CN104574363B - A kind of full reference image quality appraisement method for considering gradient direction difference - Google Patents
A kind of full reference image quality appraisement method for considering gradient direction difference Download PDFInfo
- Publication number
- CN104574363B CN104574363B CN201410766714.3A CN201410766714A CN104574363B CN 104574363 B CN104574363 B CN 104574363B CN 201410766714 A CN201410766714 A CN 201410766714A CN 104574363 B CN104574363 B CN 104574363B
- Authority
- CN
- China
- Prior art keywords
- mrow
- mtd
- msub
- mtr
- msubsup
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of full reference image quality appraisement method for considering gradient direction difference, this method carries out gray processing processing to reference picture and distorted image first and obtains luminance component, casts out colour information；The average LPF of 2 dimensions is carried out to luminance component, the space down-sampling of wide high all progress 1/2 obtains the correspondence image that wide height is original image size half；Gradient intensity of the reference picture after down-sampling with each location of pixels of distorted image on level, vertical, leading diagonal, minor diagonal altogether four direction is calculated using Prewitt gradient operators；Reference picture and gradient intensity difference of the distorted image on four direction on each location of pixels are calculated, and counts the maximum for obtaining gradient intensity difference on each location of pixels four direction；Finally overall image quality is evaluated using the statistical average of all pixels position gradient intensity difference maximum.
Description
Technical field
The present invention relates to the technical field of image quality evaluation, more particularly to a kind of full ginseng for considering gradient direction difference
Examine image quality evaluating method.
Background technology
Image quality evaluation has important application in many image processing fields.Current image quality evaluating method is main
It is divided into subjective assessment and the class of objective evaluation two.Subjective assessment is in certain environment and appointed condition by lineup's class observer
Under, by carrying out subjective scoring to distorted image, the quality index of image is finally obtained by the method for statistics again.Due to most
The final recipient of number image processing system is people, therefore subjective assessment is the most accurate；But subjective evaluation method flow is complicated,
Take time and effort, picture quality can not most importantly be carried out evaluating and monitoring in real time, be difficult to apply to image procossing system
Parameter adjustment is instructed in system.Conversely, objective evaluation extracts characteristics of image and obtains quality index by computation model first, can
Automatic quality evaluation is realized, and is readily applied in real time image processing system.Method for evaluating objective quality is according to reference to figure
The available or utilization power of picture can be generally divided into three classes, i.e., full reference, reduction reference and reference-free quality evaluation.Refer to matter entirely
Amount, which is evaluated, is directed to the available situation of reference picture full detail；Reduce reference mass evaluation and there was only partial information for reference picture
Or the available situation of feature；And reference-free quality evaluation is for reference picture disabled situation completely, such as video communication system
The quality-monitoring of middle receiving terminal.
It is still at most mean square error (MSE) or Y-PSNR to be applied at present in full reference image quality appraisement
(PSNR).MSE or PSNR is averaged by the brightness of statistical-reference image and distorted image average on each location of pixels
Error, calculates simple, and with clear and definite physical significance, but do not account for sense of the human visual system (HVS) to picture quality
Know characteristic, it is not quite identical with human eye subjective assessment.There is scholar to be modeled HVS characteristics afterwards, trial is anti-with mathematical modeling
Perception characteristics of the HVS to image is reflected, and frequency domain decomposition is carried out to image, overall evaluation quality is obtained by measuring the collimation error.
This kind of quality evaluating method effect based on HVS is preferable, but due to complete not enough to the research of HVS complete mechanisms at present, it is this kind of
Also there are many restrictions in method, and computation complexity is generally higher, limits its practical application.
In order to reach preferable balance between quality evaluation performance and computation complexity, there is scholar to attempt by extracting human eye
Sensitive image low-level feature, and by the difference degree of comparison reference image and the low-level feature of distorted image come evaluation image
Quality.It is also the important evidence that human eye understands picture material because border or gradient information are the important low-level features of image, because
This feature frequently as image quality evaluation.But current evaluation algorithms merely with the strength information of gradient, seldom consider mostly
The difference of Gradient direction information, and distortion can make the intensity of gradient and direction change simultaneously.And the present invention can be well
The problem of solving above.
The content of the invention
Present invention aims at provide a kind of full reference image quality appraisement method for considering gradient direction difference, this method
When carrying out image quality evaluation, the difference of gradient intensity information is not only make use of, and efficiently utilize gradient direction
The difference of information, therefore performance is evaluated with better quality, and also this method calculates simple, and complexity is very low, is adapted to large quantities of
The automated graphics quality evaluation of amount.
Method flow：
Step 1：Input reference picture and distorted image to be evaluated；
Step 2：Reference picture and distorted image are subjected to gray processing processing, only retains monochrome information and carries out follow-up image
Quality analysis, casts out colour information；
Step 3：The average LPF of 2 dimensions, the average low pass filtered of 2 dimensions are carried out to the luminance component of reference picture and distorted image
Ripple device is defined as follows：
To the space down-sampling of the picture traverse after LPF and height all progress 1/2, it is original image to obtain wide height
The corresponding reference image R of size half and corresponding distorted image D；
Step 4：Reference image R after the down-sampling obtained to step 3 uses Prewitt gradient operators with distorted image D
Calculate gradient intensity of each location of pixels on level, vertical, leading diagonal, minor diagonal altogether four direction, Prewitt ladders
Convolution kernel of the degree operator on level, vertical, leading diagonal, minor diagonal four direction is defined as follows：
If the gradient intensity figure for the four direction that reference picture is obtained with distorted image after Prewitt gradient operator convolution
It is expressed as Rh、Rv、Rd、RsAnd Dh、Dv、Dd、Ds, that is, it is defined as below；
WhereinRepresent two-dimensional convolution operation；
Step 5：Calculate reference picture and the gradient disparities of each location of pixels of distorted image respectively on four direction, meter
Calculate formula as follows：
WhereinWithRepresent reference image R with distorted image D at certain respectively
Location of pixels is the gradient intensity of (i, j) on individual direction,Then represent that reference image R exists with distorted image D
Location of pixels is the gradient intensity difference of (i, j) on some specific direction；
Step 6：Statistical-reference image R and each location of pixels of distorted image D gradient intensity difference on four direction
MaximumFormula is as follows：
Step 7：By the calculation for calculating reference image R and distorted image D all pixels position gradient intensity difference maximum
Art averagely evaluates overall image quality, and calculation formula is as follows：
The step 4 of the method for the invention calculates reference picture using Prewitt gradient operators and existed with distorted image respectively
Level, vertical, the gradient intensity figure of leading diagonal, minor diagonal four direction, gradient during compared to image quality evaluation before
The main improvement of information extraction is：In view of the perhaps complexity on border in true picture, the gradient for being extracted four direction is strong
Spend information, and the gradient of most of quality evaluating method calculated levels and vertical both direction before.
The step 5 of the method for the invention calculate reference picture with each location of pixels of distorted image respectively in level, hang down
Directly, leading diagonal, the gradient intensity difference on minor diagonal four direction, the improvement compared to most of existing models are：It is right
Reference picture has been respectively compared the gradient intensity difference on four direction with distorted image each position pixel, more make use of
Gradient direction information.
The step 6 statistical-reference image R of the method for the invention is with each location of pixels of distorted image D on four direction
The maximum of gradient intensity difference.
The step 7 of the method for the invention is by calculating the arithmetic average of all pixels position gradient intensity difference maximum
To evaluate overall image quality；Improvement compared to most of existing models is：Gradient intensity information is not only make use of, also effectively
The image quality evaluation that Gradient direction information is referred to entirely is make use of, evaluation effect is more preferable.
Beneficial effect：
1st, the present invention is applied to image quality evaluation, not only solves the difference of gradient intensity information, but also pass through system
Count reference picture and distorted image respectively four direction gradient intensity difference maximum, efficiently solve gradient direction
The difference of information, with better image quality evaluation effect.
2nd, the present invention calculates simple, and complexity is very low, is adapted to large batch of automated graphics quality evaluation.
Brief description of the drawings
Fig. 1 is the image quality evaluating method overall flow figure for considering gradient direction difference of the invention.
Embodiment
The invention is described in further detail with reference to Figure of description.
As shown in figure 1, a kind of full reference image quality appraisement method of consideration gradient direction difference of the present invention, this method
Specific implementation step is as follows：
Step 1：Input reference picture and distorted image to be evaluated；
Step 2：Reference picture and distorted image are subjected to gray processing processing, monochrome information is extracted and carries out follow-up image
Quality analysis, casts out colour information；
Step 3：The average LPF of 2 dimensions, the average low pass filtered of 2 dimensions are carried out to the luminance component of reference picture and distorted image
Ripple device is defined as follows：
To the space down-sampling of the picture traverse after LPF and height all progress 1/2, it is original image to obtain wide height
The corresponding reference image R of size half and corresponding distorted image D；
Step 4：Reference image R after the down-sampling obtained to step 3 uses Prewitt gradient operators with distorted image D
Calculate gradient intensity of each location of pixels on level, vertical, leading diagonal, minor diagonal altogether four direction, Prewitt ladders
Convolution kernel of the degree operator on level, vertical, leading diagonal, minor diagonal four direction is defined as follows：
If the gradient intensity figure for the four direction that reference picture is obtained with distorted image after Prewitt gradient operator convolution
It is expressed as Rh、Rv、Rd、RsAnd Dh、Dv、Dd、Ds, then it is defined as follows；
WhereinRepresent two-dimensional convolution operation；
Step 5：Calculate reference picture and the gradient disparities of each location of pixels of distorted image respectively on four direction, meter
Calculate formula as follows：
WhereinWithRepresent reference image R with distorted image D at certain respectively
Location of pixels is the gradient intensity of (i, j) on individual direction,Then represent that reference image R exists with distorted image D
Location of pixels is the gradient intensity difference of (i, j) on some specific direction；
Step 6：Statistical-reference image R and each location of pixels of distorted image D gradient intensity difference on four direction
MaximumFormula is as follows：
Step 7：By the calculation for calculating reference image R and distorted image D all pixels position gradient intensity difference maximum
Art averagely evaluates overall image quality, and calculation formula is as follows：
Claims (5)
1. a kind of full reference image quality appraisement method for considering gradient direction difference, it is characterised in that comprise the following steps：
Step 1：Input reference picture and distorted image；
Step 2：Reference picture and distorted image are subjected to gray processing processing, only retains monochrome information and carries out follow-up picture quality
Analysis, casts out colour information；
Step 3：The average LPF of 2 dimensions, the average low pass filter of 2 dimensions are carried out to the luminance component of reference picture and distorted image
It is defined as follows：
<mrow>
<msub>
<mi>T</mi>
<mrow>
<mi>a</mi>
<mi>v</mi>
<mi>e</mi>
<mi>L</mi>
<mi>o</mi>
<mi>w</mi>
<mi>P</mi>
<mi>a</mi>
<mi>s</mi>
<mi>s</mi>
</mrow>
</msub>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>4</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>4</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>4</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>4</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
To the space down-sampling of the picture traverse after LPF and height all progress 1/2, it is original image size to obtain wide height
The corresponding reference image R of half and corresponding distorted image D；
Step 4：Reference image R after the down-sampling obtained to above-mentioned steps 3 uses Prewitt gradient operators with distorted image D
Calculate gradient intensity of each location of pixels on level, vertical, leading diagonal, minor diagonal altogether four direction, Prewitt ladders
Convolution kernel of the degree operator on level, vertical, leading diagonal, minor diagonal four direction is defined as follows：
<mrow>
<msub>
<mi>g</mi>
<mi>h</mi>
</msub>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>,</mo>
<msub>
<mi>g</mi>
<mi>v</mi>
</msub>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>g</mi>
<mi>d</mi>
</msub>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>,</mo>
<msub>
<mi>g</mi>
<mi>s</mi>
</msub>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
If the gradient intensity figure for the four direction that reference picture is obtained with distorted image after Prewitt gradient operator convolution is distinguished
It is expressed as Rh、Rv、Rd、RsAnd Dh、Dv、Dd、Ds, that is, it is defined as below；
<mrow>
<msub>
<mi>R</mi>
<mi>h</mi>
</msub>
<mo>=</mo>
<mi>R</mi>
<mo>&CircleTimes;</mo>
<msub>
<mi>g</mi>
<mi>h</mi>
</msub>
<mo>,</mo>
<msub>
<mi>R</mi>
<mi>v</mi>
</msub>
<mo>=</mo>
<mi>R</mi>
<mo>&CircleTimes;</mo>
<msub>
<mi>g</mi>
<mi>v</mi>
</msub>
<mo>,</mo>
<msub>
<mi>R</mi>
<mi>d</mi>
</msub>
<mo>=</mo>
<mi>R</mi>
<mo>&CircleTimes;</mo>
<msub>
<mi>g</mi>
<mi>d</mi>
</msub>
<mo>,</mo>
<msub>
<mi>R</mi>
<mi>s</mi>
</msub>
<mo>=</mo>
<mi>R</mi>
<mo>&CircleTimes;</mo>
<msub>
<mi>g</mi>
<mi>s</mi>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>4</mn>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>D</mi>
<mi>h</mi>
</msub>
<mo>=</mo>
<mi>D</mi>
<mo>&CircleTimes;</mo>
<msub>
<mi>g</mi>
<mi>h</mi>
</msub>
<mo>,</mo>
<msub>
<mi>D</mi>
<mi>v</mi>
</msub>
<mo>=</mo>
<mi>D</mi>
<mo>&CircleTimes;</mo>
<msub>
<mi>g</mi>
<mi>v</mi>
</msub>
<mo>,</mo>
<msub>
<mi>D</mi>
<mi>d</mi>
</msub>
<mo>=</mo>
<mi>D</mi>
<mo>&CircleTimes;</mo>
<msub>
<mi>g</mi>
<mi>d</mi>
</msub>
<mo>,</mo>
<msub>
<mi>D</mi>
<mi>s</mi>
</msub>
<mo>=</mo>
<mi>D</mi>
<mo>&CircleTimes;</mo>
<msub>
<mi>g</mi>
<mi>s</mi>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>5</mn>
<mo>)</mo>
</mrow>
</mrow>
WhereinRepresent two-dimensional convolution operation；
Step 5：Reference picture and the gradient disparities of each location of pixels of distorted image respectively on four direction are calculated, calculate public
Formula is as follows：
<mrow>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>S</mi>
<mi>h</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>R</mi>
<mi>h</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>D</mi>
<mi>h</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>S</mi>
<mi>v</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>R</mi>
<mi>v</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>D</mi>
<mi>v</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>S</mi>
<mi>d</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>R</mi>
<mi>d</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>D</mi>
<mi>d</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>S</mi>
<mi>s</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>R</mi>
<mi>s</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>D</mi>
<mi>s</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>6</mn>
<mo>)</mo>
</mrow>
</mrow>
WhereinWithRepresent reference image R with distorted image D in some direction respectively
Upper location of pixels is the gradient intensity of (i, j),Then represent that reference image R is special at some with distorted image D
Determine the gradient intensity difference that location of pixels on direction is (i, j)；
Step 6：Statistical-reference image R and the maximum of each location of pixels of distorted image D gradient intensity difference on four direction
ValueFormula is as follows：
<mrow>
<msubsup>
<mi>S</mi>
<mi>max</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>=</mo>
<munder>
<mrow>
<mi>m</mi>
<mi>a</mi>
<mi>x</mi>
</mrow>
<mrow>
<mi>k</mi>
<mo>&Element;</mo>
<mo>{</mo>
<mi>h</mi>
<mo>,</mo>
<mi>v</mi>
<mo>,</mo>
<mi>d</mi>
<mo>,</mo>
<mi>s</mi>
<mo>}</mo>
</mrow>
</munder>
<msubsup>
<mi>S</mi>
<mi>k</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
</mrow>
Step 7：Put down by the arithmetic for calculating reference image R and distorted image D all pixels position gradient intensity difference maximum
Overall image quality is evaluated, calculation formula is as follows：
<mrow>
<mi>S</mi>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mi>M</mi>
<mo>&times;</mo>
<mi>N</mi>
</mrow>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<msubsup>
<mi>S</mi>
<mi>max</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</msubsup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>8</mn>
<mo>)</mo>
</mrow>
<mo>.</mo>
</mrow>
2. a kind of full reference image quality appraisement method for considering gradient direction difference according to claim 1, its feature
Be, the step 4 of methods described using Prewitt gradient operators calculate respectively reference picture and distorted image level, it is vertical,
The gradient intensity figure of leading diagonal, minor diagonal four direction, what gradient information during compared to image quality evaluation before was extracted
Mainly improving is：In view of the perhaps complexity on border in true picture, the gradient intensity information of four direction is extracted, and it
The gradient of preceding most of quality evaluating method calculated levels and vertical both direction.
3. a kind of full reference image quality appraisement method for considering gradient direction difference according to claim 1, its feature
It is, the step 5 calculating reference picture of methods described is diagonal in level, vertical, master respectively with each location of pixels of distorted image
Gradient intensity difference on line, minor diagonal four direction, the improvement compared to most of existing models is：To reference picture with
Distorted image each position pixel has been respectively compared the gradient intensity difference on four direction, more make use of gradient direction to believe
Breath.
4. a kind of full reference image quality appraisement method for considering gradient direction difference according to claim 1, its feature
It is, the step 7 of methods described is whole to evaluate by the arithmetic average for calculating all pixels position gradient intensity difference maximum
Body picture quality；Improvement compared to most of existing models is：Gradient intensity information is not only make use of, ladder is also effectively utilized
The image quality evaluation that degree directional information is referred to entirely, evaluation effect is more preferable.
5. a kind of full reference image quality appraisement method for considering gradient direction difference according to claim 1, its feature
It is, methods described is applied to image quality evaluation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410766714.3A CN104574363B (en) | 2014-12-12 | 2014-12-12 | A kind of full reference image quality appraisement method for considering gradient direction difference |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410766714.3A CN104574363B (en) | 2014-12-12 | 2014-12-12 | A kind of full reference image quality appraisement method for considering gradient direction difference |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104574363A CN104574363A (en) | 2015-04-29 |
| CN104574363B true CN104574363B (en) | 2017-09-29 |
Family
ID=53090335
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410766714.3A Active CN104574363B (en) | 2014-12-12 | 2014-12-12 | A kind of full reference image quality appraisement method for considering gradient direction difference |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104574363B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105809691A (en) * | 2016-03-09 | 2016-07-27 | 华侨大学 | Full-reference screen image quality evaluation method |
| CN107578399B (en) * | 2017-07-25 | 2020-12-15 | 天津大学 | Full-reference image quality evaluation method based on boundary feature segmentation |
| CN108053393A (en) * | 2017-12-08 | 2018-05-18 | 广东工业大学 | A kind of gradient similarity graph image quality evaluation method and device |
| CN113222943B (en) * | 2021-05-18 | 2022-05-03 | 宁波智能装备研究院有限公司 | Image deformation estimation method based on mixed grid transformation model |
| CN115187918B (en) * | 2022-09-14 | 2022-12-13 | 中广核贝谷科技有限公司 | Method and system for identifying moving object in monitoring video stream |
| CN117853484B (en) * | 2024-03-05 | 2024-05-28 | 湖南建工交建宏特科技有限公司 | Intelligent bridge damage monitoring method and system based on vision |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5956418A (en) * | 1996-12-10 | 1999-09-21 | Medsim Ltd. | Method of mosaicing ultrasonic volumes for visual simulation |
| CN101482973A (en) * | 2009-01-21 | 2009-07-15 | 西安交通大学 | Partial reference image quality appraisement method based on early vision |
| CN101489130A (en) * | 2009-01-21 | 2009-07-22 | 西安交通大学 | Complete reference image quality assessment method based on image edge difference statistical characteristic |
| CN101621709A (en) * | 2009-08-10 | 2010-01-06 | 浙江大学 | Method for evaluating objective quality of full-reference image |
| CN101853504A (en) * | 2010-05-07 | 2010-10-06 | 厦门大学 | Image quality evaluating method based on visual character and structural similarity (SSIM) |
| CN102708568A (en) * | 2012-05-11 | 2012-10-03 | 宁波大学 | Stereoscopic image objective quality evaluation method on basis of structural distortion |
-
2014
- 2014-12-12 CN CN201410766714.3A patent/CN104574363B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5956418A (en) * | 1996-12-10 | 1999-09-21 | Medsim Ltd. | Method of mosaicing ultrasonic volumes for visual simulation |
| CN101482973A (en) * | 2009-01-21 | 2009-07-15 | 西安交通大学 | Partial reference image quality appraisement method based on early vision |
| CN101489130A (en) * | 2009-01-21 | 2009-07-22 | 西安交通大学 | Complete reference image quality assessment method based on image edge difference statistical characteristic |
| CN101621709A (en) * | 2009-08-10 | 2010-01-06 | 浙江大学 | Method for evaluating objective quality of full-reference image |
| CN101853504A (en) * | 2010-05-07 | 2010-10-06 | 厦门大学 | Image quality evaluating method based on visual character and structural similarity (SSIM) |
| CN102708568A (en) * | 2012-05-11 | 2012-10-03 | 宁波大学 | Stereoscopic image objective quality evaluation method on basis of structural distortion |
Non-Patent Citations (3)
| Title |
|---|
| The Research of Image Quality Assessment Methods;Xiaonan Cui 等;《2012 International Conference on Solid State Devices and Materials Science》;20120412;485-491 * |
| 基于梯度方向信息的医学图像质量评价方法研究;郭龙,郑剑;《计算机科学》;20121215;第39卷(第12期);278-280 * |
| 基于梯度的结构相似度的图像质量评价方法;杨春玲;《华南理工大学学报（自然科学版）》;20060930;第34卷(第9期);22-25 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104574363A (en) | 2015-04-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104574363B (en) | A kind of full reference image quality appraisement method for considering gradient direction difference | |
| CN102421007B (en) | Image quality evaluating method based on multi-scale structure similarity weighted aggregate | |
| CN100559881C (en) | A kind of method for evaluating video quality based on artificial neural net | |
| CN104079925B (en) | Ultra high-definition video image quality method for objectively evaluating based on vision perception characteristic | |
| CN105825500B (en) | A kind of evaluation method and device to camera image quality | |
| CN105338343B (en) | It is a kind of based on binocular perceive without refer to stereo image quality evaluation method | |
| CN104574381B (en) | A kind of full reference image quality appraisement method based on local binary patterns | |
| CN105160647B (en) | A kind of panchromatic multispectral image fusion method | |
| CN107679503A (en) | A kind of crowd's counting algorithm based on deep learning | |
| EP2202687A2 (en) | Image quality evaluation device and method | |
| CN102222323A (en) | Histogram statistic extension and gradient filtering-based method for enhancing infrared image details | |
| CN103942525A (en) | Real-time face optimal selection method based on video sequence | |
| TW201120814A (en) | Method for determining if an input image is a foggy image, method for determining a foggy level of an input image and cleaning method for foggy images | |
| CN104318545B (en) | A kind of quality evaluating method for greasy weather polarization image | |
| CN109978854A (en) | A kind of screen content image quality measure method based on edge and structure feature | |
| CN101976444A (en) | Pixel type based objective assessment method of image quality by utilizing structural similarity | |
| CN105049838A (en) | Objective evaluation method for compressing stereoscopic video quality | |
| CN105678218A (en) | Moving object classification method | |
| CN104282019B (en) | Based on the blind image quality evaluating method that natural scene statistics and perceived quality are propagated | |
| CN106504230A (en) | Complete based on phase equalization refers to color/graphics image quality measure method | |
| CN106934770A (en) | A kind of method and apparatus for evaluating haze image defog effect | |
| CN107371015A (en) | One kind is without with reference to contrast modified-image quality evaluating method | |
| CN109257592B (en) | Stereoscopic video quality objective evaluation method based on deep learning | |
| CN104361571B (en) | Infrared and low-light image fusion method based on marginal information and support degree transformation | |
| CN106412571A (en) | Video quality evaluation method based on gradient similarity standard deviation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |