CN113344843A

CN113344843A - Image quality evaluation method, device and system

Info

Publication number: CN113344843A
Application number: CN202110383595.3A
Authority: CN
Inventors: 梁彦君
Original assignee: Quarkdata Software Co ltd
Current assignee: Quarkdata Software Co ltd
Priority date: 2021-04-09
Filing date: 2021-04-09
Publication date: 2021-09-03
Anticipated expiration: 2041-04-09
Also published as: CN113344843B

Abstract

The invention discloses an image quality evaluation method, device and system. The method comprises the following steps: determining the confirmation weight of each selected evaluation dimension according to the image type and at least one image attribute of the image to be evaluated; comparing the image to be evaluated with the selected comparison image to obtain evaluation results of the image to be evaluated and the comparison image in each evaluation dimension; and determining the comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension. The automatic portrait quality evaluation can be realized, the evaluation accuracy and the working efficiency are greatly improved, the evaluation requirements of different image types and different customer requirements can be met, and the universality is higher.

Description

Image quality evaluation method, device and system

Technical Field

The invention relates to the technical field of computer vision, in particular to an image quality evaluation method, device and system.

Background

In the process of acquiring, processing, transmitting and recording images, because the imaging equipment, the processing method, the transmission medium, the recording equipment and the like are not perfect, and the influence of factors such as object motion, noise pollution, shooting environment and the like causes that the shot images have inevitable distortion and degradation, therefore, the image quality evaluation is carried out on the images, so that the parameters and the flow of links such as imaging, processing, transmitting, recording and the like are adjusted, and the images with higher quality are necessary to be acquired. There are two modes of subjective evaluation and objective evaluation for evaluating image quality.

The subjective evaluation takes a person as an observer to evaluate the image, and strives to truly reflect the visual perception of the person; in this way, an observer generally compares an image to be evaluated with an image captured by a contrast device, and compares a plurality of sets of images captured by a plurality of sets of imaging devices to evaluate the image quality. The manual evaluation mode has low accuracy and poor working efficiency.

The objective evaluation reflects subjective perception of human eyes by means of a mathematical model, and gives evaluation results based on digital calculation, such as: US patent application publication No. US20190080443a1 discloses extracting a reference image by a computer, evaluating the quality of a target image by comparing the reference image with the target image; another example is: chinese patent application publication No. CN108596901A discloses using a data model to evaluate image quality by machine learning; the existing objective evaluation mode collects and makes a benchmark data model through the evaluation of learning experts and machines, the consumption of early-stage preparation work is large, and the benchmark model cannot be adapted to all customer requirements due to different customer requirements.

Disclosure of Invention

In view of the above problems, the present invention has been made to provide an image quality evaluation method, apparatus and system that overcome the above problems or at least partially solve the above problems.

The embodiment of the invention provides an image quality evaluation method, which is characterized by comprising the following steps:

determining the confirmation weight of each selected evaluation dimension according to the image type and at least one image attribute of the image to be evaluated;

comparing the image to be evaluated with the selected comparison image to obtain evaluation results of the image to be evaluated and the comparison image in each evaluation dimension;

and determining the comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension.

In some optional embodiments, the determining the confirmation weight of each selected evaluation dimension according to the image type and at least one image attribute of the image to be evaluated comprises:

determining the initial weight of each selected evaluation dimension according to the image type of the image to be evaluated and the selected first image attribute;

and adjusting the initial weight of each judgment dimension at least once according to at least one second image attribute to obtain the confirmation weight of each judgment dimension.

In some optional embodiments, the image attributes comprise at least one of image scene, image content characteristics, capture device, capture parameters;

the evaluation dimension includes at least one of color, brightness, sharpness, noise, saturation, contrast, sharpness, dynamic range.

In some optional embodiments, comparing the image to be evaluated with the selected comparison image to obtain the evaluation result of the image to be evaluated and the comparison image in each evaluation dimension, includes:

dividing an image to be evaluated into regions, and extracting image characteristics of the divided image regions or region intersections;

comparing the extracted image features with the image features of the corresponding regions or region intersections of the comparison image to obtain comparison results of each corresponding region or region intersection of the image to be evaluated in each evaluation dimension relative to the comparison image;

and obtaining the evaluation results of the image to be evaluated and the comparison image in each evaluation dimension according to the comparison result.

In some optional embodiments, when the comparison image is one image, the determining the comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension includes:

and performing weighted calculation on the evaluation result of each evaluation dimension and the confirmation weight of the corresponding dimension to obtain a comprehensive evaluation result of the image to be evaluated and the comparison image.

In some optional embodiments, when there is more than one comparison image, the determining the comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension includes:

for each comparison image, carrying out weighted calculation on the evaluation result of each evaluation dimension and the confirmation weight of the corresponding dimension to respectively obtain the evaluation results of the image to be evaluated and each comparison image, and determining the mean value of the evaluation results of the image to be evaluated and each comparison image to obtain the comprehensive evaluation result of the image to be evaluated and the comparison images; or

For each comparison image, performing weighted calculation on the evaluation result of each evaluation dimension and the confirmation weight of the corresponding dimension to respectively obtain the evaluation results of the image to be evaluated and each comparison image, and taking the evaluation results as the comprehensive evaluation results of the image to be evaluated and the comparison images; or

Determining the mean value of the evaluation results of the image to be evaluated and each comparison image aiming at each evaluation dimension to obtain the comprehensive evaluation result of each evaluation dimension, and performing weighted calculation on the comprehensive evaluation result of each evaluation dimension and the confirmation weight of the corresponding dimension to obtain the comprehensive evaluation result of the image to be evaluated and the comparison image; or

And determining the mean value of the evaluation results of the image to be evaluated and each comparison image aiming at each evaluation dimension to obtain the comprehensive evaluation result of each evaluation dimension as the comprehensive evaluation result of the image to be evaluated and the comparison image.

In some optional embodiments, the method further comprises:

and adding evaluation description information based on the comprehensive evaluation result of the evaluation image and the comparison image, and providing the comprehensive evaluation result and the evaluation description information for a user.

An embodiment of the present invention further provides an image quality evaluation device, including:

the weight determining module is used for determining the confirmation weight of each selected evaluation dimension according to the image type and at least one image attribute of the image to be evaluated;

the first evaluation module is used for comparing the image to be evaluated with the selected comparison image to obtain evaluation results of the image to be evaluated and the comparison image in each evaluation dimension;

and the second evaluation module is used for determining the comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension.

In some optional embodiments, the weight determining module is specifically configured to:

In some optional embodiments, the first evaluation module is specifically configured to:

when the contrast image is one:

carrying out weighted calculation on the evaluation result of each evaluation dimension and the confirmation weight of the corresponding dimension to obtain a comprehensive evaluation result of the image to be evaluated and the comparison image;

when the contrast image is more than one:

In some optional embodiments, the apparatus further comprises:

and the result output module is used for adding evaluation description information based on the comprehensive evaluation result of the evaluation image and the comparison image and providing the comprehensive evaluation result and the evaluation description information for a user.

The embodiment of the invention also provides an image quality evaluation method, which is characterized by comprising the following steps:

determining the initial weight of each selected evaluation dimension according to the image type and the image scene of the image to be evaluated;

adjusting the initial weight of each evaluation dimension according to the image content of the image to be evaluated to obtain the confirmation weight of each evaluation dimension;

In some optional embodiments, before obtaining the validation weight of each evaluation dimension, the method further includes:

and adjusting the initial weight after the initial weight is adjusted for at least one time according to the type of a camera of shooting equipment used for shooting the image to be evaluated and/or at least one item of the gender and the age of the person in the image.

the weight determining module is used for determining the initial weight of each selected evaluation dimension according to the image type and the image scene of the image to be evaluated; adjusting the initial weight of each evaluation dimension according to the image content of the image to be evaluated to obtain the confirmation weight of each evaluation dimension;

The embodiment of the present invention further provides an image quality evaluation system, including: the image quality evaluation device comprises a testing machine, at least one comparison machine and the image quality evaluation device;

the image quality evaluation device is used for acquiring an image to be evaluated from the testing machine and a comparison image from the comparison machine; and evaluating the obtained image to be evaluated and the comparison image to obtain a comprehensive evaluation result of the image to be evaluated and the comparison image.

The embodiment of the invention also provides a computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions are executed by a processor to realize the image quality evaluation method.

An embodiment of the present invention further provides an electronic device, including: the image quality evaluation method comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor executes the program to realize the image quality evaluation method.

The technical scheme provided by the embodiment of the invention has the beneficial effects that at least:

determining the confirmation weight of each evaluation dimension according to the image type and the image attribute of the image to be evaluated, comparing the image to be evaluated with the comparison image in each evaluation dimension, and obtaining the comprehensive evaluation result of the image to be evaluated and the comparison image based on the determined confirmation weight and the evaluation result of each evaluation dimension; from weight determination to image comparison and evaluation result obtaining, each stage can be automatically realized, so that human resource investment is greatly reduced, and the working efficiency and accuracy of image evaluation are greatly improved; according to the scheme, a data model is manufactured without a large amount of learning sample data, the workload of early preparation work is greatly reduced, different evaluation dimensions and weights can be selected according to different requirements for evaluation, the evaluation dimensions and weights can be selected and determined based on image types and image attributes, namely different evaluation dimensions can be used for different images to be evaluated, and different confirmation weights can be used for each evaluation dimension, so that the evaluation requirements of different types and different customer requirements are met, and the evaluation method can adapt to various different evaluation requirements.

Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.

Drawings

The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:

FIG. 1 is a flowchart illustrating an image quality evaluation method according to an embodiment of the present invention;

FIG. 2 is a diagram illustrating an example of an image storage format according to an embodiment of the present invention;

FIG. 3 is a diagram illustrating an example of image storage naming codes according to an embodiment of the present invention;

FIG. 4 is a flowchart illustrating an image quality evaluation method according to a second embodiment of the present invention;

FIG. 5 is a flowchart illustrating a first weight verification according to a second embodiment of the present invention;

FIG. 6 is a flowchart illustrating a second weight determination according to a second embodiment of the present invention;

FIG. 7 is a schematic view of a contrast focusing principle according to a second embodiment of the present invention;

FIG. 8 is a diagram illustrating an example of an RGB color model according to a second embodiment of the invention;

FIG. 9 is a diagram illustrating an example of images with different contrast ratios according to a second embodiment of the present invention;

FIG. 10 is a diagram illustrating an example of an image histogram according to a second embodiment of the present invention;

fig. 11 is a schematic structural diagram of an image quality evaluation device according to an embodiment of the present invention;

FIG. 12 is a schematic structural diagram of another image quality evaluation apparatus according to an embodiment of the present invention;

fig. 13 is a schematic structural diagram of an image quality evaluation system according to an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

In order to solve the problems of low accuracy and poor working efficiency of manual image quality evaluation in the prior art, large evaluation workload by means of a mathematical model, incapability of meeting requirements of different customers and the like, the embodiment of the invention provides an image quality evaluation method.

Example one

The embodiment of the invention provides an image quality evaluation method, the flow of which is shown in figure 1, and the method comprises the following steps:

step S101: and determining the confirmation weight of each selected evaluation dimension according to the image type and at least one image attribute of the image to be evaluated.

In this step, the image type is identified, which evaluation dimensions to use for evaluation can be determined for different image types, and different weights are used in each evaluation dimension for different image types in the comprehensive evaluation. In addition, one or more times of weight adjustment can be carried out according to the image attributes, so that the confirmation weight for comprehensive evaluation is determined finally.

The image type may include at least one of an original image type, an algorithm image type, and the like; the image attribute comprises at least one of an image scene, image content characteristics, shooting equipment and shooting parameters; the evaluation dimension includes at least one of color, brightness, sharpness, noise, saturation, contrast, sharpness, and dynamic range, and is certainly not limited to these image types, image attributes, and evaluation dimensions, and may be increased or decreased as desired in a particular application. That is, the dimension of image evaluation may be increased according to the requirement, and the adjustment times of weight confirmation of evaluation may also be increased according to the requirement, such as: the confirmation is performed one or more times according to the scene division or according to the person's gender, according to age, or the like.

Specifically, the initial weight of each selected evaluation dimension can be determined according to the image type of the image to be evaluated and the selected first image attribute; and adjusting the initial weight of each judgment dimension at least once according to at least one second image attribute to obtain the confirmation weight of each judgment dimension. For example: the first image attribute may select an image scene and the second image attribute may select image content.

Step S102: and comparing the image to be evaluated with the selected comparison image to obtain the evaluation result of the image to be evaluated and the comparison image in each evaluation dimension.

In the step, the image to be evaluated shot by the testing machine is compared with the comparison image shot by one comparison machine or a plurality of comparison machines. In comparison, the image to be evaluated may be partitioned, and the partitioned areas may be compared, or the features at the intersections of the areas may be compared. Specifically, the method comprises the steps of performing region division on an image to be evaluated, and performing image feature extraction on a divided image region or a region intersection; comparing the extracted image features with the image features of the corresponding regions or region intersections of the comparison image to obtain comparison results of each corresponding region or region intersection of the image to be evaluated in each evaluation dimension relative to the comparison image; and obtaining the evaluation results of the image to be evaluated and the comparison image in each evaluation dimension according to the comparison result.

When image comparison is carried out, different modes can be selected for comparison according to different judging dimensions. For example, comparing the color or saturation, an RGB color model may be used for comparison; comparing the brightness, wherein a histogram mode can be adopted; contrast histograms may be used to compare the sharpness. The Noise can be aligned by calculating the Signal to Noise Ratio (SNR) in the images.

Step S103: and determining the comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension.

After the evaluation results of the image to be evaluated and the comparison image in each evaluation dimension are obtained, comprehensive processing can be performed according to the confirmation weight of each evaluation dimension to obtain a comprehensive evaluation result. Generally, in order to obtain a better and more accurate evaluation result, images shot by a plurality of comparison machines are selected as comparison images to be compared with an evaluation image to be evaluated, so that different modes can be adopted when the evaluation results of each evaluation dimension are comprehensively sorted.

And when the comparison image is one image, performing weighted calculation on the evaluation result of each evaluation dimension and the confirmation weight of the corresponding dimension to obtain a comprehensive evaluation result of the image to be evaluated and the comparison image. For the situation of a contrast image, an evaluation report can be formed based on the comprehensive evaluation result obtained after the weighted calculation; optionally, the evaluation result of each evaluation dimension and the confirmation weight of the corresponding dimension may be directly used as a comprehensive evaluation result without performing weighted calculation, so as to form an evaluation report.

When more than one comparison image is available, the evaluation results of each comparison image can be sorted firstly, and then the evaluation results of each comparison image are integrated to obtain an integrated evaluation result and form an evaluation report; or the evaluation results of all evaluation dimensions can be sorted firstly, and the evaluation results of all evaluation dimensions are integrated to obtain an integrated evaluation result and form an evaluation report; of course, the evaluation results of each evaluation dimension of each comparative image and the corresponding confirmation weight can be directly used as the comprehensive evaluation result without comprehensive treatment, so that an evaluation report is formed. Obtaining the comprehensive evaluation result of the image to be evaluated and the comparison image by adopting any one or more of the following modes:

the first method is as follows:

and for each comparison image, carrying out weighted calculation on the evaluation result of each evaluation dimension and the confirmation weight of the corresponding evaluation dimension to respectively obtain the evaluation results of the image to be evaluated and each comparison image, determining the mean value of the evaluation results of the image to be evaluated and each comparison image, and obtaining the comprehensive evaluation result of the image to be evaluated and the comparison images.

The method firstly arranges the evaluation results of the image to be evaluated and each comparison image, for example, when there are two comparison images 1 and 2, the evaluation results of each evaluation dimension of the image to be evaluated and the comparison image 1 and the confirmation weight of the corresponding evaluation dimension can be weighted to calculate to obtain the evaluation result 1 of the image to be evaluated and the comparison image 1, similarly, the evaluation results of each evaluation dimension of the image to be evaluated and the comparison image 2 and the confirmation weight of the corresponding evaluation dimension can be weighted to calculate to obtain the evaluation result 2 of the image to be evaluated and the comparison image 2, and then the mean value of the evaluation result 1 and the evaluation result 2 is used as the comprehensive evaluation result of the image to be evaluated and the comparison image.

The second method comprises the following steps:

and for each comparison image, performing weighted calculation on the evaluation result of each evaluation dimension and the confirmation weight of the corresponding dimension to respectively obtain the evaluation results of the image to be evaluated and each comparison image, and taking the evaluation results as the comprehensive evaluation results of the image to be evaluated and the comparison images.

The method firstly arranges the evaluation results of the image to be evaluated and each comparison image, for example, when there are two comparison images 1 and 2, the evaluation results of each evaluation dimension of the image to be evaluated and the comparison image 1 and the confirmation weight of the corresponding evaluation dimension can be weighted to calculate to obtain the evaluation result 1 of the image to be evaluated and the comparison image 1, similarly, the evaluation results of each evaluation dimension of the image to be evaluated and the comparison image 2 and the confirmation weight of the corresponding evaluation dimension can be weighted to calculate to obtain the evaluation result 2 of the image to be evaluated and the comparison image 2, and then the evaluation result 1 and the evaluation result 2 are used as the comprehensive evaluation result of the image to be evaluated and the comparison image.

The third method comprises the following steps:

and aiming at each evaluation dimension, determining the average value of the evaluation results of the image to be evaluated and each comparison image to obtain the comprehensive evaluation result of each evaluation dimension, and performing weighted calculation on the comprehensive evaluation result of each evaluation dimension and the confirmation weight of the corresponding dimension to obtain the comprehensive evaluation result of the image to be evaluated and the comparison image.

The method firstly arranges the evaluation results of the image to be evaluated and each comparison image in each evaluation dimension, for example, when two comparison images 1 and 2 are provided, the comparison is respectively carried out in the

evaluation dimensions

1, 2 and 3, the mean value 1 of the evaluation results of the image to be evaluated and the comparison images 1 and 2 in the evaluation dimension 1, the mean value 2 of the evaluation results of the image to be evaluated and the comparison images 1 and 2 in the evaluation dimension 2, the mean value 3 of the evaluation results of the image to be evaluated and the comparison images 1 and 2 in the evaluation dimension 3 can be calculated, and the

mean values

1, 2 and 3 and the confirmation weights of the corresponding evaluation dimensions are weighted to obtain the comprehensive evaluation result of the image to be evaluated and the comparison images.

The method is as follows:

The method firstly arranges the evaluation results of the image to be evaluated and each comparison image in each evaluation dimension, for example, when two comparison images 1 and 2 are provided, the comparison is carried out in the

evaluation dimensions

1, 2 and 3 respectively, the mean value 1 of the evaluation results of the image to be evaluated and the comparison images 1 and 2 in the evaluation dimension 1, the mean value 2 of the evaluation results of the image to be evaluated and the comparison images 1 and 2 in the evaluation dimension 2, the mean value 3 of the evaluation results of the image to be evaluated and the comparison images 1 and 2 in the evaluation dimension 3 can be calculated, and the mean value 1, the mean value 2 and the mean value 3 are used as the comprehensive evaluation results of the image to be evaluated and the comparison images.

In some optional embodiments, the method further comprises:

and adding evaluation description information based on the comprehensive evaluation result of the evaluation image and the comparison image, and providing the comprehensive evaluation result and the evaluation description information for the user. An evaluation report including the comprehensive evaluation result and the evaluation description information can be formed, and the evaluation description information can be similar description information with reddish color, slightly better definition than a comparator, larger noise and the like.

The method of the embodiment can evaluate the image quality of a test machine and a (or more) contrast machines for comparison, and the evaluation dimensions can include a large item of definition, color, brightness, noise and the like, and the evaluation dimensions derived from the large item of definition, color, brightness, noise and the like, including contrast, dynamic range, saturation, sharpness and the like. Taking a testing machine and two comparison machines as examples, the electronic device may store related images in the form shown in fig. 2 and 3, as shown in fig. 2, the images taken by the testing machine, the comparison machine 1 and the comparison machine 2 are stored in folders with folder names of the testing machine, the comparison machine 1 and the comparison machine 2, respectively, the images in each file may be named and numbered according to the scene, as shown in fig. 2, the corresponding images in the folders of the testing machine, the comparison machine 1 and the comparison machine 2 use the same naming number, and in comparison, the images with the same naming number in each file are extracted for comparison. The mode is convenient for extraction and comparison, and certainly, other named storage organization modes can be adopted for storage, and only corresponding relations are needed to be established, and corresponding images can be extracted for comparison.

In the method of this embodiment, according to the image type and the image attribute of the image to be evaluated, the determination weight of each evaluation dimension is determined, and after the image to be evaluated and the comparison image are compared in each evaluation dimension, the comprehensive evaluation result of the image to be evaluated and the comparison image is obtained based on the determined determination weight and the evaluation result of each evaluation dimension; from weight determination to image comparison and evaluation result obtaining, each stage can be automatically realized, so that human resource investment is greatly reduced, and the working efficiency and accuracy of image evaluation are greatly improved; according to the scheme, a data model is manufactured without a large amount of learning sample data, the workload of early preparation work is greatly reduced, different evaluation dimensions and weights can be selected according to different requirements for evaluation, the evaluation dimensions and weights can be selected and determined based on image types and image attributes, namely different evaluation dimensions can be used for different images to be evaluated, and different confirmation weights can be used for each evaluation dimension, so that the evaluation requirements of different types and different customer requirements are met, and the evaluation method can adapt to various different evaluation requirements.

Example two

The second embodiment of the present invention provides a specific implementation process of the image quality evaluation method, the flow of which is shown in fig. 4, and the method includes the following steps:

step S201: and determining the initial weight of each selected evaluation dimension according to the image type and the image scene of the image to be evaluated.

In this embodiment, the example of performing weight determination twice is described, and the practical application is not limited to twice.

Referring to fig. 5, the image type is determined to be an original image or an algorithm image. When the evaluation is carried out, the image is determined to be of an algorithm class (beauty/HDR/night view.) or an original image class, so that the weight of each subsequent evaluation dimension can be confirmed. In this embodiment, an original image is used for explanation, and the evaluation methods of the algorithm images are similar, except that the weight settings of the evaluation dimensions may be different.

Then, the first weight confirmation is performed, at this time, it is required to confirm an image scene, for example, a scene in the image is a day or night scene, and a person or no person is in the image, so as to confirm that the obtained image scene may be one of a person in the day, a person in the night or a person in the night. And then determining the initial weight of each judgment dimension according to the image scene, namely the proportion of each judgment dimension in judgment.

The first weight confirmation is performed according to the day or the night, with or without people, which is to perform the large direction confirmation according to the large scene difference: for example, a person scene in a night scene can emphasize the brightness weight of the face of the person during judgment; during evaluation, the weight of the overall color and the overall brightness of the picture is emphasized by an unmanned scene in the daytime; this weight may be a fixed weight confirmation.

Step S202: and adjusting the initial weight of each evaluation dimension according to the image content of the image to be evaluated to obtain the confirmation weight of each evaluation dimension.

After the first weight confirmation is performed, the initial weight of each evaluation dimension is obtained, and the initial weight can be adjusted once or for many times. Taking the first adjustment as an example, referring to fig. 6, the second weight confirmation is performed, and in this case, it is necessary to recognize the content in the image and perform the weight adjustment according to the image content. For example, if the image content is recognized as a building/landscape, an animal/person, an object/text, or others (such as artistic elements, ink and wash paintings), the second image confirmation is performed according to the recognized image content, that is, the weight adjustment is performed after the recognition is completed, and when the image content is different, the weights of the judgment dimensions may be different. Of course the image content is not limited to the ones listed. The secondary weight confirmation can be correspondingly changed according to the customer requirements or the main direction targets. If the customer wants the saturation of the landscape color to be more gorgeous, the evaluation weight of the landscape color module is increased.

Step S203: and comparing the image to be evaluated with the selected comparison image to obtain the evaluation result of the image to be evaluated and the comparison image in each evaluation dimension.

In this embodiment, when the images are compared to obtain the evaluation result, the evaluation result is described as an example of the comparison score, and in practical application, the comparison score is not limited, and different evaluation parameters may be adopted to output the corresponding evaluation result.

And starting to carry out contrast scoring on the images after the weight of each evaluation dimension is confirmed. The score for each scene contrast may be set to a range of 1-10 or 1-100 equally-spaced values.

Taking color dimension as an example, pixel point RGB information of an image to be evaluated to be compared and a contrast image can be extracted, and the extraction mode can correspondingly distinguish the image to be evaluated from the image to be compared aiming at the manned/unmanned scene: the picture can be equally divided into N (preferably N is larger than or equal to 16) modules aiming at an unmanned scene, RGB information of pixel points is extracted at equally divided cross points, RGB numerical value comparison is carried out, grading of color dimensionality is obtained, and an image overall color evaluation result is obtained comprehensively. Therefore, judgment description of color dimension, such as color cast and the like, can be obtained.

Aiming at the manned scene, the RGB information of the pixel points can be correspondingly extracted from the face of the person, and the RGB information of the pixel points can be extracted from other parts of the image. And carrying out RGB numerical value comparison on the pixel points extracted from the two parts to obtain the score of the color dimension, and comprehensively obtaining the overall color evaluation result of the image. Accordingly, judgment description of color dimension, such as whole color cast or partial color cast, can be obtained.

The contrast of the saturation and the brightness dimension is similar to the color dimension, and the RGB information of the pixel points of the image to be evaluated and the contrast image to be compared is extracted for comparison. Taking the evaluation of the brightness dimension as an example, if all extracted G values of the image to be evaluated are greater than the G value of the contrast image, the brightness of the image to be evaluated is greater than the brightness of the contrast image.

The contrast evaluation of the saturation dimension may refer to or directly introduce the RGB color model for evaluation, the evaluation of the brightness dimension may use an image histogram or a G value in RGB for evaluation, a higher point on the histogram is more points under the brightness, and the contrast evaluation may also use an image histogram.

The contrast evaluation of the definition dimension may be performed by comparing the whole image or selecting a partial region in the image, and the contrast of the definition may be performed by using a contrast histogram, as shown in fig. 7, the contrast focusing basic principle is performed by comparing the eye regions in the image, the leftmost column in fig. 7 may see the black-white boundary position where the focus is the black eyeball and the white eyeball in the eye, the second column on the left is an AF comparison range, the third column is a calculated inverse difference amount within the comparison range, and the rightmost column is an inverse difference amount histogram, as can be seen from fig. 3, the definition of the bottommost one of the 7 images of the eye portions in the left column is the best.

The comparison of the noise dimensions can be evaluated by calculating the signal-to-noise ratio SNR in the image. Generally, a flat area or a dark area in an image is selected as a calculation sample area, and an algorithm formula is required to be introduced in the calculation, so that the SNR value can be calculated by using the existing algorithm.

Step S204: and determining the comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension.

The evaluation results are collated as described in step S103 in example one. Taking a testing machine, a lower limit machine and an upper limit machine as an example, comparing an image to be evaluated shot by the testing machine with a comparison image 1 shot by the lower limit machine to obtain comparison scores of the image to be evaluated and the comparison image 1 in each evaluation dimension, comparing the image to be evaluated shot by the testing machine with a comparison image 2 shot by the lower limit machine to obtain comparison scores of the image to be evaluated and the comparison image 2 in each evaluation dimension, and then performing summarizing and sorting in at least one mode in step S103. When averaging, weighted average may also be performed, for example, the weighted ratio of the contrast scores of the tester and the upper limit machine and the contrast scores of the tester and the lower limit machine is 6: 4.

of course, the comprehensive score is not calculated, and the comparison scores of the image to be evaluated and each comparison image in each evaluation dimension are directly collected together to form an evaluation report together with the confirmation weight of each evaluation dimension. Language descriptions may also be added to the assessment report, such as: the whole picture is reddish, the definition is slightly worse than that of the upper limit machine and slightly better than that of the lower limit machine, the whole brightness is equal to that of the contrast machine, and the whole noise is more, and the like.

In some optional embodiments, before the confirmation weight of each evaluation dimension is obtained in step S202, at least one adjustment may be performed again, and the initial weight after the first adjustment may be adjusted at least once according to a camera type of a shooting device used when shooting the image to be evaluated and/or at least one of a person gender and a person age included in the image, so as to obtain the confirmation weight of each evaluation dimension after the adjustment. Such as: when the weight adjustment is confirmed, the distinction of evaluation lenses, such as a wide-angle lens, a main camera lens, a macro lens, a front camera lens, a long-focus camera lens and the like, can be added, and the evaluation weight is modified correspondingly according to the characteristics of the lenses.

The RGB color model mentioned in the embodiment of the present invention can be illustrated with reference to fig. 8, where the space of the RGB color model is a unit cube, each point in the region in the cube corresponds to a different color, that is, each point may correspond to a different color from the origin, and is represented by a vector from the origin to the point, three coordinate values are ratios of three colors, red (R), green (G), and blue (B), where black is fixed at the origin, and white is fixed at the point (1,1, 1). In digital systems where this unit space is discretized, each component is typically represented by an 8-bit integer, thus requiring a 24-bit representation for each pixel. When the color model is used for judging, the RGB color model can be used, and the judgment can also be carried out based on YUV and HSV color models.

The principle of contrast focusing mentioned in the embodiments of the present invention is briefly described as follows: generally, the sharpest point of an image is also the point with the largest contrast, a camera drives a lens, the focus points are changed along the axis pointing to a shot object, an image is obtained on each focus point, similarly to point-by-point scanning, the image obtained on each focus point is firstly digitized, the digitized image is actually an integer matrix and is transmitted to an image processor, then the contrast value is calculated, the focus point with the largest contrast value is screened out through contrast, the lens is driven, the focus point is placed on the focus point with the largest contrast value, the correct focus point is obtained, whether the focus is focused or not is determined according to the value with the largest contrast value, and focusing is finished. When the method is reflected on the screen of the electronic equipment, the method is a process of 'drawing a bellows' from blurring to clearness and then from blurring to clearness and finally clearness. This determination can achieve very high focusing accuracy, as well as practical use. This focusing technique is called contrast focusing. The contrast focusing process is a simple process of finding the maximum value, and the implementation by a program is a relatively simple matter, and the basic purpose is as follows: focusing is accomplished with a minimum number of samples.

The contrast and histogram referred to in the examples of the present invention are briefly described as follows: a histogram may describe the contrast of an image, which is a measure of the difference in brightness between bright and dark areas in a scene of an image. A broad histogram may reflect a certain image having a higher contrast, whereas a narrower histogram may reflect a certain image having a lower contrast. Such contrast difference may be caused by illumination conditions and other various factors in combination. Images taken in foggy weather have low contrast; images taken under some strong light have a higher contrast. As shown in fig. 9, the left half is an example of an image with low contrast, and the right half is an example of an image with high contrast.

The image histogram is a histogram for representing the brightness distribution in an image, and is given by counting the number of pixels with certain brightness in an image or a certain range of brightness. As shown in fig. 10, the horizontal axis represents luminance values of 0-255, the vertical axis represents the number of pixels corresponding to luminance in the image, the left side of the histogram is pure black, and the right side of the histogram is pure white, and the peak of the histogram shown in fig. 10 is a position to the left in the middle, which indicates that there are many dark gray or dark parts in the picture. .

Based on the same inventive concept, an embodiment of the present invention further provides an image quality evaluation device, which may be disposed in an electronic device, for example: a terminal device, a server, and the like, the configuration of which is shown in fig. 11, and which includes:

and the weight determining module 11 is configured to determine the confirmation weight of each selected evaluation dimension according to the image type and at least one image attribute of the image to be evaluated.

The first evaluation module 12 is configured to compare the image to be evaluated with the selected comparison image to obtain an evaluation result of the image to be evaluated and the comparison image in each evaluation dimension.

And the second evaluation module 13 is configured to determine a comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension.

Optionally, the weight determining module 11 is specifically configured to determine an initial weight of each selected evaluation dimension according to the image type of the image to be evaluated and the selected first image attribute; and adjusting the initial weight of each judgment dimension at least once according to at least one second image attribute to obtain the confirmation weight of each judgment dimension.

Optionally, the first evaluation module 12 is specifically configured to perform region division on an image to be evaluated, and perform image feature extraction on a divided image region or a region intersection; comparing the extracted image features with the image features of the corresponding regions or region intersections of the comparison image to obtain comparison results of each corresponding region or region intersection of the image to be evaluated in each evaluation dimension relative to the comparison image; and obtaining the evaluation results of the image to be evaluated and the comparison image in each evaluation dimension according to the comparison result.

Optionally, the first evaluation module 12 is specifically configured to:

when the contrast image is one:

when the contrast image is more than one:

In some optional embodiments, the apparatus further comprises:

and the result output module 14 is used for adding the evaluation description information based on the comprehensive evaluation result of the image to be evaluated and the comparison image, and providing the comprehensive evaluation result and the evaluation description information for the user.

An embodiment of the present invention further provides another image quality evaluation device, a structure of which is shown in fig. 12, including:

the weight determining module 21 is configured to determine an initial weight of each selected evaluation dimension according to the image type and the image scene of the image to be evaluated; and adjusting the initial weight of each evaluation dimension according to the image content of the image to be evaluated to obtain the confirmation weight of each evaluation dimension.

The first evaluation module 22 is configured to compare the image to be evaluated with the selected comparison image to obtain an evaluation result of the image to be evaluated and the comparison image in each evaluation dimension.

The second evaluation module 23 is configured to determine a comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension.

The weight determining module 21 is specifically configured to adjust the initial weight after the initial weight is adjusted for at least one time according to a camera type of a shooting device used when the image to be evaluated is shot and/or at least one of a person gender and a person age included in the image.

Based on the same inventive concept, an embodiment of the present invention further provides an image quality evaluation system, where the system structure is shown in fig. 13, and the system includes: the image quality evaluation system comprises a testing machine 1, at least one comparison machine 2 and an image quality evaluation device 3;

the image quality evaluation device 3 is used for acquiring an image to be evaluated from the testing machine 1 and acquiring a comparison image from the comparison machine 2; and evaluating the obtained image to be evaluated and the comparison image to obtain a comprehensive evaluation result of the image to be evaluated and the comparison image.

The embodiment of the invention also provides a computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions are executed by the processor to realize the image quality evaluation method.

An embodiment of the present invention further provides an electronic device, including: the image quality evaluation method comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor realizes the image quality evaluation method when executing the computer program.

With regard to the system and apparatus in the above embodiments, the specific manner in which the respective modules perform operations has been described in detail in relation to the embodiments of the method, and will not be elaborated upon here.

In the prior art, subjective image quality evaluation is generally performed artificially and subjectively by using a comparison machine, and multiple groups of pictures of multiple shooting devices need to be manually compared to perform image quality evaluation. When the machine learning mode is adopted for evaluation, a large amount of sample data needs to be collected to manufacture the benchmark data model, the consumption of early preparation work is large, the benchmark model cannot be adapted to all customer requirements, and the application range is narrow. The method can replace the artificial judgment mechanism, quantizes the judgment mode, avoids the inaccuracy of judgment caused by the long time of artificial picture viewing, improves the working efficiency and reduces a large amount of time and labor cost required by artificial picture viewing judgment.

The image quality evaluation method, the system and the device can compare the image effects of a plurality of comparison machines and give the scores for the reference or direct use of users. And a large amount of experts or machines are not needed for standard model drawing at the previous stage. All judgment standards are quantized, the image quality can be objectively judged by using an algorithm and image information, and the aesthetic fatigue caused by overlong human picture viewing time is replaced, so that errors can be caused in judgment. In addition, the image can be subjected to region division, and the image quality effect can be compared by a plurality of dimension quantitative evaluation in partial pixel points or partial image regions.

When the Image quality is evaluated, the Image quality evaluation method, the Image quality evaluation system and the Image quality evaluation device mainly divide an Image frame equally by Exchangeable Image File (Exchangeable Image File EXIF) information of the Image, intercept pixel point information of each region intersection or an Image of a certain region to evaluate and compare, and can calculate and compare by combining an algorithm and the existing functions and principles. EXIF refers to detailed information of an image, and may include image size, exposure time, ISO rate, and the like, and this scheme may also be extended to refine to separate analysis of each small term, such as analyzing brightness of all images, i.e. extracting histogram information of all images for comparison analysis, and analyzing noise of all images, i.e. extracting flat or dark area pictures of all images for SNR calculation comparison. More detailed dimensions such as focusing, sharpness, saturation and the like can be subjected to independent extraction analysis and are not limited to overall analysis evaluation.

Unless specifically stated otherwise, terms such as processing, computing, calculating, determining, displaying, or the like, may refer to an action and/or process of one or more processing or computing systems or similar devices that manipulates and transforms data represented as physical (e.g., electronic) quantities within the processing system's registers and memories into other data similarly represented as physical quantities within the processing system's memories, registers or other such information storage, transmission or display devices. Information and signals may be represented using any of a variety of different technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that may be referenced throughout the above description may be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

It should be understood that the specific order or hierarchy of steps in the processes disclosed is an example of exemplary approaches. Based upon design preferences, it is understood that the specific order or hierarchy of steps in the processes may be rearranged without departing from the scope of the present disclosure. The accompanying method claims present elements of the various steps in a sample order, and are not intended to be limited to the specific order or hierarchy presented.

In the foregoing detailed description, various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments of the subject matter require more features than are expressly recited in each claim. Rather, as the following claims reflect, invention lies in less than all features of a single disclosed embodiment. Thus, the following claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.

Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. Of course, the storage medium may also be integral to the processor. The processor and the storage medium may reside in an ASIC. The ASIC may reside in a user terminal. Of course, the processor and the storage medium may reside as discrete components in a user terminal.

For a software implementation, the techniques described herein may be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein. The software codes may be stored in memory units and executed by processors. The memory unit may be implemented within the processor or external to the processor, in which case it can be communicatively coupled to the processor via various means as is known in the art.

What has been described above includes examples of one or more embodiments. It is, of course, not possible to describe every conceivable combination of components or methodologies for purposes of describing the aforementioned embodiments, but one of ordinary skill in the art may recognize that many further combinations and permutations of various embodiments are possible. Accordingly, the embodiments described herein are intended to embrace all such alterations, modifications and variations that fall within the scope of the appended claims. Furthermore, to the extent that the term "includes" is used in either the detailed description or the claims, such term is intended to be inclusive in a manner similar to the term "comprising" as "comprising" is interpreted when employed as a transitional word in a claim. Furthermore, any use of the term "or" in the specification of the claims is intended to mean a "non-exclusive or".

Claims

1. An image quality evaluation method, comprising:

2. The method of claim 1, wherein determining the validation weight for each selected evaluation dimension based on the image type and at least one image attribute of the image to be evaluated comprises:

3. The method of claim 2, wherein the image attributes comprise at least one of image scene, image content characteristics, capture device, capture parameters;

4. The method of claim 1, wherein comparing the image to be evaluated with the selected comparison image to obtain the evaluation result of the image to be evaluated and the comparison image in each evaluation dimension comprises:

5. The method of claim 1, wherein when the comparison image is one image, the determining the comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension comprises:

6. The method of claim 1, wherein when there is more than one comparison image, the determining the comprehensive evaluation result of the image to be evaluated and the comparison image according to the evaluation result of each evaluation dimension and the confirmation weight of each evaluation dimension comprises:

7. The method of any of claims 1-6, further comprising:

8. An image quality evaluation apparatus characterized by comprising:

9. The apparatus of claim 8, further comprising:

10. An image quality evaluation method, comprising:

11. The method of claim 10, wherein prior to obtaining the validation weight for each evaluation dimension, further comprising:

12. An image quality evaluation apparatus characterized by comprising:

13. An image quality evaluation system, comprising: a testing machine, at least one comparing machine and the image quality evaluation device according to any one of claims 8 to 9 and 12;

14. A computer storage medium having computer-executable instructions stored therein, which when executed by a processor implement the image quality assessment method according to any one of claims 1 to 7 and 10 to 11.

15. An electronic device, comprising: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the image quality assessment method according to any one of claims 1 to 7 and 10 to 11 when executing the program.