You are on page 1of 14

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230717890

The Visual Inspection of Product Surfaces

Article  in  Food Quality and Preference · August 2012


DOI: 10.1016/j.foodqual.2012.08.006,

CITATIONS READS

12 6,577

3 authors:

Nathalie Baudet Jean Luc Maire


Université Savoie Mont Blanc Université Savoie Mont Blanc
14 PUBLICATIONS   73 CITATIONS    76 PUBLICATIONS   299 CITATIONS   

SEE PROFILE SEE PROFILE

Maurice Pillet
Université Savoie Mont Blanc
220 PUBLICATIONS   931 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Industrial Performance View project

Inertial Tolerancing View project

All content following this page was uploaded by Maurice Pillet on 14 February 2019.

The user has requested enhancement of the downloaded file.


The Visual inspection of product surfaces
Abstract
Many companies practice the human visual inspection of their production. In some cases,
this inspection takes the aesthetics aspects of the product surface into account. However, only
a few companies have developed methods for this type of inspection. In this article, how a
sensory analysis test can be applied for the visual inspection of product surface is described.
A synthesis of the specifications for an effective visual inspection of products with high added-
value is detailed. These specifications can also be adapted for any type of product, including
food products.

1. Introduction

The consumer’s perception of the quality of a product depends largely on the quality
of its surfaces. It is particularly the case for luxury products, but has nowadays become so for
an increasing number of products. For food products for example, the texture (Pereira et al.,
2006) or the appearance (Munkevik et al., 2008) significantly influence what the consumer
purchases. In the automotive industry, the tactile perception (Giboreau et al., 2001) and the
haptic perception (Soufflet et al., 2004) of the product surface are important in consumer
choice. Obviously, the visual perception of the surface is also very important (Creusen and
Schoormans, 2005). To ensure that each manufactured product meets the expected visual
characteristics, a visual inspection must be carried out.
The bibliography concerning the visual inspection methods is very important. It
includes the inspection conditions (Gaaras and Pomplun, 2008; Garret et al., 2001, Jebaraj et
al., 1999), the training techniques (Rao et al., 2006; Chabukswar et al., 2003; Rebsamen et al.,
2010; Nickles III et al., 2003), different ways of controlling (Lee et al., 2009; Schütte et al.,
1999; Gilden et al., 2010) or methods to detect defects (Zamuner, 2011; Hassan and Diab,
2010).
However, visual inspection is often described as the way to detect a product’s
functional anomalies (for example, the specific color of a fruit which gives one the indication
of its ripeness). This said, visual inspection also sometimes includes also an aesthetic
objective (for example, the specific color of a fruit which does not seduce the consumer).
Sometimes, even the two objectives (functional or aesthetic) are complementary. For
example, the extra carbon dioxide adds brightness to a white wine and increases its longevity.
Thus, brightness can be inspected to ensure that the product will keep its functional properties
over time. Since the intensity of the brightness is also a significant aesthetic criterion in
consumer choice, visual inspection can also be carried out to ensure that the product meets
aesthetic requirements. “The success of the food industry depends on the consumer’s
continuing confidence that the appearance of the product is a true indicator of the subsequent
acceptability of the eating quality of the food” (MacDougall, 2003).
The visual inspection is often carried out by one single inspector who assesses the
quality of the product by referring to either a set of standard products or to his own
experience. When he detects an anomaly, he has to scrap the product. However, in some
cases, he has to decide if the anomaly is critical or not. He needs to evaluate the intensity of
this anomaly and its impact on the quality of the product. Unfortunately, this process is poorly
formalized (Debrosse et al., 2010), particularly in the case of aesthetic objectives. The
evaluation of the anomaly is very subjective because it depends on the inspector’s level of
knowledge, his know-how and his perception of the importance of the anomaly.
Therefore, a method must be developed to reduce the variability in the results of visual
inspection for any product (if possible). It is the main aim of our research program, called
INTERREG IV, which brings together the University of Savoy and Lausanne Federal
Polytechnic School (EPFL) and several Swiss and French companies. Figure 1 gives some
examples of the products whose visual inspection must be improved (S.T. Dupont and
Fournier).

Figure 1 Visual inspection in S.T. Dupont and Fournier

The method that we propose, and that is described in this paper, is based on the use of
sensory analysis methods.

2. Visual inspection of surface appearance

2.1. Different types of surface deviation

For a visual inspection with an aesthetic objective, the process used to decide if the
product complies is much more complex than for a functional objective. The nonconformity
on an aesthetic criterion can be defined by an unacceptable deviation between an ideal part
and the real part. We have defined three types of deviation:

- A deviation from a reference: it is expressed by differences in the characteristics of


the product/component (e.g.: color, texture, etc.)

- A deviation from the intent of the designer: it is expressed by the differences in the
configuration of components, from one component to the one next to it (e.g. an
contrast in color between two components)

- A local deviation: it is expressed by the presence of an anomaly: mark, pollution,


heterogeneity or distortion (e.g. a slight dent on the surface).
During visual inspection, the inspector has to detect a deviation in order to identify
which type it is, then describe it and finally quantify its intensity. Sensory analysis can be
used to help him. Sensory tests have proved useful in the food industry to express the
differences in human perception (taste, touch, hearing, smell and sight) in an identical way as
if one was using a measuring instrument. Sensorial analysis is classically used to develop new
products or to compare products with those of the competitors. It can also be used as a quality
control tool to check sensory characteristics. Although we did not find any examples of the
use of sensory analysis in the visual inspection of products in previous research, this method
seems to be perfectly suitable to describe a deviation on a surface and to quantify its intensity.
2.2. Sensory tests

There are two main groups of sensory tests: analytic and hedonic. Hedonic (or
affective) tests are intended to identify the consumer preferences for two or more products
and to give subjective information about how well these products are likely to be accepted
(Lawless and Heymann, 2010). Analytical tests are intended to identify and/or evaluate
differences between two or more products.
As said previously, the purpose of visual inspection is to assess differences and not
preferences. Visual inspection is therefore only used for analytical tests. These tests can be
divided into two subgroups (Depledt, 2009):
- Discrimination testing which is used to determine the probability that a very slight
difference can be perceived (triangular test, duo-trio test, two out of five test,
paired comparison test, n-Alternative Forced Choice method, A-Not-A test,
Sorting method, …) .
- Descriptive analysis which is used to quantify a difference when it is undoubtedly
perceived. There are three main methods associated with this analysis: ranking,
scaling and profiling.
In visual inspection, the deviation is generally clearly perceived. Visual inspection is
thus only used for descriptive analysis. Table 1 details the characteristics of each method.
Table 1 Tests in sensory descriptive analysis
Method Dimension Aim

Mono Attribute To classify three or more simultaneously presented products


Ranking with regard to a specific attribute (e.g. sweetness, hardness,
Multi products etc.)

Mono Attribute To evaluate one or more products (not necessarily


Scaling simultaneously presented) on a scale of intensity with regard
Mono/Multi products to a specific attribute

Multi Attributes To evaluate all the attributes of one or more products (not
Profiling necessarily simultaneously presented) in order to build up a
Mono/Multi products sensorial profile

Ranking methods cannot be used for visual inspection because the difference between
products is expressed using relative values. However, these methods can be used during
inspector training in order to verify the capacity of inspectors to perceive and rank a
difference in products according to a sensory attribute.
Scaling methods can be used for visual inspection in the case of a deviation from a
standard (e.g. a difference between the color of the controlled product and the color of the
model product) and in the case of deviation from the intent of the designer (e.g. a difference
between the color of two adjacent components of a product). In the case of local deviation,
these methods can be used if the difference is perceived on the same design (e.g. a scratch on
a mono-color product).
When a local deviation is perceived by the inspector, it is not possible to compare this
product with another anomaly-free product, or to compare the perceived anomaly with
another one. That is especially true when the inspection applies to several types of products
whose shape designs and colors are different. In this context, the same anomaly seen on two
different products, or the same anomaly in two different places on the same product, will not
necessarily lead to the same decision. For example, Figure 2 shows how the same scratch
(same depth, width and length) may lead to different inspection results. In cases ‘a.’ and ‘d.’,
the product is accepted (the scratch is located respectively in the direction of the décor and
close to the edge of the product). However it will be refused in cases ‘b.’ and ‘c.’.

a. Parallel to the decor b. Across the decor

c. In the middle of a polished decor d. Close to the edge


Figure 2 The same anomaly in two different decors

In the case of local deviation, a prior characterization of the anomaly is therefore


necessary. Profiling methods can be used, not to obtain a sensorial profile of the product, but
to obtain a sensorial profile of the anomaly detected on the product (i.e. the characterization of
the anomaly using a set of attributes). The question is: “How does one know which attributes
can be used to profile an anomaly?”

3. Sensory profile of an anomaly

3.1 Types of anomaly

Before determining the attributes of a given anomaly, it is necessary to clearly define


what can be considered as an anomaly.
The eye of an inspector is able to identify shapes, colors and edges. Therefore, he
detects an anomaly when he perceives a local deviation of color, edge or shape. For this local
deviation, two categories of anomalies can be identified: the progressive anomalies and non-
progressive anomalies. A progressive anomaly has an intensity that will evolve over time and
that will be critical for the customer (e.g. corrosion detected during control which will spread
significantly over time). The detection of a progressive anomaly during visual inspection
leads the inspector to reject the product immediately. On the other hand, a non-progressive
anomaly retains the same intensity (e.g. a scratch detected during the inspection the length of
which never develops over time). In this case, the inspector has to decide whether the piece
should be rejected or not. Sensory profile methods are therefore only used for non-progressive
anomalies.
It is obviously impossible to identify all of the anomalies that can occur on a product
(even if some companies try to list all the anomalies that inspectors could find). For example,
are a 1-mm scratch and a 2-mm scratch two separate anomalies? Or are they the same
anomaly? We therefore decided to list all the different types of anomalies rather than all the
possible anomalies. In Guerra et al. (2009), we listed four types of non-progressive anomalies:
marks, stains, deformations and particles. More generally, for visual inspection, all anomalies
can be classified as:
- Mark: something that damages the surface (scratches, scuffs, dent, etc.).
- Heterogeneity: anything that makes the surface lose its homogeneity (a stain, a
color difference, etc.).
- Pollution: anything undesirable that is added to a surface (a hair, a dirt particle, a
black stain, etc.).
- Distortion: anything that changes the shape of the surface (an overly-polished
surface, non-regular line of light, etc.).
The advantage of this classification is to minimize the vocabulary used to describe
anomalies. Another advantage is that it can be used for all types of products.

3.2 Anomaly attributes


If possible, anomaly attributes must be chosen in accordance with NF ISO 11035. This
standard defines an attribute like "the term referring the subject to an element of product
perception" (ISO, 1995). It must have properties such that an assessment can be made using a
scale of intensity and must meet a number of principles, including that of being relevant
(suitable for describing anomalies) and accurate (which can be understood by the inspectors).
Attributes must also be discriminating (allowing one to differentiate between the anomalies)
and, wherever possible, independent of each other (each covering different aspects of the
event).
When an inspector controls a product, he focuses his attention firstly on the anomaly,
and then on its environment. To build up the list of attributes, we therefore consider:
- The factual description of an anomaly, which refers to how to describe intrinsic
characteristics of the anomaly
- The context of the anomaly, which includes the local context and general context
(how the anomaly is perceived in its overall environment)
The factual description includes two parameters:
a. The viewing conditions, which refer to the conditions in which the anomaly is
perceived, and which regroups five attributes:
o Distance or magnification: it characterizes the distance from which the
anomaly is perceived. This attribute depends on the product being
controlled (e.g. an anomaly visible from up to 1 meter for furniture or an
anomaly only visible using a magnifying glass for a watch component).
o Orientation or light effect: it characterizes the viewing angles one has on
the anomaly. Figure 3 shows the movements that the inspector has to do
during visual inspection. Three distinct phases occur during the movement,
called “light effects” (Guerra, 2009):
• Black effect: the light beam arrives perpendicular to the part surface
and this is reflected perpendicularly. The inspector perceives a matt
surface (Figure 3 a.).
• Shiny effect: the incident light beam returns a reflected beam so that the
subject is placed in the path of the beam. The inspector is dazzled by
the part surface (Figure 3 b.).
• Mirror effect: the incident light beam reaches the surface tangentially.
The inspector sees his reflection in the part surface (Figure 3 c.).

a. black effect b. shiny effect c. mirror effect


Figure 3. Light effects

o Light intensity: the visibility of the anomaly depends on the light intensity
(in the control room: 1000 lx recommended; 200 lx, etc.).
o Time: it represents the time to find the anomaly during the detection period
(e.g. visible immediately or visible after a thorough scan)
o Direction: it characterizes the orientation of the part in which the anomaly
is visible (e.g. the anomaly is visible only when the part is oriented
vertically).
Among the five attributes related to viewing conditions, time is the least
reproducible. It can vary from one inspector to another according to the
detection procedure and also sometimes to the “luck” in finding the anomaly.
The light intensity attribute is hardly used because the workstations are rarely
equipped to be able to change the light intensity in a uniform and constant way.
So during the detection and evaluation steps of the product, the light intensity
is often constant (for the partners of the INTERREG program, the intensity is
fixed at 1000 lx).
The use of distance attribute depends on the size of the anomaly in relation to
the company’s standard. In the watch industry for example, the use of a
magnifying glass is common and the anomaly will be more critical if it is
visible with the naked eye. Sometimes, the distance is fixed and the inspection
is done from a standard distance.
On the other hand, the orientation is an attribute that is often used to
characterize the viewing conditions of the anomaly.
b. The anomaly description, which refers to all its physical characteristics;
includes two attributes:
o Size (e.g. length, width, thickness, etc.).
o Shape (e.g. round, elongated, etc.).

The context of the anomaly includes two parameters:


a. The local context, which refers to how the anomaly is perceived in its
immediate environment, itself including two attributes:
o Shape contrast: it characterizes the shape of the anomaly relative to other
shapes (c.f. figure 2 above).
o Color contrast: it characterizes the color of the anomaly relative to other
colors in the decor (e.g. an anomaly whose color is very different from the
other colors).
b. The general context, which refers to how the anomaly is perceived in its
overall environment (e.g. the anomaly is visible whatever the final
configuration of the product).
All identified attributes can be used for any type of anomaly (mark, pollution,
distortion or heterogeneity). However, the definition of the level of each attribute’s intensity
depends on the type of anomaly and the level of quality reached by the company.
3.3 Types of standard

The choices of attributes to be used for visual inspection, as well as the definition of
each level of each attribute, are very important. However, it is also important to choose how
the company wants to share these definitions. Three main types of standards can be used.
The choice of the standard depends on the product and on the characteristics of the
product to control (Costell, 2002):
- Product standard: it includes acceptable variation limits for each sensory attribute
of the ingredient or raw material used for the product. This standard offers the
advantage of being easily obtained, maintained and reproduced.
- Mental standard: it is developed by one or more experts who define the desired
level of the sensory characteristics of the product to be reached. These experts
have demonstrated their ability to recognize and evaluate the sensory properties of
the product.
- Written standard: here, it is the written definitions of a given critical attribute that
drive consumer acceptance. It includes a definition of the key attributes, the
perceptible variations of which depend on the raw material and on the
manufacturing process, the common defects and unacceptable characteristics.
In the case of visual inspection whose purpose is aesthetic, the product standard is
often difficult to use. This is especially true when the results of the evaluation of the anomaly
depend on its position on the product (Figure 2). It is indeed difficult to build up a collection
of products (or pictures of products) which reflects all the possible product anomalies. In
addition, even if it can be built, this collection is difficult to maintain over time (the
management of this collection can, for example, cause other anomalies). When the product
changes over time (e.g. a perishable product), it is impossible to maintain a product standard.
The mental standard is commonly used in visual inspection. Experts using this
standard are often appointed from among marketing managers and/or quality managers. Their
choice often depends much more on their hierarchical position than their ability to perform a
sensory evaluation of products. ISO 8586-2 (1994) clearly specifies what skills and qualities
required for experts to form the mental standard (Table 2).
Table 2 Types of assessors in sensory analysis (ISO, 1994)
Type of assessor Definition Characteristics

Assessor chosen for his/her ability to


Select assessor
perform a sensory test

Selected assessor with a high degree of Good consistency of judgments, both within a session
sensory sensitivity and experience of and from one session to another
Expert assessor sensory methodology, who is able to make
a consistent and repeatable sensory Good long-term sensory memory
assessment of various products

Expert assessor who has additional Extensive experience in the relevant specialist field.
experience as a specialist in the product
and/or the process and/or marketing, and Highly-developed ability to recognize and evaluate
who is able to perform the sensory analysis sensory properties.
of the product and to evaluate or predict the Mental retention of reference standards.
Specialized effects of variations relating to raw
expert assessor materials, processing, etc. Recognition of key attributes.
Deductive skills which may be applied to problem-
solving.
Good ability to describe and communicate conclusions
or to take appropriate action.

Reducing the variability of the results of a visual control therefore requires the
selection of experts within the knowledge of sensory analysis and their training with the
criteria and attributes used to create the mental standard. The limited use of mental standards
is due to the difficulty in ensuring that all the experts have the same concept of quality, i.e.
using the same criteria to assess quality (Costell, 2002). This limitation is even more
important given that the standard must sometimes be shared by several people (about one
hundred people, between experts and inspectors).
The difficulties associated with the use of a product standard and a mental standard
can be reduced with the use of a written standard that will contain the definition of each
attribute and each criterion. The use of a written standard has the undeniable advantage of
serving as a unique and common support document to all inspectors and experts, thus helping
to reduce the variability of the results of visual inspection.

4. Specifications for visual inspection

An anomaly can therefore be profiled. We will now describe how to detect them, how
to evaluate them and how to decide if the product is acceptable or not. Figure 4 details the
visual inspection process.
Exploration is the first step. The inspector has to look over the product surface
carefully to check to see if there is a deviation and, if that is the case, to identify what type of
deviation it is. In the case of a local deviation, further analysis must be carried out because
each type of anomaly can involve different sensory attributes.
Evaluation is the next step. If the inspector detected a deviation from a reference or
from the intent of the designer, he will use scaling to assess it. In the case of a local deviation,
the inspector has to use descriptive profiling. In this case, he has to evaluate each anomaly
attribute that has been previously defined for each type of anomaly.
Decision is the last step. The inspector calculates the total intensity of the anomaly
with a method that correlates the evaluation of the attributes and the expert’s decision
regarding the degree of compliance. This decision is made according to tolerances which can
vary with the product’s value or the client.

Figure 4. Visual inspection process

4.1 Exploration step

Many factors impact the performance of the exploration step. Table 3 lists some of
these factors (Debrosse et al., 2010)
Table 3. Influencing factors in the exploration step.
Factors

Material Size, weight, color, texture, etc.


Manpower Visual acuity, age, experience, training, fatigue,
concentration, motivation, etc.
Mean Lighting system, workstation, magnification, etc.
Method Observation standard, magnification, light effects, time, etc.
Conditions Unwanted light, noise, dust, natural lighting, etc.

The BP X 10-042 (2006) manual written by AFNOR recommends the use of


workstations to prevent unwanted light. The workstation also guarantees that the exploration
conditions are identical for all inspectors (same light system, same light, same observation
background color, etc.) and thus limits some of the variability in the visual inspection results.
However, it is not always possible. In Mobalpa Fournier SA for example, these conditions
cannot be applied because parts can vary widely in size and the control has to be carried out at
several different stages in the production process. There is not only a final inspection, and it is
important to note that the evaluation step is performed under the same conditions as the
detection step.
Then the lighting system must be adapted to minimize these factors. The NF X 35-103
standard (1990) and the Safety data sheet ED 85 (INRS, 2009) give some principles to
respect. Table 4 lists some of the specifications for light system.
Table 4 Light system specification for visual inspection
Area lighting 750 – 1000 lux
Color rendering index (CRI) >85
Color temperature ~4000K
Perpendicular to the work plan, but can be adapted to facilitate the task (but avoid
Orientation
dazzle)
Louver use Yes (parabolic louver or opaque depending on the inspected product)

On factors like manpower, the training of inspectors is decisive in order to improve


this step. During training, it is possible to explain to them what an anomaly is made up of,
what the different types of anomalies are, how to detect a progressive anomaly, etc. Giving
the inspectors some feedback on how they perform the exploration task, in order to make
them more committed to this task is also an important part of the learning process (Megaw,
1979)
For the exploration step, it is also possible to define an appropriate strategy, such as
(Nickles III, 2003)
- A random exploration strategy: each zone of the part is likely to be explored,
possibly several times.
- A systematic exploration strategy: each zone of the part is controlled once but
systematically.
Arani et al. (1984) and Wang et al. (1997) showed that the systematic exploration
strategy significantly increased the performance of detection. Indeed, it limits the risk of
forgetting some areas to explore and at the same time reduces the exploration time, each area
being explored only once. This is the strategy which has been chosen to define an observation
standard, such as the example given in Figure 5. This standard includes:
- The "where" and the "what", which define the zones to observe and the elements to
inspect,
- The "how", which defines what path the inspector’s eyes should follow during
exploration. This definition of the path prevents the inspector from focusing
exclusively on certain anomalies and / or focusing only on those most recently
detected,
- The "when", which defines at what stage of the manufacturing process the
inspection takes place,
- The "how long", which defines an exploration time for each zone.
Figure 5. Example of a detection procedure.

4.2 Evaluation and decision

The most critical point of the visual inspection process is how to make the link
between evaluation and decision. In order to make a decision, the inspector must have a
method that correlates the intensity of each attribute of the anomaly with the quality standard
of the product.
Several methods are proposed in literature, for example the criteria / level table, the
tree-like evaluation table or the corrected hierarchical evaluation table (Baudet et al., 2011).
Erreur ! Source du renvoi introuvable. shows the table we used in our research program
with one of our industrial partners (Mobalpa Fournier SA).

Figure 6 Corrected hierarchical evaluation table

The inspector reads the table from left to right, adds up all the attribute values in order
to obtain the anomaly’s final intensity value (the attribute values can also be weighted).
This value allows him to decide whether the product should be accepted or not. For
example, in the case of Figure 6, if the anomaly he has detected is visible at arm's length, if it
is visible from all angles, if it causes a break in the form, if it is the same color as the product
and if it is on a visible part when the furniture is mounted and closed (+1), the inspector
decides to refuse the product (final intensity = 3 + 1 + 1 + 0 + 1).
All the tables make a linear correlation between the attributes’ values and the quality
standard of the product. However, when a non-linearity exists, a neural network can be used.
An example of the use of this method is described in [Baudet et al, 2012].

5 Conclusion

In this work, we have proposed how sensory analysis can be applied to the visual
inspection of product surfaces (mainly when the product has an anomaly (scratch, pit, etc.),
that can influence aesthetic perception. Thus, for local deviation, sensory profiling is the most
adequate test to be used. Inspectors can assess how its attributes influence the quality
perception of the product, using an anomaly profile. However, visual inspection is carried out
by several persons, so the choice of a good standard is necessary. We proposed to use the
written standard, since it is the best way to communicate, to maintain and to share quality
standard of the company.
Sensory analysis is only a tool to improve visual inspection. If a company wants to
control this process, it must control all the steps (exploration, evaluation and decision).
Exploration requires good conditions to improve and facilitate the detection of anomalies.
When a workstation cannot be set up, the light system must at least be adequate. When an
exploration method had to be created, systematic strategies proved to be the best solution to
develop an observation standard. Inspector training must be implemented, as it is important to
explain what an anomaly is, inform the inspectors of the different types of anomaly that exist
and how to detect them and to motivate them.
The evaluation and decision steps are closely linked. To improve these steps, a
company must choose the right evaluation attributes and then develop a method which
correlates evaluation attributes with the quality standard. Inspector training on the evaluation
attributes and the use of the method is obviously necessary.
The proposed methodology for visual inspection has been proved to be very robust. It
can be used for any type of product (lighters, watches, furniture, packaging, etc.) when its
surface has a local deviation. It allows one to assess the influence that the anomaly causes on
the aesthetic perception of the product. For this methodology, the choice of evaluation
attributes and their levels can vary from product to product but, in general, the parameters will
always be the same (factual description, local and general context).

Acknowledgements

Financial support of this work was provided by the European INTERREG VIA
France-Switzerland program. The authors wish to acknowledge the reviewer for allowing the
improvement of this work.
References

Arani, T., Karwan, M. H., Drury, C. G. (1984) A variable-memory model of visual search. Human Factors, 26(6), 631-639.
Baudet, N., Pillet, M., & Maire, J. L. (2011). Visual inspection of product: a comparison of the methods used to evaluate surface anomalies.
International Journal Metrology and Quality Engineering, 2, 31 – 38.
Baudet, N., Pillet, M., & Maire, J. L. (2012). Neural network for improvement of the visual quality process. International Conference on
Kansei Engineering and Emotion Research, Penghu, Taiwan.
BP X 10-042, (2006). Caractérisation sensorielle des matériaux : Recommandations pratiques pour l’analyse visuelle de la matière première
au produit fini. AFNOR
Chabukswar, S., Gramopadhye A. K., Melloy, B. J. & Grimes, L. W. (2003). Use of aiding and feedback in improving visual search
performance for an inspection task. Human Factors and Ergonomics in Manufacturing, 13(2), 115-136.
Costell, E. (2002). A comparison of sensory methods in quality control. Food Quality and Preference, l3, 341-353.
Creusen, M. E. H. & Schoormans, P. L. (2005). The different roles of product appearance in consumer choice. The Journal of Product
Innovation Management, 22, 63-81.
Debrosse, T., Pillet M., Maire J-L., & Baudet N. (2010) Sensory Perception of Surface Quality: Industrial Practices and Prospects,
International Conference on Kansei Engineering and Emotion Research, Paris, France.
Depledt, F. (2009). Evaluation sensorielle : Manuel méthodologique. (3th ed.) Lavoisier, (Chapter 4).
Garaas, T. W. & Pomplun, M. (2008). Inspection time and visual-perceptual processing. Vision Research, 48, 523-537.
Garret, S. K., Melloy, B. J. & Gramopadhye, A. K. (2001). The effects of per-lot and per-item pacing on inspection performance.
International Journal of Industrial Ergonomics, 27, 291-302.
Giboreau, A., Navarro, S., Faye, P., & Dumortier, J. (2001). Sensory evaluation of automotive fabrics: the contribution of categorization
tasks and non-verbal information to set-up a descriptive method of tactile properties. Food Quality and Preference, 12, 311-322.
Gilden, D. L., Thornton, T. L. & Marusich, L. R. (2010). The serial process in visual search. Journal of Experimental Psychology: Human
Perception and Performance, 36(3), 533-542.
Guerra, A.S. (2009) Métrologie sensorielle dans le cadre du contrôle qualité visuel. PhD Thesis, Université de Savoie, laboratoire SYMME,
2009.
Hassan, M. H., & Diab, S. L. (2010). Visual inspection of products with geometrical quality characteristics of known tolerances. Ain Shams
Engineering Journal, 1, 79 – 84.
ISO (1994). Sensory analysis. General guidance for selection, training and monitoring of assessors. Part 2: Experts. International Standard
no. 8586-2, Switzerland: International Organisation for Standardization.
ISO (1995) Analyse sensorielle – Recherche et sélection de descripteurs pour l’élaboration d’un profil sensoriel, par approche
multidimensionnelle, Switzerland: International Organisation of Standardisation. International Standard, no.11035
Jebaraj, D., Tyrrell, R. A., & Gramopadhye A. K. (1999). Industrial inspection performance depends on both viewing distance and
oculomotor characteristics. Applied Ergonomics, 30, 223-228.
Lawless, H. T., & Heymann, H. (2010) “Sensory evaluation of food – Principles and Practices”. (2° Ed) Springer (Chapter 1).
Lee, F. C. H., & Chann A. H. S. (2009). Effects of magnification methods and magnifier shapes on visual inspection. Applied Ergonomics,
40, 410-418.
MacDougall, D. B. (2003) Sensory evaluation – Appearance. Encyclopedia of Food Sciences and Nutrition (2° Ed), 5161-5167
Megaw, E. D. (1979) Factors affecting visual inspection accuracy. Applied Ergonomics, 10(1), 27-32
Munkevik, P. Hall, G., & Duckett, T. (2007). A computer vision system for appearance-based descriptive sensory evaluation of meals.
Journal of Food Engineering, 78, 246-256.
NF X35-103 (1990) Egornomie – Principes d’ergonomie visuelle applicables à l’éclairage des lieux de travail. AFNOR
Nickels III, G. M., Melloy, B. J., & Gramopadhye A. K. (2003). A comparison of three levels of training designed to promote systematic
search behavior on visual inspection. International Journal of Industrial Ergonomics, 32, 331-339.
Pereira, R., Matia-Merino, L., Jones, V., & Singh, H. (2006). Influence of fat on the perceived texture of set acid milk gels: a sensory
perspective. Food Hydrocolloids, 20 (2-3), 305-313.
Rao, P., Bowling, S. R., Khasawneh, M. T., Gramopadhye A. K., & Melloy, B. J. (2006). Impact of training standard complexity on
inspection performance. Human Factors and Ergonomics in Manufacturing, 16(2), 109-132.
Rebsamen, M., Boucheix, J. M., & Fayol, M. (2010). Quality control in the optical industry: From a work analysis of lens inspection to a
training programme, an experimental case study. Applied Ergonomics, 41, 150-160.
Schütte, M., Dettmer, U., Klatte, H., & Lauring, W. (1999). Comparison of different layouts of inspection instruction for the production
department of a company in the electronic industry. International Journal of Industrial Ergonomics, 23, 439-450.
Soufflet, I., Calonnier, M., & Dacremont, C. (2004). A comparison between industrial experts’ and novices’ haptic perceptual organization: a
tool to identify descriptors of the handle of fabrics. Food Quality and Preference, 15(7-8), 689-699.
Vandevyer, B. (2005) Fiche pratique de securité – Eclairage artificiel au poste de travail. INRS, ED 85
Wang, M. J. J., Lin, S. C., Drury, C. G. (1997) Training for strategy in visual search, International Journal of Industrial Ergonomics, 20, 101-
108.
Zamuner, G., “Application of Artificial Vision to the Quality Inspection of Surfaces in the Case of Luxury Products”, PhD Thesis, Ecole
Polytechnique Federale de Lausanne, Laboratoire de Production Microtecnique (LPM), 2011.

View publication stats

You might also like