ISO/TS 19567-2:2019

TECHNICAL ISO/TS
SPECIFICATION 19567-2
First edition
2019-07
Corrected version
2019-11
Photography — Digital cameras —
Part 2:
Texture analysis using stochastic
pattern
Photographie — Caméras numériques —
Partie 2: Analyse de la texture en utilisant un modèle stochastique
Reference number
ISO/TS 19567-2:2019(E)
©
ISO 2019

---------------------- Page: 1 ----------------------
ISO/TS 19567-2:2019(E)

COPYRIGHT PROTECTED DOCUMENT
© ISO 2019
All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may
be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting
on the internet or an intranet, without prior written permission. Permission can be requested from either ISO at the address
below or ISO’s member body in the country of the requester.
ISO copyright office
CP 401 • Ch. de Blandonnet 8
CH-1214 Vernier, Geneva
Phone: +41 22 749 01 11
Fax: +41 22 749 09 47
Email: copyright@iso.org
Website: www.iso.org
Published in Switzerland
ii © ISO 2019 – All rights reserved

---------------------- Page: 2 ----------------------
ISO/TS 19567-2:2019(E)

Contents Page
Foreword .iv
Introduction .v
1 Scope . 1
2 Normative references . 1
3 Terms and definitions . 1
4 Test conditions and methods . 1
4.1 General . 1
4.2 Environmental conditions . 2
4.3 Apparatus and hardware . 2
4.4 Arrangement of measuring equipment . 2
4.4.1 Reflective test chart . 2
4.4.2 Transmissive test chart . 2
4.4.3 Lighting . 3
4.4.4 Camera settings . 3
4.5 Test Chart . 4
4.5.1 General. 4
4.5.2 Chart generation . 4
5 Analytical approach. 5
5.1 General . 5
5.2 Measurement method . 6
6 Presentation of results . 8
6.1 SFR curve . 8
6.2 Single numerical values . 9
6.2.1 General. 9
6.2.2 SFR10 . 9
6.2.3 SFR50 .10
6.2.4 Texture acutance .10
Annex A (informative) Differentiation of cyclic pattern and stochastic pattern .12
Annex B (informative) Exposure condition for the measurement .14
Annex C (informative) Inconsistency between results of the measurement method specified
in this document and subjective evaluations .17
Bibliography .20
© ISO 2019 – All rights reserved iii

---------------------- Page: 3 ----------------------
ISO/TS 19567-2:2019(E)

Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards
bodies (ISO member bodies). The work of preparing International Standards is normally carried out
through ISO technical committees. Each member body interested in a subject for which a technical
committee has been established has the right to be represented on that committee. International
organizations, governmental and non-governmental, in liaison with ISO, also take part in the work.
ISO collaborates closely with the International Electrotechnical Commission (IEC) on all matters of
electrotechnical standardization.
The procedures used to develop this document and those intended for its further maintenance are
described in the ISO/IEC Directives, Part 1. In particular, the different approval criteria needed for the
different types of ISO documents should be noted. This document was drafted in accordance with the
editorial rules of the ISO/IEC Directives, Part 2 (see www .iso .org/ directives).
Attention is drawn to the possibility that some of the elements of this document may be the subject of
patent rights. ISO shall not be held responsible for identifying any or all such patent rights. Details of
any patent rights identified during the development of the document will be in the Introduction and/or
on the ISO list of patent declarations received (see www .iso .org/ patents).
Any trade name used in this document is information given for the convenience of users and does not
constitute an endorsement.
For an explanation of the voluntary nature of standards, the meaning of ISO specific terms and
expressions related to conformity assessment, as well as information about ISO's adherence to the
World Trade Organization (WTO) principles in the Technical Barriers to Trade (TBT), see www .iso .org/
iso/ foreword .html.
This document was prepared jointly by Technical Committee ISO/TC 42, Photography.
A list of all parts in the ISO 19567 series can be found on the ISO website.
Any feedback or questions on this document should be directed to the user’s national standards body. A
complete listing of these bodies can be found at www .iso .org/ members .html.
This corrected version of ISO/TS 19567-2:2019 incorporates the following correction:
— The mention of the collaboration with IEC/TC 100, Audio, video and multimedia systems and
equipment, was removed from the Foreword.
iv © ISO 2019 – All rights reserved

---------------------- Page: 4 ----------------------
ISO/TS 19567-2:2019(E)

Introduction
In a general context, texture refers to the visual and tactile surface quality derived from the physicality
of a material and the roughness or graininess of its surface. For digital still camera images, texture is
restricted to the visual surface quality and the characteristic of texture reproduction in the captured
image can be interpreted as the reproduction of the low contrast fine details. This document specifies
the measurement of how cameras reproduce texture defined as low contrast fine details.
The on going tendency to utilize smaller sensors with higher pixel counts in some cameras leaves
a very small amount of light reaching each individual pixel. With the signal getting smaller and the
noise level remaining at a certain level, it is necessary to reduce the noise in the image processing after
capturing the image. Although the algorithms used for noise reduction have been developed over time,
they are still not able to differentiate texture in the actual scene from the unwanted noise introduced
by the capturing system. This decreases the image quality and it is therefore helpful to have a method
to measure the loss of texture. Texture may also be enhanced to increase the acutance of the image.
The texture reproduction is dependent on frequency and contrast because the noise reduction and the
acutance enhancement, etc., are nonlinearly dependent on the values of the surrounding pixels.
This document specifies methods to measure texture reproduction using test charts with a stochastic
pattern. Annex A talks about the differentiation of this document from ISO/TS 19567-1, which deals
with cyclic pattern. The test charts described here are based on randomly arranged circles of various
sizes and colour with a limited contrast. This provides a target with known structure and spatial
statistics similar to natural images. The measurement results are presented in SFR (Spatial Frequency
Response) curves from which a single value representing the overall texture content is derived.
In general if one measured SFR is greater than the other across all measured spatial frequencies, a
larger amount of texture is reproduced in the corresponding image. If two SFRs have a crossover point
and the larger SFR depends on the frequency range, relative ordering of texture preservation quality
is less clear. Comparison of the measurement results can provide important information about the
relative texture reproduction of the captured images.
While the measurement method specified in this document is for objective evaluations of texture
reproduction for images, their relationship to subjective evaluations of texture reproduction with visual
perception is important to give attention, since image quality for camera/photograph users generally
accords with subjective evaluation. Annex C explains possible inconsistency between measurement
results using the method described in this document and subjective evaluations, due to different
condition of noise, with experimental results for images.
© ISO 2019 – All rights reserved v

---------------------- Page: 5 ----------------------
TECHNICAL SPECIFICATION ISO/TS 19567-2:2019(E)
Photography — Digital cameras —
Part 2:
Texture analysis using stochastic pattern
1 Scope
This document specifies a protocol to measure the texture reproduction in images captured and
processed by digital cameras including cameras in other devices e.g. in camera phones.
This document specifies protocols for the measurement of texture reproduction using test charts with
stochastic pattern.
NOTE The measurement method specified in this document is for objective evaluations of texture
reproduction, of which the results are sometimes inconsistent with subjective evaluations (See Annex C).
2 Normative references
There are no normative references in this document.
3 Terms and definitions
For the purposes of this document, the following terms and definitions apply.
ISO and IEC maintain terminological databases for use in standardization at the following addresses:
— ISO Online browsing platform: available at http:// www .iso .org/ obp
— IEC Electropedia: available at http:// www .electropedia .org/
3.1
texture
low contrast fine details, which appear in objects
EXAMPLE Low contrast fine details, which is visible in foliage, fur, sand, textiles, grass, or masonry surfaces.
3.2
texture reproduction
response in the output image of cameras to the texture of the object in the scene
4 Test conditions and methods
4.1 General
The measurement shall be carried out using digital images of the texture test chart captured by a
digital still camera.
The following measurement conditions should be used as nominal conditions when measuring the
texture reproduction of a digital still camera. If it is not possible or appropriate to achieve these nominal
operating conditions, the actual operating conditions shall be listed along with the reported results.
© ISO 2019 – All rights reserved 1

---------------------- Page: 6 ----------------------
ISO/TS 19567-2:2019(E)

4.2 Environmental conditions
The measurement shall be carried out in the following environment unless otherwise stated:
— Temperature: 23 °C ± 3 °C.
4.3 Apparatus and hardware
Either a reflective chart or a transmissive chart may be used. The light flux from the target shall be
diffuse and shall not include any specular component.
Each test chart shall be specified, together with the lighting conditions such as illuminance, luminance
and colour temperature of illumination.
4.4 Arrangement of measuring equipment
4.4.1 Reflective test chart
The arrangement of the measuring equipment for a reflective test chart shall be set up as shown in
Figure 1. The camera shall be positioned so that it casts no shadow on the chart. The lamps shall be
positioned at an angle which avoids direct specular reflection from the test chart entering the camera.
Key
1 lamp
2 digital camera
3 baffles to prevent direct illumination of the camera lens by the lamps
Figure 1 — Arrangement of measuring equipment for reflective test chart
4.4.2 Transmissive test chart
The arrangement of the measuring equipment for a transmissive test chart shall be set up as shown in
Figure 2.
2 © ISO 2019 – All rights reserved

---------------------- Page: 7 ----------------------
ISO/TS 19567-2:2019(E)

Key
1 digital camera
Figure 2 — Arrangement of measuring equipment for transmissive test chart
4.4.3 Lighting
The default colour temperature of the illumination shall be 5 700 K ± 1 000 K. For specific measurements
a different colour temperature may be required. In this case the colour temperature shall be reported
together with the results. Any illuminance level of the test chart may be applied for the texture
reproduction measurement, however the illuminance level in the range from 1 000 lx to 2 000 lx (in
2 2
the case of transmissive chart, from 57,3 cd/m to 115 cd/m for 18 % grey) is recommended when
the measurement has no specified special purpose (e.g. low light performance). Non-uniformity of
illumination on the chart shall be less than 10 %. The light source(s) should be positioned to provide
uniform illumination and produce no glare or specular reflections from the target. A flickering light
source is not recommended as it may cause banding artefacts to occur in the captured image. In the
case of using a flickering light source, although it is not recommended, the exposure time shall not be
shorter than one period of the flickering to minimize the banding artefacts caused by the light source.
4.4.4 Camera settings
The exposure should be adjusted to give the output value for the background grey near the centre of the
measured chart to be the value corresponding to the input value defined below in the camera’s output
colour space.
(input luminance for grey) = (max input luminance) × (grey reflectance)
For example in the case of an sRGB camera, the output Y for 18 % grey is 118 (8-bit) (see IEC 61966-2-1
and its amendment for details).
The deviation of the exposure should be between +5 % and −10 % of the aforementioned target
exposure. For example, when the output colour space of the camera is sRGB 8-bit, the mean output luma
Y value for the 18 % grey should be 118 (8-bit) +2, −6.
The exposure shall be in the range of the aforementioned deviation when the measurement results of
multiple cameras are compared. (Annex B explains the basic concept for this stipulation.) The exposure
may be adjusted by the exposure bias setting of the camera, or by adding a white or black card to the
test chart.
White balance should be adjusted to render the centre of the image as neutral as possible.
© ISO 2019 – All rights reserved 3

---------------------- Page: 8 ----------------------
ISO/TS 19567-2:2019(E)

The focusing shall be in the best practically attainable focus.
For a camera with user selectable compression ratio (e.g. JPEG), the compression ratio should be
minimum to minimize the artifacts of compression. The texture reproduction for raw DSC image data
should be measured for the output of the software that converts raw DSC image data into the final
images. The name and the setting of the software shall be reported with the results of the measurement.
Other settings, such as “sharpness”, “noise reduction”, shall be in the default mode (factory shipping
condition) if those settings are not reported.
4.5 Test Chart
4.5.1 General
“Test chart” shall be “Dead leaves chart with markers” dominating less than a quarter and greater than
a fifth of the image height surrounded by an 18 % background. The field of view of the camera shall be
covered by the 18 % grey background. The dead leaves chart (also known as spilled coins chart) consists
of circles with a random size and colour laid on top of each other. See Figure 3 for a representation of
the dead leaves chart.
A vector file of the recommended reference texture chart in EPS format, approximately 8 000 × 8 000
pixels in size is available at:
http:// standards .iso .org/ iso/ ts/ 19567/ -2/ ed -1/ en
The chart can be printed in various sizes, resolutions and contrast versions. However to reach
comparable results the following instructions shall be used when capturing the chart. In order not to
measure the resolution of the chart but the resolution of the camera the smallest circles in the chart
shall be significantly smaller than the pixel of a camera it is projected on. The largest circle in the
image shall cover at least 50 pixels in diameter. The contrast of the chart should be ±9 % around 18 %
reflectance.
Additional charts at different mean grey and contrast levels may be used in addition to the above
mentioned 18 % ± 9 % grey one.
The size of the Dead leaves with markers should be square and it should cover at least 350 × 350 pixels
of the camera under test. To fulfil this and the image height requirement the number of pixels of the
camera under test shall have more than 1 400 × 1 400 pixels. In case the camera under test has fewer
pixels the height can be larger than a quarter of the field of view which shall be reported together with
the results. For cameras with very low distortion it may cover the whole image but generally it should
only cover a quarter of the image height to minimize potential distortion for registering the image with
the original structure. The pattern can also be smaller and integrated into multi-purpose charts. It is
recommended to place it close to the centre of the field of view to avoid lens performance related issues.
The chart shall be printed in a size that the actual printing resolution (not just the smallest printed
dot) is at least double the camera sampling rate under the above mentioned chart and image height
requirements. The provided sRGB encoded image shall be printed in a colour managed way to ensure
the correct tonal and colour range.
4.5.2 Chart generation
The procedure for generating a list of circles for the dead leaves target is fairly straightforward and
is in part outlined in Reference [10]. It is based on an occlusion model, with circles generated with a
uniform distribution in digital level in the range between 0,09 and 0,27 of the maximum digital output
level for each colour channel in a linear (in reflection or transmission) encoded image. A probability
2
distribution should be selected that achieves a power spectral density that closely approximates 1/f ,
which makes the chart scale-invariant. This can be accomplished with circle radii chosen according to
3
a 1/r probability distribution, although there shall be an upper and lower bound for the radii (r and
min
r respectively) to avoid full coverage by the smallest or largest circles, as discussed in Reference [10].
max
4 © ISO 2019 – All rights reserved

---------------------- Page: 9 ----------------------
ISO/TS 19567-2:2019(E)

Typically, r is chosen such that the image size of the smallest circles will be much less than the image
min
pixels when the target is projected onto the camera sensor array; r is selected to be much larger
max
than r , but generally not larger than the width of the dead leaves image in the chart, W. The centre
min
of each circle is randomly placed on a square canvas of width W + 2r ; during the actual rendering
max
(printing) phase it is assumed that only the centre W × W area is reproduced. Circles whose entire area
falls outside of the inner W × W image are obviously excluded from the final list.
Figure 3 — Dead leaves chart with markers (central part of the test chart)
5 Analytical approach
5.1 General
Texture reproduction is measured from an image of the dead leaves target as described in this technical
specification. The frequency characteristics of the texture reproduction are measured for various
frequencies using a full reference method with the known spatial information of the pattern to obtain
the complete transfer function.
To be able to neglect the local influence of geometric distortion on the target, the test pattern shall
cover less than one quarter and more than one fifth of the image height and shall be located as close to
the image centre as possible.
The dependency of the evaluation on the distortion has been tested for distortion levels of the Dead
Leaves chart up to 3 % local geometric distortion according to ISO 17850. Any significant distortion
above 3 % caused deviations in the measured spatial frequency response curve. Therefore this
measurement method is only valid for distortion levels of the Dead Leaves chart up to 3 %. Distortion
levels above that would need to be corrected prior to performing this analysis but be aware that
distortion correction may affect the SFR measurement. In the case that distortion correction is
© ISO 2019 – All rights reserved 5

---------------------- Page: 10 ----------------------
ISO/TS 19567-2:2019(E)

performed for the measurement; the notification that distortion correction is performed for the
measurement shall be reported with the measurement result.
NOTE Some cameras perform internal distortion correction without notification to the user. In these cases
the correction is treated as part of the camera characteristic.
5.2 Measurement method
The measurement is performed as follows.
— Step 1: The image data shall be linearized using either the tone curve information of the output
colour space or a dedicated OECF measurement as stated in ISO 14524 and be reduced to a single
luminance channel Y. Also the reference data shall be reduced to a single luminance channel Y. The
factors shown in Formula (1) are consistent with the sRGB standard IEC 61966-2-1. They are slightly
different from other ISO TC 42 standards but more recent and more accurate.
YR=+0,,21260 71520GB+ ,0722 (1)
— Step 2: The corners of the dead leaves target shall be located with at least half a pixel accuracy using
the four markers placed in the corners of the target.
— Step 3: The vector version of the reference data shall be transformed projectively as described in
Reference [11] to fit the image data using the corners of the target as located in the image data in step
2 to calculate the transformation matrix [M] [see Formula (3)] by solving the matrix shown in
proj
[11]
Formula (2) (u |v ) are the coordinates of the target vertices in the image and (x |y ) are
1−4 1−4 1−4 1−4
the corresponding coordinates of the reference vector data vertices. The projectively transformed
vector data of the reference is then converted into an uncompressed or lossless compressed
reference image with the same pixel count as the test image using a raster image processor.
In case of using the Qt open source platform the vector-raster conversion is performed the following
way: The class QPainter is needed for drawing the circles, choosing a brush (Qt::solidPattern) and
a pen (Qt::SolidLine) with a pen width of 0,30 pixels. The resulting reference image can be saved by
using the class QPixmap.
u  xy 10 00 −−xu −yu  a
1 11 11 11
     
v 00 01xy −−xv yv
b
 
1  11 11 11   
     
u xy 10 00 −−xu yu c
2 22 22 22
 
   
     
v 00 01xy −−xv y vv d
2 22 22 2 2
 
   
= × (2)
     
u xy 10 00 −−xu yu
e
3 33 33 33
 
   
v 00 01xy −−xv yv f
     
3 33 33 33
 
   
u xy 10 00 −−xu yu g
 4   44 44 44  
   
 
v 0 000 xy 1 −−xv yv
h
4 44 44 44  
   
ab c
 
 
M = de f (3)
 
   
proj
 
gh 1
 
— Step 4: For further processing, image and reference shall be cropped square so that the image edge
m
length is 2 pixels, where m is an integer, and the reference shall be cropped to the matching image
area and not contain any parts of the marker. The cropped square image area with i × i pixels
max max
shall cover the largest area possible of the dead leaves chart excluding markers. In addition, the
respective mean value Y is subtracted from both, image and reference [see Formula (4)]. To avoid
leakage, a window function [Formula (5)] with the parameter r = 0,25 is applied to the crops,
strongly decreasing from the edge to prevent losing too much image information.
6 © ISO 2019 – All rights reserved

---------------------- Page: 11 ----------------------
ISO/TS 19567-2:2019(E)

YY=−Y
im im im
(4)
YY=−Y
refref ref
1 r

2π
12+−cos/xr 0≤  
{}()
  
r
2 2

 r r
wx()= 1 ≤ 
2 2


1 r
2π
11++cos x −rr/2 1−≤x<1

()
{} 
r
2 2

with x being the horizontal normalized coordinate of the pixel around which the window is created
with the pixel count ranging from 0 to i -1 normalized by i . The same has to be applied to y
max max
which is the normalized vertical coordinate to get to a two dimensional window.
— Step 5: The signals of image Y( f , f ) and reference target X( f , f ) shall be calculated using the
h v h v
Fourier transformation on the mean-corrected and windowed 2-D data. Based on Y( f , f ) and X( f ,
h v h
f ), the cross power density ϕ ( f , f ) of target and image and the auto power density ϕ ( f ,
v YX_org h v XX_org h
f ) of the target are calculated as described in Formula (6):
v
*
φ ff,,= Xf fX⋅ ff,
() () ()
XX_orgh vh vh v
(6)
**
φ ff,,=Yf fX⋅ ff,
() () ()
YX_org hv hv hv
— Step 6: A smoothing step on the cross power density ϕ (f , f ) and the auto power density ϕ
YX_org h v XX_
(f , f ) shall be performed by transforming them back into the spatial domain, then applying a
org h v
narrow window [see Formula (7)]. The results are then transformed into the frequency domain again.
2ππn 4 n 6πn
wn =−aa cosc+aaos + cos 01≤≤nN−
()
() () ()
01 2 3
N N N
with thecooefficients:
a =0,35875
0
(7)
a =0,48829
1
a =0,14128
2
a =0,01168
3
with n being the pixel coordinate around which the window is created and N being the width/
height of the window.
— Step 7: To
...