Note: Descriptions are shown in the official language in which they were submitted.
CA 02202765 1997-04-15
1
METHOD FOR IDENTIFYING MISCELLANEOUS OBJECTS,
SPECIES OR ITEMS, AND USES THEREOF.
The invention relates to a method for identifying
miscellaneous objects, species or items and also
to the various uses of this identification method.
Recognition systems which operate-on the basis of
dichotomising check digits exist for the
identification of miscellaneous objects, species
or items, especially for the identification of
"weeds". To be more precise, the known prior art
permits the identification of a species by the
elimination by chosen successive criteria of all
the species that do not meet the criteria in
question. The disadvantages of those expert
systems are numerous.
On the one hand, they are complex and their use
requires detailed knowledge of the subject-matter.
Then, these methods have absolutely no flexibility
because they use a determined order of criteria
which cannot be modified. For example, in order to
identify a botanical species, it is first of all
necessary to meet general criteria, such as, for
example: herbaceous plant, either never having
CA 02202765 1997-04-15
2
flowers, or of the ligneous creeper type, or even
bushes and shrubs, etc. If, for example, the
species is a tree, which may be difficult to
identify if the plant has just germinated, it is
necessary to reply, for example, to questions such
as: carrying cones or carrying catkins; it is then
necessary to identify whether the leaves are
narrow and acicular or, for example, small and
squamiform, etc.
In addition to the fact that it seems clear that,
without a detailed knowledge of the subject-
matter, the rLsk of choosing incorrect replies is
great, it also seems that from the moment at which
an error has been committed, the species to be
identified, having been eliminated, will not
appear.
In brief and to summarise, these known methods of
the prior art are considerably lacking in
flexibility, they require a detailed knowledge of
= the subject-matter, do not accept an error of
observation and judgement, do not tolerate lack of
information, require a reply to each question and,
furthermore, take a long time to implement.
JOURNAL OF COMPUTING IN HIGHER EDUCATION, Fall
1991, USA, volume 3, N . 1, ISSN 1042-1726, pages
CA 02202765 1997-04-15
3
85-103, RAGAN L.C. :"Hypermedia in the plant
sciences : the Weed Key and Identification
System/videodisc" describes a search system for
identifying plant species on the basis of
characteristics selected by the user. This system
is designed to take into account the situation
where some data are missing or uncertain and
therefore to use only information which an
untrained user is capable of providing: the latter
is able not to provide information on some
criteria or can reply "uncertain" to some criteria
in order not to eliminate the species sought by an
incorrect reply on a criterion. Once information
has been given on one or more criteria, the search
can be initiated at any time and the system then
provides a list of species corresponding to
this/those criterion/a or a negative reply if no
species meets the criteria. The user can cause
texts and images relating to the selected species
to be displayed.
However, this system is dichotomising by nature
because, if there is an incorrect reply on a
single criterion, the corresponding species are
systematically eliminated and can be found again
only by taking up the search again with (a)
corrected criterion/a or corrected information.
CA 02202765 2004-08-31
4
The aim of the invention is to overcome these disadvantages_
According to the present invention, there is provided a method for identifying
miscellaneous objects, species or items by means of a list formed by sets of
identifying
features, each set corresponding to a known object, species or item, the
method
cornprising the following steps: the user enters said list in a computer
memory, the user
interactively interrogates the list by entering successively in the computer,
in the order he
desires, features of the object, species or item to be identified which seem
to be
determining and which he selects each tirne beforehand by pointing the cursor,
in a
composite picture of the object, species or item to be identified, on those of
the
predeternvned portions which have the feature in question, the composite
picture being
displayed on the computer screen and being updated as the selected features
are entered,
and after each feature has been entered, the computer identifies and displays
on the screen
a limited number of known objects, species or items corresponding to at least
some of the
features entered up to that point, by means of a calculation of probabilities
effected for
each known object, species or item appearing in the list, the calculation
being based on a
weighting between the features.
Preferred embodiments of the invention advantageously provide the user with a
method for identifying miscellaneous objects, species or items, which method
is very
flexible, does not require a detailed knowledge of the subject-matter, being
usable by
anyone having a minimum power of observation, tolerates errors of observation
and
judgement while avoiding the systematic elimination of a subject even if some
of the
replies made to the identifying features or criteria selected by the user
prove to be
erroneous, and leaves it to the user to choose the identifying features or
criteria to which
he wishes to reply.
Advantageously, after selection and verification, other criteria are
optionally
entered for identification (if the selected subject is not satisfaetory) or to
check the
previous identification (as a safety measure).
CA 02202765 2004-08-31
According to a preferred embodiment of the method, after selection or
verification, given selection criteria retained at the begirming are
optionally corrected by
removing those criteria or by -- --- = -------
CA 02202765 1997-04-15
6
modifying them, which enables the reliability of
the diagnosis to be improved.
In a preferred embodiment, the computer is caused
to calculate and display, in respect of the
identified objects, species or items, an --
identification probability corresponding to a
weighting between exact criteria and non-exact
criteria. Advantageously, for the,probability
calculation carried out by the computer, a greater
weight is allocated to some criteria than to
others, as a function of the "discriminating"
nature of thE criterion in question. Thus, for
example, if the colour of a flower is not always
determining, the fact that the plant which carries
the flower has or does not have thorns is much
more determining.
Of the numerous uses of the method of the
invention, attention has turned very especially to
the identification of botanical species; in such
an application, the selection criteria which have
been entered in the memory and which can be chosen
by the user comprise usual identifying criteria or
features such as: habit, phyllotaxis, simple or
composite leaves, thorns, type of flower, type of
root, etc..., the user choosing as desired, and in
the order he considers the most appropriate to the
CA 02202765 1997-04-15
7
case in point, the selection criteria which he
chooses and which he can "click" onto in the
composite picture of the botanical species, which
composite picture is updated progressively as the
selected criteria are retained by the user.
The invention and its implementation will be
appreciated more clearly by means of the following
description given with reference to the appended
drawings which illustrate, by way of example, some
procedures for identifying botanical species.
In the drawings:
- Figure 1 illustrates what appears on
the user's computer screen at the beginning of the
procedure for identifying a botanical species,
- Figure 2 shows a fresh screen on which
the user will determine the.habit of the subject
to be identified,
- Figure 3 shows what appears on the
screen after the composite picture has been
modified by the computer system in order to
integrate the habit selected by the user,
- Figure 4 shows a fresh screen on which
the user will select the type of leaves
corresponding to the subject to be identified,
- Figure 5 shows the following screen on
which the composite picture integrates the fresh
CA 02202765 1997-04-15
8
additional criterion selected by the user in
relation to the type of leaves of the subject,
- Figure 6 indicates the species selected
by the computer which meet the criteria entered,
allocating to each species of the list an
identification probability which corresponds to
the probability that the specimen corresponds to
that species,
- Figure 7 shows the following screen of
the procedure in which the user has returned to
the illustration of Figure 5, wishing to refine
his selection by a fresh criterion,
- Figure 8 illustrates the following
screen enabling the user to select the additional
criterion retained, namely, in this case,
phyllotaxis,
- Figure 9 shows the following screen in
which the composite picture comprises all the
criteria previously entered, the computer this
time identifying only a single species which
corresponds 100% to the criteria entered,
- Figure 10 illustrates in the form of a
drawing the species identified by the computer for
comparison by the user with the subject,
- Figure 11 illustrates the following
screen in which the user has asked for an enlarged
view of the drawing and, in particular, of the
plantlet in order to verify the identification,
CA 02202765 1997-04-15
9
- Figure 12 illustrates the composite
picture displayed at the end of another
identification procedure by the computer, after
the user has entered several criteria and, on the
basis of those criteria, a species appears to be
identified with a fairly high degree of
probability.
Reference will first of all be made to Figure 1 in
which the user has before him on the computer
screen a typical composite picture of a plant
basically comprising a stem 1, several leaves
distributed along the stem, a flower 3, and roots
4.
In addition, the right-hand side of the screen
shows on a larger scale a leaf 5 having a given
number of places enclosed in rectangles permitting
the identification of, for example: at 6, the
attachment of the leaf 5 to the stem 1, at 7, the
naissance of the leaf, at 8, the lateral edge of
the leaf, at 9 the tip of the leaf, etc...
Appearing on the extreme right of the screen are
also clear inscriptions and logos: respectively
"no species identified" 0%; the drawing of a leaf
or a grass; "phyllotaxis"; "habit"; and finally
the button "calculation". The use and the meaning
CA 02202765 1997-04-15
of these expressions and logos will appear
hereinafter.
With the subject to be identified before his eyes,
the user first of all notices that the plant is a
creeping plant. Since this criterion seems to him
to be particularly determining, the user "clicks"
onto the logo "habit" in Figure 1. In accordance
with conventional information science procedures,
the software contained in the computer responds to
th.is "click" of the mouse by displaying, as
indicated in r'igure 2, a screen on which several
diagrammatic :iabits appear.
On this screen, the user "clicks" onto the
crawling habit as shown in a frame. The following
screen then appears in Figure 3 in which the
composite picture of the plant is shown again but
with the modification of a crawling habit.
The user now enters a new criterion relating to
the type of leaf. He clicks onto a leaf of the
composite picture or, if desired, onto the
corresponding logo at the right-hand side of the
screen, the computer displaying the screen of
Figure 4 which offers the user simple leaves (at
the top of the screen) or composite leaves (at the
bottom of the screen). Because the subject has
CA 02202765 1997-04-15
11
composite leaves, the user clicks onto the
composite leaves as shown diagrammatically by the
rectangle framing the lower line. In Figure 5, the
composite picture is updated: crawling habit,
composite leaves. In this situation, it will be
noted that the computer displays "two species 100%
identified"; this means that only two species
contained in the computer's memory meet the set of
criteria which have just been entered. The user
can either consult the species in question or
refine the procedure.
If he wishes to consult the list of species, he
clicks onto calculation, which causes the screen
illustrated in Figure 6 to appear with, at the top
left, the small-scale reminder of the updated
composite picture and, below, the list of species
found in decreasing weight of probability.
The two 100% species: Tribulus terrestris and
Zornia glochidiata mean that these two species
have all of the criteria identified. The other
species found, with weightings of from 62% to 38%,
mean that, for those species, at least one of the
criteria does not correspond.
In this situation, the user chooses to refine the
procedure; he clicks onto the composite picture at
CA 02202765 1997-04-15
12
the top left of Figure 6 returning to the screen
of Figure 7 which is identical to that of Figure
5. He then clicks onto "phyllotaxis"; the computer
displays the screen of Figure 8, proposing three
possible leaf attachments to the stem. The user
selects opposite phyllotaxis as indicated in
framed form on the left of the screen of Figure 8.
The computer then displays the screen of Figure 9,
indicating a single 100% identified species, that
is to say, having all the criteria concerned.
The user then calls up that species and verifies
the agreement Df the images, photos and drawings
with his specimen. Figure 10 shows only the
drawing as it appears on screen. In the case
concerned, the identified species was Tribulus
terrestris. The identification is confirmed by
verification as illustrated in Figure 11 of a
full-screen drawing of the plant with enlargement
of the plantlet which appears on the right of the
screen.
In Figure 10, it will be observed that the
computer has selected approximately 20 different
species and has allocated to them probability
coefficients corresponding to the number of
correct criteria weighted by the relative
importance of the criteria as indicated above.
CA 02202765 1997-04-15
13
It will thus be appreciated, for example, that
Zornia glochidiata, which had been identified as
corresponding 100% to the screen of Figure 6, is
now retained only with a 57% probability; the
difference is due to the fact that the user has
entered an important additional criterion, namely
opposite phyllotaxis, which Zornia glochidiata
does not meet and it is therefore=finally not the
plant to be identified.
Referring to Figure 12,,this Figure shows the
typical case where the user has succeeded in
identifying a plant fairly rapidly and easily,
although he has been mistaken in the determination
of some identifying criteria.
The screen of Figure 12 gives the composite
picture of a plant of which the various criteria
as identified in the various rectangles marked 10
to 17 in the Figure have been entered for
identification by the user, namely, respectively:
naissance of the leaf on the stem, starting shape
of the leaf, lateral edge of the leaf, tip of the
leaf, cross-section of the stem, general shape of
the leaf, pilosity of the upper face of the leaf
and pilosity of the lower face.
CA 02202765 1997-04-15
14
In the case under consideration, the computer has
identified the plant as being "Mollugo nudicaulis"
with 92% probability, the user having been
mistaken in the identification of some criteria
which, however, were of secondary importance. By
comparison with the species selected by the
computer and full-screen display of the photos and
drawings of the species in question, the user can
verify whether the species in question is the
correct one. If it is not, he can refer to the
following species which he can consult,
advantageously first verifying the species
selected by the computer in descending order of
probability.
At this stage the user can optionally also
correct, by verification, any criterion which he
has entered erroneously or he can add an
additional criterion, such as, for example, the
phyllotaxis or the presence or absence of a flower
or the shape of the roots and attachment, this
being effected in case of doubt or to verify the
accuracy of the diagnosis.
It will be appreciated from the above description
that the identification method of the invention is
very versatile because it is very tolerant of a
lack of information and even an error in secondary
CA 02202765 1997-04-15
features or criteria of which the identification
is not always easy. Of course, identification will
be effected all the more rapidly if the user
enters determining criteria, that is to say, of
substantial weight and reliable identification.
In addition, although the invention has been
illustrated in connection with the identification
of botanical species, it will be appreciated that
the method can also be used in connection with the
identification of any desired object or item which
can be defined by any criteria of appearance,
shape, weight and, generally, having any intrinsic
qualities whatever. The invention could thus be
used, for example, in the identification of
persons, in the identification of industrial -
objects, etc. It would seem that its most
immediate and useful applications must concern
identification in the biological field, for
example, the identification of viruses, fungi,
animals, agricultural varieties and plants. The
embodiments indicated in connection with the
identification of some botanical varieties can be
immediately extrapolated to identification
broadened to cover the entire biological domain.