对焦系统及方法
US20110292234A1 (en) * 2010-05-26 2011-12-01 Canon Kabushiki Kaisha Image processing apparatus and image processing method
CN102270303A (zh) * 2011-07-27 2011-12-07 重庆大学 敏感图像的联合检测方法
CN103024165A (zh) * 2012-12-04 2013-04-03 华为终端有限公司 一种自动设置拍摄模式的方法和装置

Also Published As

Publication number Publication date
US20170178339A1 (en) 2017-06-22
US10026184B2 (en) 2018-07-17
CN104837007B (zh) 2018-06-05
HK1208976A1 (zh) 2016-03-18
CN104837007A (zh) 2015-08-12

Similar Documents

Publication Publication Date Title
US10803554B2 (en) Image processing method and device
US11222399B2 (en) Image cropping suggestion using multiple saliency maps
US10453204B2 (en) Image alignment for burst mode images
WO2015120771A1 (zh) 一种数字图像质量分级的方法和装置
US9454712B2 (en) Saliency map computation
US9665962B2 (en) Image distractor detection and processng
US9299004B2 (en) Image foreground detection
WO2016124103A1 (zh) 一种图片检测方法及设备
US8837867B2 (en) Method and system to detect and select best photographs
US9406110B2 (en) Cropping boundary simplicity
US10019823B2 (en) Combined composition and change-based models for image cropping
US20150117783A1 (en) Iterative saliency map estimation
WO2017197959A1 (zh) 一种图片处理方法、装置及设备
CN104572735B (zh) 一种图片标注词推荐方法及装置
US9245347B2 (en) Image Cropping suggestion
WO2021164550A1 (zh) 图像分类方法及装置
US9606975B2 (en) Apparatus and method for automatically generating visual annotation based on visual language
WO2012101697A1 (ja) 画像管理装置、画像管理方法、プログラム、記録媒体、集積回路
CN107885787B (zh) 基于谱嵌入的多视角特征融合的图像检索方法
US11669566B2 (en) Multi-resolution color-based image search
WO2015120772A1 (zh) 一种计算商品图像牛皮癣分值的方法和装置
US9443134B2 (en) Propagating object selection across multiple images
US20160239944A1 (en) Image Resolution Enhancement Based on Data from Related Images
Yin et al. Crowdsourced learning to photograph via mobile devices
CN106469437B (zh) 图像处理方法和图像处理装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 15749115

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 15116828

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 15749115

Country of ref document: EP

Kind code of ref document: A1