=Paper=
{{Paper
|id=Vol-3156/paper40
|storemode=property
|title=Analysis of the effectiveness of means to achieve optimal color balancing in obtaining a digital photographic image
|pdfUrl=https://ceur-ws.org/Vol-3156/paper40.pdf
|volume=Vol-3156
|authors=Bohdan Kovalskyi,Myroslava Dubnevych,Tetyana Holubnyk,Inna Skarga-Bandurova,Ludmyla Maik,Zoryana Selmenska
|dblpUrl=https://dblp.org/rec/conf/intelitsis/KovalskyiDHSMS22
}}
==Analysis of the effectiveness of means to achieve optimal color balancing in obtaining a digital photographic image==
https://ceur-ws.org/Vol-3156/paper40.pdf
Analysis of the Effectiveness of Means to Achieve Optimal Color
Balancing in Obtaining a Digital Photographic Image
Bohdan Kovalskyia, Myroslava Dubnevycha , Tetyana Holubnyka, Inna Skarga-Bandurovab
Ludmyla Maika, Zoryana Selmenskaa
a
Ukrainian Academy of Printing, Lviv, Ukraine
b
Oxford Brookes University, United Kindom
Abstract
The publication considers one of the most important indicators of the quality of a digital
halftone image - colour reproduction, in particular colour balancing. The authors
systematized the factors influencing this indicator, compiled an informational multilevel
model of the importance of factors of influence and determined the priority of each of them.
Based on the model of priority influence of factors, it is determined that in order to achieve
optimal colour reproduction of digital images, the highest priority is the spectral composition
of lighting, and the next most important is the format of data digitization. Based on the results
of mathematical modelling, an experimental study of the influence of data digitization format
on the degree of colour balance and determined that the optimal result provides data
digitization in raw format with subsequent processing of the data file in specialized software.
The authors also investigated that the use of automatic balancing in white on the degree of
colour balancing has a significant impact on the method of exposure metering mode, which is
recommended for consideration in the practical implementation of the photographic process.
Recommendations for optimal adjustment of the workflow of formation and subsequent
processing of qualitative characteristics of digital photo images are given.
Keywords 1
digital halftone image, colour balance, raw format, matrix of photosensitive elements,
spectral characteristics, factors influencing the color balance, methods of exposure metering,
auto white balance mode, RAW converter, gradation characteristics.
Introduction
Digital photography today has confidently taken the position of the only means of obtaining the
halftone photographic image in the modern workflow of optical information recording. But the
predominantly obtained photographic images require significant improvement in their quality
characteristics in graphic editors and other software products. Further complicating the process of
post-photographic processing is the absence of a standard procedure for an objective assessment of
the quality of a digital photographic image, and the list and maximum permissible values of quality
indicators are not regulated by any regulatory document. The international ISO standards standardize
only a certain list of technical indicators of the quality of recording tools, but do not describe the
methodology for evaluating and quality indicators of digital photographic images. In particular, the
ISO 12232 standard defines how digital camera manufacturers set the exposure index, ISO value,
output level, how exposure is recorded in metadata. A number of other ISO standards relate to the
construction, operation and testing of digital cameras: resolution (ISO 12233), noise (ISO 15739),
IntelITSIS’2022: 3rd International Workshop on Intelligent Information Technologies and Systems of Information Security, March 23–25,
2022, Khmelnytskyi, Ukraine
EMAIL: bkovalskyi@ukr.net (B. Kovalskyi), dubnevychmyroslava@gmail.com (M. Dubnevych); tanagolubnik@gmail.com (T.
Holubnyk), iskarga-bandurova@brookes.ac.uk (I. Skarga-Bandurova), ludmyla.maik@gmail.com (L. Maik), zorselm@gmail.com (Z.
Selmenska)
ORCID: 0000-0002-5519-0759 (B. Kovalskyi), 0000-0002-5519-0759 (M. Dubnevych); 0000-0002-8325-9813 (T. Holubnyk), 0000-0003-
3458-8730 (I. Skarga-Bandurova), 0000-0001-8552-0942 (L. Maik), 0000-0002-9514-7923 (Z. Selmenska)
©️ 2021 Copyright for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
tone reproduction (ISO 14524), shading (ISO 17957), geometric distortion (ISO 17850), chromium
ISO 19084) , image flare (ISO 18844), shooting delay (ISO 15781), low light (ISO 19093) and image
stabilization (ISO 20954-1), test procedures for color characteristics (ISO 17321-1) and cameras in
general /TR 19247) [1, 6].
Analyzing information about the perception of visual information by a person, we can summarize
the following list of quality indicators of a digital photographic image: geometric parameters,
structural characteristics, gradation and color characteristics. The human brain reacts especially
critically to shortcomings in the structural indicators of the image [9, 10] (the presence of digital
noise, reduced sharpness and non-reproduction of small details of objects). But gradation loss and
color reproduction disturbances are no less important for adequate perception of information by the
consumer (reader).
The color image when shooting with any photographic equipment, including digital, depends on
many factors such as the distribution of the brightness of the subject (the ability to reflect or absorb
certain zones of the radiation spectrum of the light source), lighting features (power and spectral
composition of the light flux), optics characteristics (lens aperture) , the correct choice of exposure
(combination of shutter speed and aperture), the type and characteristics of separating media (a device
for registering single-color components of complex radiation), the characteristics of photosensitive
sensors.
Sensors of the photosensitive matrix register colors almost objectively, that is, if there is a
dominant radiation in the spectrum of the light source, this shade will prevail in the image. This
negative moment can be minimized by using the white balance function. The software of all models
of cameras (of different technical classes and purposes, including cameras for smartphones) has this
option. At the same time, depending on the development of the software of a particular camera, this
function can be implemented according to different algorithms: automatic balance, by source type, by
color temperature, and manual white balance. The latter algorithm also requires the use of special test
objects for adjusting the white point parameters to implement the color balancing process.
Complicating the process of obtaining a digital tone image with satisfactory color characteristics is
also the problem of integrating color management systems into the workflow of photography due to
changing lighting conditions.
Thus, the search for the optimal implementation of the balance of color reproduction of a digital
photographic image is an urgent problem that requires an integrated approach, taking into account the
significance of the factors influencing the specified quality indicator.
Related Works
Let us consider what physical phenomena underlie the process of forming the color characteristics
of photographic images. To obtain a full-color image in additive space, it is necessary to fix its three
single-color components as accurately as possible: red, green and blue partial image. The accuracy of
fixing these arrays of information is one of the determining factors influencing the color
characteristics of a digital photographic image (color reproduction). The chain of transformation of
color information from a real object (subject) to its digital image (digital color halftone photo image)
is as follows (Fig. 1). An object with certain spectral characteristics (the ability to reflect the visible
zones of the spectrum described by the function ρ0 (λ), receives radiation from a light source with a
spectral distribution of energy E0 (λ). Each natural or artificial light source is characterized by a
certain spectral composition of radiation, in which, depending on the color temperature, one or
another zone of the visible spectrum predominates. The luminous flux reflected by the object of
shooting E1 (λ), modulated according to the spectral composition , enters the light energy receiver - a
matrix of photosensitive elements, composed of charge-coupled devices (CCD) or complementary
metal oxide semiconductors (CMOS) through the optical system of the lens throughput τ0 (λ).
Both types of semiconductors (CCD and CMOS) are not characterized by selective
photosensitivity to individual zones of the visible spectrum. On fig. Figure 2 shows the dependence of
the relative sensitivity of transceivers of both types of matrices on the radiation wavelength. As can be
seen from these dependences, the magnitude of the photoelectric effect is practically the same in the
entire range of the visible spectrum (from 400 to 700 nm).
Light source E0( λ)
Object of shooting ρо(λ)
Light E1( λ)
Lens
τ0(λ)
Three separating light filters
τс1(λ)τс2(λ)τс3(λ)
Figure: 1. Scheme of the process of transforming color information from the subject into a
photographic image
Figure: 2. Spectral photosensitivity of typical CCD and CMOS arrays
The spectral characteristics of CCD and CMOS elements necessitate the use of separating devices
in photosensitive matrices [1] for separating a polychromatic light flux into three monochromatic
components. The vast majority of models of modern digital photographic equipment (except for a few
models equipped with a multilayer Foveon matrix, which, however, is of limited use) use arrays of
light filters (Bayer array) with subsequent mathematical interpolation of color data in each image
pixel (debayering algorithms). The transmittance of three separating light filters ( τ𝑐1 (λ), τ𝑐2 (λ),
τ𝑐3 (λ),) installed in front of light-sensitive transceivers determines the quantity and quality of the
light flux entering the receiver.
Thus, the integral response of the light energy receiver, which describes the characteristics of a
section of a color photographic image, is defined as (1):
r = f ( a ) = f 0 ( ) E1 ( ) 0 ( ) c ( ) d .
0 (1)
The color characteristics of a photographic image in digital form depend on a whole list of factors,
among which one of the most significant is the spectral characteristics of separating media, according
to the sources [2, 3] [8], are far from ideal.
The status of the spectral sensitivity of light energy receivers of digital photographic equipment
from three different manufacturers according to a standard set of light filters is shown in fig. 3 as a
dependence of the relative response of the receiver on the light filter on the wavelength [5]. The blue
dash-dotted line describes the sensitivity of the receiver to the blue filter, the green line to the green
filter, and the red line to the red filter, respectively. It can be seen from the depicted graphic
dependences that all sets of light filters are wide-zoned (transmission zones of light filters within the
set overlap), which will inevitably lead to the registration of light fluxes of some individual
wavelengths of radiation by two or even three light filters. This is the reason for the inaccuracy of
fixing the color information and the subsequent inaccuracy of the color image of the photographic
object in the digital image.
It is possible to prevent the occurrence of color reproduction deficiencies caused by the
imperfection of the separating media by using color management systems. The purpose of color
management systems is to coordinate the color gamut of various colorimetric systems of input and
output devices (visualization) of information, as well as the conversion of color coordinates from one
color system to another [8, 11 ]. It is especially difficult to achieve satisfactory color reproduction in
digital photographic images due to the technical limitations of using color management systems in the
digital photographic process.
All color transformation operations are carried out using ICC profiles, which describe, on the basis
of standardized colorimetric systems, the possibilities of reproducing the color gamut of a particular
device (in particular, a digital camera). For image input devices, a number of factors affect the final
color rendering result: the optical system, the characteristics of the separating media and devices
(filters, light sources), and software.
а b
c
Figure: 3. Status of the spectral sensitivity of receivers of CCD and CMOS matrices of various
configurations (as a dependence of the relative response of the receivers on the wavelength of light
radiation) (a – full-frame Kodac CCD; b – Sony column-buffered CMOS array; c – Agilent CMOS array)
[1, 5]
The reasons for the limited use of the color management system in the workflow of the digital
photographic process, as shown in [4, 8], are significant technological and technical limitations:
1. When photographing with digital photographic equipment, the exposure is carried out each time
under changing lighting conditions: both in terms of the intensity of the light flux and with its
different spectral composition, which is especially critical for color reproduction. The profile obtained
in this way will be a strictly narrow application for frames taken only under these specific lighting
conditions. In each particular case, the profiling procedure should be repeated.
2. Profiling the process of photographing with digital photographic equipment is very laborious
and has a number of features:
– ensuring the uniformity of the lighting conditions of the photographed object (the use of the
same light sources with a strictly controlled color temperature, the absence of extraneous illumination
of a different spectral composition in the frame, the uniformity of the illumination of the object);
– the test object must be located in the frame in way to avoid reflections from light sources and
nearby foreign objects, which will violate the correspondence of the colorimetric coordinates of the
fields of the test object itself;
- illumination of the test object must be strictly uniform in terms of the intensity of the light flux,
and the exposure itself must be accurately calculated to avoid under- and overexposure;
– when photographing, it is necessary to deactivate the algorithms for improving the qualitative
and quantitative characteristics of a photographic image, which are used by default by the software for
photographic equipment at the stage of processing data from a photosensitive matrix.
3. When constructing a profile of input devices, test objects with a large number of control fields
(from several hundred to several thousand) are used, the absolute number of which has a direct impact
on the accuracy of the result. Each type of photographic material, on which a test object for
calibration is produced, is characterized by its own unique features in color reproduction.
а) b)
c)
Figure: 4. Distribution of luminescence energy over the spectrum of light sources: a) -
daylight; b) - LED lamp of a warm glow spectrum; c) - fluorescent mercury lamp [10]
Thus, color management systems can be used only in a limited list of photography genres: fashion
photography, food photography, portrait photography and other types of staged scenes, static in time,
before photographing which it is possible to carry out profiling. Other things like reportage, sports
photography, street photography, due to the dynamics of the scenes and the rapid change of events in
the plane of the frame, do not allow for preliminary photography of the test object to build an ISS
profile. If you do not profile the camera, then the photo image cannot avoid the appearance of color
reproduction flaws, which, in turn, must be eliminated during post-photographic processing in graphic
editors.
Another significant factor influencing the color characteristics of a digital photographic image is
the spectral composition of the light at which the frame is exposed. Different light sources have
different spectral composition of the luminous flux, which is described by the color temperature (Fig.
4).
In conventionally white radiation from different light sources, a certain part of the visible spectrum
always predominates. Sensors of the photosensitive matrix of the photorecording system react to the
light flux reflected from the photographed object, and the spectral composition depends both on the
spectral characteristics of the object itself and on the spectral composition of the illumination.
Since light-sensitive receivers fix light radiation objectively, in order to achieve the
correspondence of colors in the photographic image to the color gamut of the subject and eliminate
the excessive tint due to the peculiarities of the spectral composition of the radiation of the light
source in the software of cameras of various technical classes and purposes, in particular, and cameras
of smartphones, the available function white balance. Depending on the development of the software
of a particular camera, this function can be implemented using different algorithms: automatic
balance, source type, color temperature, and manual white balance. The latter algorithm also requires
the use of special test objects for adjusting the white point parameters to implement the color
balancing process.
The color balance of a photographic digital image can be improved at the stage of post-photo
processing in RAW converters. This is a class of software products that work with a special data
digitization format, which is called the RAW format. The working environment of all RAW
converters without exception allows you to adjust the color balance using the appropriate tools, since
the RAW file format provides for recording information without processing by the camera software.
To sum up, there are several algorithms for achieving color balance in a digital photographic
image. Information sources do not contain data on an objective analysis and comparison of the results
of applying each of the white balance implementation options, except for a purely description of the
sequence of applying specific options from the camera software menu or graphic editors [13]. It is
appropriate to compare the results that provide the listed means. In addition, it is known that in order
to achieve optimal color balance, already at the exposure stage (obtaining a digital photographic
image), it is necessary to provide an optimal level of illumination according to the exposure meter.
Built-in and external exposure meters, depending on the selected method of estimating (measuring)
illumination in the plane of the frame, take into account the brightness of a larger or smaller number
of objects and the background. From the ratio of dark and light objects in the plane of the frame, the
exposure metering system of photographic equipment sets the value of the exposure pair. It can be
assumed that not only tone reproduction, but also the color balance will be different for each metering
method.
Information model of the significance of factors influencing the accuracy of
balancing the colors of a digital photograph
To solve the problems formulated above, we will first of all build an information model for
determining the significance of factors influencing the accuracy of color balancing of a digital
photographic image at the stage of its acquisition. For this purpose, a technique was chosen that
provides for the selection of many factors that directly affect the quality of the impact on the balance
of colors of a digital photograph, the formation of the initial graph of relationships between them, the
implementation of iterative procedures over the reach matrix, and the synthesis of the factor model by
the method of structuring relations [7].
Based on the analysis carried out, we highlight the main factors influencing the balance of colors.
They are as follows: status of the spectral sensitivity of receivers (SS), white balancing algorithm
(WBA), application of color management systems (ICC), spectral composition of illumination (SL),
spectral characteristics of the subject (OS), type of photosensitive matrix (TM), and format data
digitization (FF).
We will consider the process of influencing the color balance of a digital photographic image as a
certain function, the arguments of which will be the listed factors.
PR = F ( s1 , s2 , s3 , s4 , s5 , s6 , s7 )
, (2)
where s1 – status of spectral sensitivity of receivers (SS); s2 – photosensitive matrix type (TM) s3
− spectral warehouse lighting (SL); s4 − spectral characteristics of the subject (OS); s5 − white
balance algorithm (WBA); s6 − data digitization format (FF); s7 − application of color management
systems (ICC).
Certain factors in terms of terminology and essence are referred to linguistic variables, which in
the tasks of the prepress process can be indicators of the quality of a digital photographic image:
geometric parameters, structural characteristics, gradation and color characteristics. To do this, we
will build an initial graphical model (directed graph), taking into account expert judgments on
pairwise effects (relations) between factors (Fig. 5).
Spectral
sensitivity
status (s1)
Color
Image sensor management
type (s2) system (s7)
Spectral Data
state of Digitization
lightening (s3) Format (s6)
Spectral
characteristics White
of the object balance
(s4) algorithm (s5)
Figure: 5. The initial graph of relations between the factors for balancing the colors of a
digital photo
Source graph Fig. 5 is used to order the factors according to the importance of the impact on the
process under study, which will result in a multilevel model of the factors influencing the color
balance of the digital photographic image. To synthesize a linguistic model, we use the tools of matrix
theory and system analysis.
Using the existing graphical model - an analogue of the semantic network, we build a binary reach
matrix (Table 1), which simulates possible options for getting from each vertex of the graph to other
vertices.
The matrix is constructed by filling in the table, the binary elements of which are determined by
the following logical rule:
1, if bi you can get into b j
bi j = (3)
0, if bi you can not get into b j
Almost top 𝑠𝑗 (j=1, 2, …, 7) the original graph in Fig. 5 is considered reachable from the top 𝑠𝑖
(i=1, 2, …, 7), if from the latter it is possible to get to an arbitrary path, taking into account transitions
through other vertices. The result of the analysis of all vertices leads to a subset of reachable vertices
𝐷(𝑠𝑖 ).
At the same time the top si we will consider the predecessor of the vertex sj , if it is reached from
it, and their totality forms a subset P(si ). Finally, the section of subsets of reachable vertices and
predecessors forms a separate subset:
Z ( si ) = D ( si ) P ( si )
, (4)
which determines a certain level of priority for the action of factors related to these vertices. An
additional condition for this is to ensure equality
P ( si ) = Z ( si )
, (5)
the implementation of dependencies (4) and (5) using iterative tables leads to the formation of the
corresponding levels, the initial of which is the highest in terms of the priority of the impact on the
process under study. To determine the specified level, we use the reach matrix and mathematical
dependencies (4) and (5), on the basis of which we build Table. 2.
Table 1
Reach Matrix
SS ТМ SL OS WBA FF ICC
SS 1 0 0 0 1 0 1
ТМ 1 1 0 0 1 0 0
SL 1 0 1 0 1 0 1
OS 0 0 0 1 1 0 0
WBA 0 0 0 0 1 0 0
FF 0 0 0 0 1 1 1
ICC 0 0 0 0 1 0 1
Table 2
Results
D ( si ) P ( si ) D ( si ) P ( si )
1,5,7 1,2,3 1
1,2,5 2 2
1,3,5,7 3 3
4,5 4 4
5 3,4,5,6 5
5,6,7 6 6
5,7 1,3,6,7 7
Subset - the numbers of reachable vertices or the numbers of single elements of the corresponding
rows of the reach matrix are entered in the second column of the table; the third column defines a
subset of the vertices of the predecessors - the numbers of unit elements of the columns of this matrix.
In this case, dependency (5) means the fulfillment of the condition of equality of the numbers of
factors specified in the second and third columns of the table, as a result of which a certain level of the
hierarchy of factors is formed in the resulting graphical model.
As can be seen from Table. 2, the coincidence of numbers is fixed for factor 3 - spectral
illumination. This factor will be considered the highest in terms of the priority level of influence on
the process of influencing the color balance of a digital photographic image.
According to the methods of system analysis and mathematical modeling of hierarchies [7], we
remove from the table. 2 is the third line, and in the second and third columns of this table we cross
out the number 3, respectively. We get a table that is the basis for calculating the next iteration - the
basis of the next most important level of the hierarchy of factors.
Analysis of the table. 3 is produced according to the above algorithm. It is easy to see that the
coincidence of numbers is fixed for factor 6 - the format of data digitization, which forms the next
level of the hierarchy from the top.
Actions similar to those described above are given in table 4, in which the row with the number 6
is extracted, and this number is missing in the second and third columns of the table.
Table 3
D ( si ) P ( si ) D ( si ) P ( si )
1,7 1,2,3 1
1,2 2 2
4,5 4 4
5 3,4,5,6 5
5,6,7 6 6
7 1,3,6,7 7
Table 4
D ( si ) P ( si ) D ( si ) P ( si )
1,7 1,2,3 1
1,2 2 2
4,5 4 4
5 3,4,5,6 5
7 1,3,6,7 7
From Table. 4 we obtain factors 2 and 4 - the spectral composition of the illumination and the
spectral characteristics of the subject. As a result of repeating the procedures, we get:
Table 5
D ( si ) P ( si ) D ( si ) P ( si )
1,7 1,2,3 1
5 3,4,5,6 5
7 1,3,6,7 7
Actions similar to those described above are given in table. 5, in which the row with number 1 is
extracted, and this number is missing in the second and third columns of the table.
Table 6
D ( si ) P ( si ) D ( si ) P ( si )
5 3,4,5,6 5
7 1,3,6,7 7
Tab. 6 leads to the exclusion of two factors at once 5 - the white balancing algorithm, 7 - the use
of color management systems.
SL
S3
FF
S6
TM OS
S2 S4
SS
S1
WBA ICC
S5 S7
Figure: 6. Multilevel model of factors influencing the color balance of a digital photographic image
Using the iterative analysis data and taking into account [7], we synthesize a multilevel structured
graphical model (Fig. 6), which clearly displays the place of each factor and reproduces the
relationships between them specified in the original model in Fig. 5.
On the basis of a multilevel synthesized model of the effect on the color balance of a digital
photographic image, we construct a priority model in which (Fig. 7) the priority of the factor is
determined by the level of its placement. The highest priority among the selected factors has the
spectral composition of illumination, which is associated with the next factor - the data digitization
format.
Taking into account the factors reflected in the qualitatively weighted original graph of Fig. 6 and
using the tools of the system analysis matrix theory, a linguistic model is synthesized, resulting in a
multilevel model of factors influencing the color balance of digital photographic image. To ensure the
quality of the impact on the color balance of a digital photographic image, a model of priority impact
was built by the method of structuring relationships.
Determining the degree of influence of the exposure metering method on
the accuracy of the balance from a white digital photo
The constructed model of the priority influence of factors influencing the color balance of a
digital photographic image found that the format of data digitization is one of the most priority
factors. Therefore, based on the information received, we will conduct a practical study of the effect
on the quality characteristics of digital photographic images, the color balance tools of the camera
software when digitizing information immediately after exposure, and the corresponding color
balance tools of the RAW converter. In addition, we will determine how the method of exposure
metering with automatic white balance affects the color characteristics of a photographic image. Such
a factor of influence on the accuracy of color balancing is not noted by any researchers of the
described technological processes, however, the authors’ preliminary studies of the influence of this
factor on the gradation (quantitative) characteristics of a color digital photographic image indicate its
significance, which suggests an effect on the mentioned factor and on qualitative (color )
photographic image characteristics.
Influence on the color balance of a digital
photographic image
Spectral composition of illumination S3
(SL)
Data Digitization Format S6
Image sensor type Spectral characteristics
S2 of the object S4
Spectral sensitivity status S1
приймачів
White balance S5 Color management S7
algorithm system
Figure: 7. Model of the priority influence of factors of influence on the color balance of a digital
photographic image
There are several ways to adjust the color balance in a photographic image immediately after
exposure: automatically, by source type, and manual balance by sample. The latter requires a white or
neutral gray reference and is only available on high-end camera models. White balance according to
the type of light source, as studies have shown, does not always guarantee the best result, since there
is often a discrepancy between the description of the spectral characteristics of common light sources
and their spectral characteristics (for example, fluorescent and gas discharge lamps) [9].
Thus, automatic white balance is often used. When using this option, according to the
manufacturers of digital photographic equipment, the camera software analyzes the image and
determines the brightest part of it, which the application tries to bring to white. The hue that the
software fixes on the brightest object is considered parasitic and is removed during processing from
the entire image.
The process of image analysis itself is of interest, in particular, what part of the frame plane of a
photographic image is taken into account. It can be assumed that metering methods that differ in the
light measurement area affect the result of collecting data for color balancing. As you know, four
methods of exposure metering are used, which differ in the area of light measurements in the frame:
matrix or segment (taking into account the frame area), center-weighted (80% of measurements are
made in the central part of the frame, 20% - along the periphery), partial (measurement in the center
of the frame within the area that occupies up to 10% of the frame area) and spot (measuring in the
very center of the frame plane in a very narrow area - 2-5% of the total area).
To test the formulated hypothesis, an experiment was carried out with the following technical
shooting parameters: manual setting of exposure conditions according to the data of the built-in
exposure meter, ISO 200 sensitivity, automatic white balance mode, file recording format: JPEG.
Under these exposure conditions, a photographic image of the test object (an achromatic stepped
scale) was obtained on differently colored backgrounds using various exposure metering methods.
The color characteristics of the obtained photographic images are shown in Figs. 8 and fig. 9.
The presented graphic dependences demonstrate the content of the primary colors of additive
synthesis in the fields of the achromatic scale on photographic images from similar indicators of the
test object.
The presented characteristics show that the selected metering method is essential for automatic
white balance.
Regardless of the background color, with the segment metering method, which is characterized
by the largest area for collecting information about the light reflected by the subject, there is a
significant imbalance of colors, namely, the content of the color component is minimized, which
corresponds to the color of the background.
The degree of color imbalance is directly proportional to the background brightness, which for a
green background is - 67, 54 - fig. 8 (a), and yellow - 96.63 - fig. 8 (b), respectively.
When applying the point method of exposure metering and adjusting the plane for collecting
information on the quantity and quality of the light radiation according to the neutral gray field of the
achromatic scale, we obtained a much better, close to ideal, degree of color balance (Fig. 9).
Subsequent studies have shown that the achromatic background of a frame of different brightness
does not affect the color characteristics of a photographic image with all exposure metering methods.
Thus, the conducted studies allow us to give the following practical recommendations. If the
background in the frame differs significantly in brightness from the scene-important object, then the
exposure parameters of the shooting should be selected either in manual mode, or in order to avoid
errors in semi-automatic mode using the spot metering method. When photographing scenes with a
high-brightness color background (such as an autumn or spring landscape with a large plane of yellow
leaves or greenery, commercial subject matter photography), you should definitely use the partial or
spot metering method, which will provide satisfactory gradation and automatic color balance.
250 RGBфот
240 B
230
220 R
210
200 G
190
180
170
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
RGBориг
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250
a)
250 RGBфот B
240
230
220
G
210
200
190
180 R
170
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
RGBориг
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250
b)
Figure: 8. Color characteristics of the photo image obtained in the automatic white balance
mode and the segment metering method (a – green background, coordinates Lab 67,54;-47,17;
33,85), b - yellow background, coordinates Lab 96,63;-4,07; 68,65)
250 RGBфот
240
230
R G B
220
210
200
190
180
170
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10 RGBориг
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250
a)
250 RGBфото
240
230 B
220
G
210
200 R
190
180
170
160
150
140
130
120
110
100
90
80
70
60
50
40
30
20
10
RGBориг
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250
a
b)
Figure: 9. Color characteristics of a photographic image obtained in the auto white balance
mode and by the spot metering method (a - green background, coordinates Lab 67,54;-47,17; 33,85),
b - yellow background, coordinates Lab 96,63; -4,07; 68,65)
The central focus point, within which the illumination is measured, should be aimed at a scene-
weighty object.
Segment or center-weighted exposure metering method is appropriate to use when objects in the
frame are diverse in brightness and color without a predominance of a certain color or brightness.
The relationship between the format of digitizing a photographic image
and the balance of colors
Since the presented information model showed a significant influence of the digitization format
on the degree of color balance, we will study what color characteristics are formed in photographic
images when digitized in jpeg and raw formats and when using various white balancing algorithms.
The following graphic dependencies represent the color characteristics of photographic images
digitized in jpeg format and processed with white balance tools from the camera software (Fig. 10),
digitized in raw format (Fig. 11) and digitized in raw format, followed by color balance in RAW-
converter (Fig. 12). In all cases, the exposure was carried out under the light of a warm LED lamp, the
spectrum of which is dominated by the red zone of the spectrum.
According to the mutual arrangement of the curves in Figs. 10, we can come up with
unambiguous conclusions about the degree of color balance.
Achromatic shades on a perfectly balanced color image should be formed by the same color
content of additive synthesis, which in turn should be displayed on graphic dependencies by the
location of the content curves of red, blue and green colors close to each other.
If the curves on the coordinate plane are located in any other way, this indicates an imbalance in
colors.
Yes, fig. 10a shows the color characteristics of the photographic image with automatic white
balance, which demonstrate the imbalance of colors: an increased content of the red component. At
the same time, a photographic image processed by manual white balance using an achromatic color
sample demonstrates an ideal color balance (Fig. 10 (b)).
On fig. 11. Graphical dependences of the color content of additive synthesis on a gray scale of
raw photographic images in RAW format are shown.
The degree of color balance in these raw photos is similar to the corresponding JPEG photos: the
best color balance is in the photo obtained by manual white balancing on an achromatic pattern (Fig.
11(b)). It is appropriate to note that the color balancing on the analyzed RAW photographs is
different. That is, the color balance algorithm at the stage of data processing by the camera software,
contrary to what is common in some information sources of data, has a certain effect on the data
digitized in the RAW format.
Reproduction of gradation, which can also be evaluated on the obtained graphic dependencies,
also differs for photographic images in RAW and JPEG formats. If a photo in JPEG format has
significant loss of gradation in the shadow range, then in RAW format it is characterized by smooth
reproduction of gradation over the entire range of tones.
After processing in the RAW converter, all photographic images received a satisfactory color
balance (Fig. 12). The result of the white balance, made in the RAW converter, does not depend on
the conditions under which the white balance was taken, but only on the choice of the neutral area of
the photo image as a sample for color balance.
This method of achieving color balance provides good results and is not difficult to implement,
since it does not require the use of any additional tools (neutral color standards, in particular), does
not complicate the technological process with lengthy adjustments and additional exposure to adjust
the custom white balance.
a)
b)
Figure: 10. Color characteristics of photographic images processed by means of the camera software
white balance (a – automatic white balance; b – manual white balance (by achromatic pattern))
a)
b)
Figure: 11. Color characteristics of photographic images in RAW format without processing (a -
automatic white balance; b - manual white balance (according to an achromatic sample))
a)
b)
Figure: 12. Color characteristics of photographs processed in the RAW converter (a - automatic
white balance; b - manual white balance (according to an achromatic sample))
Conclusions
The following conclusions can be drawn from the conducted research. Since there are a large
number of light sources with an unbalanced spectral composition, there will always be a need to use
color balancing tools designed to eliminate excess color cast. The consumer has several white balance
algorithms that can be applied both at the time of processing data from the photosensitive matrix by
the camera software, and at the stage of post-photo processing. Among all the white balance methods
at the stage of photography, the best result was provided by manual white balance. However, studies
show that this mode requires the use of additional material support, and the result of color balancing is
no better than the balance obtained using the RAW converter. In this case, the qualitative
characteristics of the processed images do not depend on the characteristics of the original image, but
only on the selected neutral tone sample, according to the color characteristics of which the white
balance is realized. In the course of the described studies, another interesting fact was revealed:
despite the fact that the RAW format is described as a format without data processing by the camera
software, the use of various white balance tools during photography leads to different color gamuts
and the degree of color balance in photographic images. RAW format.
When using the automatic white balance tool, be sure to take into account that the metering
method affects the final result, so it is appropriate to use a point or partial method of measuring
lighting in the frame and set the central point on a plot-weighty object. Since, in addition to improving
the color characteristics of the photo image, the RAW format provides better gradation characteristics,
and the implementation of the white balance function does not require the use of cost test objects,
therefore, the optimal solution would be to take photographs in the following sequence: under
appropriate scene lighting, perform an exposure test about neutral gray colors, to shoot with automatic
white balance and digitize photo images in RAW format and then batch work out the color balance of
all photo images in the RAW converter environment. If there are objects of neutral color (white, gray
or black) in the frame, the test object (gray card, etc.) can be omitted. Such a sequence of
implementation of the technological process will reduce its complexity and ensure high quality
indicators of the photographic image.
References
[1] S. Mehta, A. Patel, J. Mehta, CCD or CMOS Image sensor for photography. International
Conference on Communications and Signal Processing (ICCSP). 2015. doi:
https://doi.org/10.1109/iccsp.2015.7322890.
[2] J. Duran, A. Buades A Demosaicking Algorithm with Adaptive InterChannel Correlation.
Image Processing On Line. 2015. Vol. 5. P. 311–327. doi:
https://doi.org/10.5201/ipol.2015.145.
[3] M.Sahani, S. K. Rout, A. K. Panigrahi, A. S. Acharya. Modified color histogram equalization
with variable enhancement degree for restoration of skin color. 2015 International Conference
on Communications and Signal Processing (ICCSP). 2015. doi:
https://doi.org/10.1109/iccsp.2015.7322561.
[4] Z.Yanhua, L. Jiaogang. Research on the Expression of Photography Language under Digital
Technology, 2020 5th International Conference on Electromechanical Control Technology
and Transportation (ICECTT), 2020, pp. 228-230, doi: 10.1109/ICECTT50890.2020.00058.
[5] E.Gurdiel, J. Y.Hardeberg. Spatially dependent white balance for fill flash photography, 2018
Colour and Visual Computing Symposium (CVCS), 2018, pp. 1-5, doi:
10.1109/CVCS.2018.8496644.
[6] M.Afifi, M. S. Brown. Deep White-Balance Editing, IEEE/CVF Conference on Computer
Vision and Pattern Recognition (CVPR), 2020, pp. 1394-1403, doi:
10.1109/CVPR42600.2020.00147.
[7] I.Ogirko, M. Dubnevych, T. Holubnyk, N. Pysanchyn O. Ogirko, Information Technology to
Determine the Effectiveness of Advertising Communication International Scientific-Practical
Conference on Problems of Infocommunications Science and Technology, PIC S and T 2018
- Proceedings, 2019, pp. 752–756, 8632016.
[8] B.Kovalskyi, M. Dubnevych, T. Holubnyk, N. Pysanchyn, B. Havrysh, Development of a
technology for eliminating color rendering imperfections in digital photographic images
Eastern-European Journal of Enterprise Technologies, 2019, 1(2-97), pp. 40–47.
[9] Adjusting white balance settings to improve photos. URL:
https://www.adobe.com/creativecloud/photography/discover/white-balance.html.
[10] J. Bosley Understanding White Balance – A Beginner’s Guide. URL:
https://photographylife. com/definition/white-balance.
[11] V. Chad. How to Correct White Balance in Photoshop. URL:
https://expertphotography.com/ correct-white-balance-photoshop/.
[12] Ch. Hoiberg How to Master White Balance Like a Pro. URL:
https://capturelandscapes.com/ master-white-balance-like-pro/.
[13] K. Munger RAW and Jpeg White Balance Problems. URL:
https://photojottings.com/rawand-jpeg-white-balance-problems/.