IS WEBCAM PERFORMANCE SUFFICIENT FOR THE
INVENTORY CONTROL OF INDUSTRIAL WHOLESALE ITEMS
WITH NO CUSTOMER INVENTORY BALANCE RECORDS?
Case: Technical Wholesale Items
Ari Happonen, Erno Salmela
Department of Industrial Engineering and Management, Lappeenranta University of Technology
Skinnarilankatu 34, Lappeenranta, Finland
Jukka Nousiainen
Department of Information Technology, Lappeenranta University of Technology
Skinnarilankatu 34, Lappeenranta, Finland
Keywords: Remote monitoring, Warehouses, VMI, Inventory management, Machinery Industry.
Abstract: This research considers the technical performance of modern webcams for the remote monitoring of the
inventory balance of industrial wholesale items. In this case study suppliers were technical wholesalers and
the customers were from machinery industry. Paper presents a study on the suitability of webcams for the
remote monitoring. In order to establish a remote monitoring system, the images must meet certain quality
criteria, thus enabling the assessment of inventory levels. Paper presents image quality tests performed on a
webcams in varied conditions, and compares the results to the Finnish Illuminating Engineering Society’s
lighting regulations for industrial work. The quality tests aimed to ascertain whether the image quality was
sufficient in typical industrial conditions. The results indicate that the image quality of modern webcams is
sufficient for remote monitoring within the limits set by the distance of the camera from the objects and by
lighting conditions. However, according to the tests, the technical performance of surveillance cameras on
the market five years ago, taking budget constraints into consideration, was not adequate for monitoring.
1 RESEARCH MOTIVATION
The studied cases utilize the VMI (Vendor-Managed
Inventory) model, which is inventory replenishment
carried out by the wholesaler delivering items to
customer in accordance with the agreed management
models. For customer items could play a critical role
because their absence may stop entire production
lines. Shortages must be avoided and usually this is
achieved by using large stock buffers and frequent
visual assessment based inventory checks for
customer’s replenishment needs. To reduce
customer site visits a reachearch on camera based
remote monitoring was started for at least visual
inspections level of inventtory balance information.
Wholesalers have recently promoted VMI
research on the use of technology in remote
monitoring of items (Happonen and Salmela, 2007).
But still there are many items which cannot be
monitored easily (cost and space efficiently).
Typical reason for limited use of technology in
inventory management services is that tehcnology is
concidered to be too expencive compared to manual
work. The objective of this research is to ascertain
whether it is possible to produce images that reveal
the inventory levels of items (order point data) in
Finnish industrial facilities. As the research of VMI
concentrates on collaboration and forecasting
(Vigtil, 2006; Salmela and Happonen, 2007; Holweg
et al. 2005) as a way to improce order-delivery
process of the supply chain and as the information
exchange is not as easily arranged between different
supply chain parthners as it should be, this research
study considered the capabilities of the modern day
web cameras as a information source of the item
balance information on the VMI relationship.
507
Happonen A., Salmela E. and Nousiainen J. (2008).
IS WEBCAM PERFORMANCE SUFFICIENT FOR THE INVENTORY CONTROL OF INDUSTRIAL WHOLESALE ITEMS WITH NO CUSTOMER
INVENTORY BALANCE RECORDS? - Case: Technical Wholesale Items.
In Proceedings of the Fourth International Conference on Web Information Systems and Technologies, pages 507-511
DOI: 10.5220/0001530305070511
Copyright
c
 SciTePress

2 INDUSTRIAL CONDITIONS
AND CAMERAS
Industrial facilities do not offer optimal environment
for the technical equipment. Dust, humidity and
temperature in the facilities could cause problems,
but usually these may be prevented by a casing etc.
In the industrial facilities in this study, the storage
facilities adjacent to the production area were quite
clean. In addition, storage areas were mostly
relatively clean too. The camera usability problems
lie more on illumination and focal length areas.
2.1 Illumination in Storage Spaces
With regard to the quality of the images, one of the
most important factors is illumination. The Finnish
legislation does not defined any minimum values for
industrial illumination levels, but the adequacy of
illumination in work spaces may be evaluated e.g.
based on publication 9 of the Illuminating
Engineering Society of Finland in 1986:
Valaistussuositukset, sisävalaistus (Lighting
recommendations, indoor lighting). The illumination
levels recommended for storage conditions are
presented in Table 1. Levels which exceed 50% of
the recommendation are considered sufficient. Based
on field research general lighting in modern
production plants, exceeds the recommended limits.
However, the lighting in warehouses varied more.
2.2 Technical Features of the Camera/
Camera Technology
If the lighting conditions are poor, the picture is
taken with a long exposure and/or increased
exposure sensitivity. A long exposure and increased
sensitivity create a random noise in the image. This
noise may be technically removed to some extent,
but due to such filtration, the sharpness of the image
will suffer (McClelland and Eismann, 2003).
If the image quality is not sufficient for
assessment of inventory levels and the quality
cannot be enhanced through post-processing an
efficient monitoring system is not possible. The
factors of the image quality are optics, the camera’s
imaging sensors, and physical imaging conditions
(e.g. illumination). Optics focal length may become
a problem since webcams are typically designed for
access monitoring, where as wide an angle as
possible is needed (Hedgecoe, 1992). So the
interrelationship of the camera angle and image
quality may be unfavourable to remote monitoring.
2.3 Capturing an Image and the
Importance of the Time of Day
Capturing images during the day is usually difficult
due to “disturbances” caused by people and
machines. Therefore, images should be captured
outside the customer’s regular working hours. For
instance, if the image is captured at night, no
disruptions are caused by people or machines except
when people work in three shifts. On another hand,
the amount of lighting may be insufficient to take
the pictures at the night time. Fortunately at least
some modern cameras enable the remote control of
external equipment (such as light fixtures), which
helps to avoid such problems.
2.4 Analysing the Image
Classifying items inventory balance category level
from images and placing orders compose an
essential part of a camera monitoring system. The
label or mark that indicates the item “count” and
order point must be identified either manually or
automatically. Analysing the image on a computer
and marking the appropriate order point would allow
the implementation of a system in which the
computer places orders independently.
Table 1: Illumination Recommendations for Storage
Spaces by Finnish Illumination Engineering Society.
Space
and
usage
Illum.
(for
work)
Illum.
(general)
Notes
Small
items on
shelves
300 lux 200 lux
Localised light
sources are
recommended
Medium
size
items on
shelves
200 lux 200 lux
Light sources
should be
between the
shelves not over
a shelf
Large
items on
shelves
100 lux 150 lux
Corridor
spaces
150 lux 150 lux
WEBIST 2008 - International Conference on Web Information Systems and Technologies
508

3 CAMERAS AND TEST GOALS
Image quality tests were conducted in laboratory
environment on a modern camera (Axis 207MW),
and on one that represents technology that dates a
few years back (Axis 2120) from the Axis
Communications producer of surveillance
equipment for industry. New camera operates on a
¼”, 1.3 megapixel CMOS sensor, which produces a
1280x1024 pixels image. The lens has a 3.6 mm
focal length and a f1.8 luminous intensity. Based on
the sensor and focal length, Figure 1 presents size of
seen area on pictures at different distances from the
object. The previous camera generation is
represented by the Axis 2120 webcam from the 2001
product line. Its highest resolution is 640x480 pixels.
0,0
2,0
4,0
6,0
8,0
10,0
Distance
Width (m)
1,3 2,7 4,0 5,3 6,7 8,0
Height (m)
123456
123456
Figure 1: Visible area on picture related to the distance on
the object from the Axis 207MW camera.
3.1 Quality Test Arrangements
The test images were captured at four different
distances and illumination levels. Illumination in the
tests was adjusted from 15 to 200 lux where the 200
lux is minimum level recommended by Illuminating
Engineering Society. This range covered
illumination levels from the worst case to the
recommended levels. Target boxes were placed at a
230 cm distance from each other, and they were
tilted at a 30° angle compared to the horizontal level
and cameras were 120 cm over box level. Items
were positioned as if they were in a typical industrial
environment. Black labels were attached to the
targets to indicate the order point. Black was chosen
as it creates good contrast on many backgrounds.
3.2 The Goal of the Measurements
The technical image quality was analysed by
measuring the noise in the pictures. The sufficiency
of the image quality for monitoring was visually
estimated for a typical machinery industry case.
The measurements aimed to shed light the
following points:
 The effect of illumination on noise
 Image quality in industrial lighting conditions
and affect of distance to image quality
 The maximum usable distance in the
recommended lighting conditions
 The adequacy of the resolution for identifying
the item inventory balance category
4 ANALYSIS OF THE
MEASUREMENTS
The images were analysed, and the sufficiency of
their quality at different parameters was checked.
4.1 Noise Measurements and Results
The images were analysed in accordance with the
ISO 15739 standard at 12 different points to analyze
the vertical and horizontal noise. Figure 2 present
measurements of the AXIS 207MW and depicts the
noise levels for each measurement point which relate
to the darkness of the point; number 1 is the darkest.
The vertical axis describes the amount of noise and
each illumination level is presented as data series. In
18 lux lighting, especially the darker measurement
points, there is considerably more noise than in other
illumination levels, but in other illumination levels
results are similar to each other.
0,00
0,50
1,00
1,50
2,00
2,50
3,00
3,50
4,00
123456789101112
Measurement point
Noise
18 lux
82 lux
147 lux
192 lux
Figure 2: Vertical noise.
Figure 3 depicts horizontal noise. Illumination
level of 18 lux clearly differs from the others as
noise level is considerably higher than in other
series. Only the very lightest points had similar noise
than higher illumination levels had.
In Figure 2 and Figure 3 there is an outlier in
measurement point 12 as the camera over exposure
IS WEBCAM PERFORMANCE SUFFICIENT FOR THE INVENTORY CONTROL OF INDUSTRIAL WHOLESALE
ITEMS WITH NO CUSTOMER INVENTORY BALANCE RECORDS? - Case: Technical Wholesale Items
509

the image affecting the measurements of point 12
which was taken account on calculations of the
average noise for the Figure 4 (last values were not
included).
0,00
0,50
1,00
1,50
2,00
2,50
3,00
3,50
123456789101112
Measurement point
Noise
18 lux
82 lux
147 lux
192 lux
Figure 3: Horizontal noise.
Figure 4 shows the noise averages for each
illumination level and indicates that the lowest
lighting level had clearly the highest noise level. The
illumination levels of 82 and 147 lux show no
considerable differences, but the 192 lux level had
slightly less noise than the others.
0,00
0,20
0,40
0,60
0,80
1,00
1,20
1,40
1,60
1,80
2,00
18 lux 82 lux 147 lux 192 lux
Amount of illumination
Noise
Horizontal noise
Vertical noise
Figure 4: Average amount of noise in different
illumination levels.
The Figure 4 shows that the 18 lux level
produces the most noise and in practice the
graininess of the image makes the interpretation of
small details rather difficult. The 82 lux illumination
level had significantly less noise, and the effect of
the noise on the image was not large. A visual
assessment revealed no remarkable difference
between 147 and 82 lux levels, which relates well on
the results of the noise analysis. Improvement was
only minor from 147 lux to 192 level. Based on the
tests, the lighting conditions in industry is
appropriate for obtaining sufficient image quality.
The lighting recommendations of the Illuminating
Engineering Society are good enough for taking
pictures with little noise.
The noise was significantly higher when
capturing images of darker tones than ones of lighter
tones. From image usability point of view it was
interesting that the surrounding environment of the
target had little impact on the noise on target area.
Therefore, even if the environment of the industrial
hall or the frames of the shelves were dark, it would
not considerably affect the interpretation of the
image. Noise in pictures of the actual monitored
items is determined by the tone of the target itself.
As the boxes in industry are often of a dark shade, at
low illumination levels, they do not provide the best
background. But in sufficient lighting of over 100
lux, there is hardly any difference between dark and
light shades with regard to the amount of noise.
However, the amount of noise does not significantly
decrease as lighting exceeds 100 lux, which
indicates that cameras can be used in normal
industrial halls, provided that the lighting conditions
are unvarying and no shades are cast on the target.
4.2 Image Quality
In the quality tests, the image quality, the visibility
of items, amount of the items and order point signals
and marks were visually estimated. The estimation
was based on how well and how efficiently details
could be distinguished from the image. The results
of the analysis are presented in Table 2.
5 FINDINGS
For the Axis 207MW the most appropriate range
proved to be 3 to 4 metres. At these distances, the
width of the view is 4–5.3 metres, which allows
more than one shelf to be monitored. Generally four-
metre distance should be enough for most cases to
allow good placement of the camera. As for the
illumination it had little impact on image quality at
the levels in industry. The older camera, AXIS 2120,
could not produce image quality required for
inventory balance level monitoring. AXIS 207MW,
on the contrary, was sufficient for remote
monitoring both in terms of image quality and other
functions.
6 CONCLUSIONS
This study aimed to establish whether the image
quality of webcams was sufficient for remote
monitoring of inventory levels. A webcam-based
system could be applied to the operative order-
delivery process of technical wholesale items with
WEBIST 2008 - International Conference on Web Information Systems and Technologies
510

no customer records on inventory balance. Although
laboratory tests indicate that the remote monitoring
of small technical wholesale items is doable,
basically monitoring is best suited for the large and
medium sized items. For small items, the system
would require labels or other signals, which would
increase maintenance costs. As camera based
monitoring itself is not easily adopted to all item
sizes and types the camera monitoring would be at
its best as a part of an extensive remote monitoring
system as an expansion of systems capabilities.
Camera monitoring is at its best when items are
consumed sporadically in large volumes. Regular
on-site replenishment inspections meet the
customer's normal replenishment needs, and camera
monitoring assures that individual consumption
peaks do not cause shortages in the meantime. The
webcam system thus helps to increase the efficiency
of operative order-delivery processes through remote
monitoring and replenishment decisions.
7 FUTURE WORK
The correlation between camera monitoring and
replenishment efficiency should be studied from a
financial perspective. Does the arrangement of the
camera and shelves have an impact on the
replenishment efficiency (e.g. installing shelves side
by side and at centralised locations) or do efficient
camera monitoring and replenishment efficiency
contradict?
As this research study was part of the TEMO –
projects research topics (Häkkinen et al., 2007) and
as the result of the scale based system have been
really promising (Happonen and Salmela, 2007) a
new development research project is on preparation
for a integrated remote inventory monitoring system
which will allow many different monitoring methods
to be used side by side to deliver complete remote
inventory monitoring system.
REFERENCES
Bavister, S., 2002. Digitaalikuvaus ja kuvankäsittely,
WSOY.
Elvander, M.S., 2005. A theoretical mapping of the VMI
concept – A literature review. Report ISRN
LUTMDU/TMTD-4016--SE, Lund University. 40
pages.
Happonen, A.., Salmela. E., 2007. Automatic & unmanned
stock replenishment process using scales for
monitoring. In Webist’07, 3rd International
Conference on Web Information Systems and
Technologies. INSTICC Press.
Hedgecoe, J., 1992. The Photographer’s Handbook, Alfred
A. Knopf, New York, 3
rd
edition.
Holweg, M., Disney, S., Holmström, J., Småros, J., 2005.
Supply Chain Collaboration: Making Sense of the
Strategy Continuum, in European Management
Journal, Volume 23, Issue 2, Pages 170-181
Häkkinen, K., Hemilä, J., Uoti, M., Salmela, E.,
Happonen, A., Hämäläinen, H., Siniluhta, E.,
Nousiainen, J., Kärkkäinen, M., 2007. VMI
teollisuudessa – Teoriaa, teknologiaa ja sovelluksia,
VTT, Finland
Illumination Engineering Society of Finland. Publisment
no 9/1986. Valaistussuositukset/Sisävalaistus.
McClelland, D., Eismann, K., 2003. Real world digital
photography, Peachpit Press, Berkeley, 2
nd
edition.
Salmela, E., Happonen, A., 2007. Applicability of CPFR
on Inventory Replenishment Operation Model of Low-
value Items. Finnish Machinery Industry - Case
Study. In Proceedings of the 12th International
Symposium on Logistics (ISL 2007), Hungary
Vigtil, A., 2006. Exchange of advance demand
information in Vendor Managed Inventory. In
Proceedings of NOFOMA 2006.
Table 2: Image usability for inventory balance detection in different distances.
Range AXIS 207MW AXIS 2120
2 m Under 20 lux; some problems to “see” the amount
of items, because of shadows by boxes on top.
Other illumination levels are ok.
Under 20 lux; the image is not clear enough for
inventory monitoring. Higher illumination levels have
lower noise levels, but low resolution is a problem.
3 m
Under 20 lux; image inspection starts to be too
slow process. Other illumination levels are
problem free.
Under 100 lux; image not clear enough for inventory
monitoring. Higher illumination levels have lower
noise levels, but low resolution and distance from the
objects makes image inspection almost impossible.
4 m Under 20 lux; for industrial usability more light is
needed, 100 lux should be sufficient.
Low resolution and “high” distance limits AXIS 2120
in over 3 metres distances.
5 m Under 100 lux; problems because of shadows by
boxes on top. Maximum range for the resolution.
Produces usable images in over 100 lux levels.
As above.
IS WEBCAM PERFORMANCE SUFFICIENT FOR THE INVENTORY CONTROL OF INDUSTRIAL WHOLESALE
ITEMS WITH NO CUSTOMER INVENTORY BALANCE RECORDS? - Case: Technical Wholesale Items
511