Pattern Recognition 41 (2008) 3092 -- 3103

Contents lists available at ScienceDirect

Pattern Recognition
journal homepage: w w w . e l s e v i e r . c o m / l o c a t e / p r

Recognition of degraded characters using dynamic Bayesian networks

Laurence Likforman-Sulem , Marc Sigelle
TELECOM Paris Tech/TSI and CNRS LTCI UMR 5141, 46 rue Barrault F-75634 Paris Cedex 13, France

A R T I C L E

I N F O

Article history:
Received 1 March 2007
Received in revised form 15 January 2008
Accepted 15 March 2008

Keywords:
Markovian models
Hidden Markov models
Dynamic Bayesian networks
Historical documents
Broken character recognition

A B S T R A C T

In this paper, we investigate the application of dynamic Bayesian networks (DBNs) to the recognition of
degraded characters. DBNs are an extension of one-dimensional hidden Markov models (HMMs) which
can handle several observation and state sequences. In our study, characters are represented by the
coupling of two HMM architectures into a single DBN model. The interacting HMMs are a vertical HMM
and a horizontal HMM whose observable outputs are the image columns and image rows, respectively.
Various couplings are proposed where interactions are achieved through the causal influence between
state variables. We compare non-coupled and coupled models on two tasks: the recognition of artificially
degraded handwritten digits and the recognition of real degraded old printed characters. Our models
show that coupled architectures perform more accurately on degraded characters than basic HMMs, the
linear combination of independent HMM scores, as well as discriminative methods such as support vector
machines (SVMs).
2008 Elsevier Ltd. All rights reserved.

1. Introduction
Since the seminal work of Rabiner [1], stochastic approaches
such as hidden Markov models (HMMs) have been widely applied to
speech recognition, handwriting [2,3] and degraded text recognition
[4,5]. This is largely due to their ability to cope with incomplete information and non-linear distorsions. These models can handle variable length observation sequences and offer joint segmentation and
recognition which are useful to avoid segmenting cursive words into
characters [6]. However, HMMs may also be used as classifiers for
single characters [7,8] or characters segmented from words by an
"explicit'' segmentation method [9]: the scores output for each character and each class are combined at the word level. Another property of HMMs is that they belong to the class of generative models.
Generative models better cope with degradation since they rely on
scores output for each character and each class while discriminative
models, like neural networks and support vector machines (SVMs),
are powerful to discriminate classes through frontiers. In case of
degradation, characters are expected to be still correctly classified
by generative models even if lower scores are given.
Noisy and degraded text recognition is still a challenging task
for a classifier [10]. In the field of historical document analysis, old
printed documents have a high occurence of degraded characters,
especially broken characters due to ink fading. When dealing with

Corresponding author. Tel.: +33 1 45 81 73 28.

E-mail address: laurence.likforman@telecom-paristech.fr (L. Likforman-Sulem).

0031-3203/$30.00 2008 Elsevier Ltd. All rights reserved.

doi:10.1016/j.patcog.2008.03.022

broken characters, several options are generally considered: restoring and enhancing characters [11--13] or recovering characters
through sub-graphs within a global word graph optimization scheme
[14]. Another solution is to combine classifiers or to combine data.
Several methods can be used for combining classifiers [15], one of
them consists of multiplying or summing the output scores of each
classifier. In the works of [16,17], two HMMs are combined to recognize words. A first HMM, modeling pixel columns, proposes word
hypotheses and the corresponding word segmentation into characters. The hypothesized characters or sub segments are then given to
a second HMM modeling pixel rows. This second HMM normalizes
and classifies single characters. The results of both HMMs are combined by a weighted voting approach or by multiplying scores. Our
approach differs with restoration methods as it aims at enhancing
the classification of characters without restoration. This is motivated by the fact that preprocessing may introduce distortions to
character images. In our previous work [18], we compared data and
decision fusion and showed that data fusion yields better accuracy
than decision fusion for HMM-based printed character recognition.
The present dynamic Bayesian network (DBN) approach is a data
fusion scheme which couples two data streams, image columns and
image rows into a single DBN classifier. It differs from the approach
presented in [16,17] where two classifiers are coupled (one classifier per stream) in a decision fusion scheme, and from a data fusion
scheme consisting of a multi-stream HMM which would require
large and full covariance matrices in order to take into account
dependencies between the streams [18].
Our study consists of building DBN models which include in a
single classifier two sequences of observations: the pixel rows and

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

the pixel columns. It can be seen as coupling two HMMs into a

single DBN classifier, as opposed to combining the scores of two
basic HMM classifiers in a decision fusion scheme. The two HMM
architectures, each including an observation stream associated with
state variables, are linked in a graphics-based representation. Two
different streams are jointly observed and the model parameters
(state transition matrices) reflect the spatial correlations between
these observations.
We apply the DBN models to broken character recognition. As
generative models, DBNs are adapted to degraded character recognition. These models also provide a certain robustness to degradation due to their ability to cope with missing information. They
have the ability to exploit spatial correlations between observations. Thus a corrupted observation in the image can be compensated by an uncorrupted one. We compare several DBN architectures
among themselves, with other fusion models like the combination of
independent HMMs, and with a SVM classifier.
The paper is organized as follows. In Section 2, we briefly introduce Bayesian networks (BN) and DBNs. In Section 3, we present
several independent or coupled models. In Section 4, we apply these
models to the problem of broken character recognition (artificial
and real). We conduct several experiments to show the advantages
of DBNs by comparing their performance with the combination of
HMM scores and with a SVM classifier. Conclusions are drawn in
Section 5.
2. Dynamic Bayesian networks
A (static) BN associated with a set of random variables X =
(X1 , X2 , . . . , XN ) is a pair: B = (G, ) where G is the structure of the
BN i.e., a directed acyclic graph (DAG) whose nodes correspond to
the variables Xi X and whose edges represent their conditional
dependencies, and  represents the set of parameters encoding the
conditional probabilities of each node variable given its parents. The
distributions are represented either by a conditional probability table (CPT) when a node and its parents represent discrete variables,
or by a conditional probability distribution (CPD) when a node represents a continuous variable. Each CPD usually follows a Gaussian
probability density function (pdf). A key property of BNs is that the
joint probability distribution factors as
P(X1 , X2 , . . . , XN ) =

N


P(Xi | Pa(Xi ))

i=1

where Pa(Xi ) denotes the parents of Xi . This property is central in

the development of fast inference algorithms. Static BNs have been
applied to on-line character recognition and signature authentication
for modelling dependencies between stroke positions or signature
components [19--21].
DBNs are an extension of static BNs to temporal processes occuring at discrete times t  1. In the following, we consider DBN models
which have two observation streams. We will use indices i = 1, 2 to
denote the two streams. The variables Xi and Yi denote the respective
1

X2

Xti ), i = 1, 2.
DBNs provide general-purpose training and decoding algorithms
based on the expectation-maximization (EM) algorithm and on inference mechanisms [22]. Model training consists of estimating model
parameters, CPTs and CPDs. Inference algorithms are performed on
the network to compute the best state sequences or the likelihoods
of observation sequences.
An HMM is a particular case of DBN where there is only one observation stream and one state sequence. The dynamic character of
DBNs makes it suitable for applications such as speech and character recognition. In [23,24], DBNs are used to model the interactions
between speech observations at different frequency bands in a way
that is robust with respect to noise.
3. Independent and coupled architectures
In this study, we couple data streams into single DBN classifiers.This coupling is performed through various DBN architectures
(graphical representations) which combine two basic HMMs: the
vertical HMM whose outputs are the columns of pixels and the
horizontal HMM whose outpouts are the image rows. In our models, the interactions are usually (but not only) performed through
states, leading to efficient models in terms of model complexity (see
Section 3.3). Brand et al. [25] have proposed coupled architectures
"coupled HMMs'' for modeling human interactions: in their models,
a state of one HMM is linked to all other HMM states of the adjacent
time-slice. This yields symmetric architectures while our coupled
architectures are highly non-symmetric.
In our framework, all character classes share the same DBN architecture. Admissible architectures do not include continuous variables with discrete children (for exact inference purposes [23]) and
have also a small number of parameters (in order to get a tractable
inference algorithm). One approach consists of learning network architecture from data [26]. This approach is tractable for static BNs
when dealing with a few observed variables but becomes rapidly too
complex in the presence of hidden state variables. Automatic architecture learning is beyond the scope of this paper and our strategy
consists of heuristically looking for various admissible architectures
and selecting those which provide the best recognition performance.

Y2

X2

X3

Y11

X21

X22
Y1

the conditional state distribution P(Xt2 | Xt1 ), the state transition dis2 ) and the two CPDs are the Gaussian pdfs P(Y i |
tribution P(Xt2 | Xt1
t

1
Y2

Y11

X21

hidden state and observation attributes in stream i. Xti and Yti are the
random variables (nodes) for Xi and Yi at time t.
We assume that the process modelled by DBNs is first-order
Markovian and stationary. In practice, this means that the parents
of any variable Xti or Yti belong to the time-slice t or t 1 only, and
that model parameters are independent of t. Parameters are thus
tied and a DBN can be represented by the first two time slices as in
Fig. 1. For each observation sequence, the network is repeated as
many times as necessary. Fig. 1 shows an example of unrolled DBN
for an observation sequence of length T = 3: the initial network is
repeated T times. Parameters for this model are given by CPTs and
CPDs: the three CPTs are the initial state distribution encoding P(X11 ),

X11

X1

3093

X22
2

Y1

Y13

Y2

X3
2

Y2

Y3

Fig. 1. Because of parameter tying, a DBN can be represented by only two time slices (left). To fit the two observation sequences {Y 1 } and {Y 2 } of length T = 3, the DBN is
unrolled and represented on 3 time slices (right).

3094

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

X1

X2

XT

X1

X2

XT

Y11

Y12

YT

Y1

Y2

YT

Fig. 2. Independent HMMs represented as DBNs: (a) vertical-HMM and (b) horizontal-HMM.

Fig. 3. Horizontal and vertical observation sequences obtained by scanning digit 3 from top to bottom and from left to right, respectively. Digit images are normalized to
size d d. Length of observation sequences is T = d, length of observation vectors is also d.

3.1. Independent architectures

We construct two basic HMMs using the DBN formalism. The vertical (resp. horizontal) HMM is constructed using the vertical (resp.
horizontal) writing stream, as depicted in Fig. 2a and b. Observations
for the vertical (resp. horizontal) HMM consist of columns (resp.
rows) of pixels (normalized values) obtained from scanning the character image from left to right (resp. top to bottom) as shown in Fig. 3.
Characters are normalized to a square of size dd pixels (see Section
4). Thus the length T of each observation sequence, either horizontal
or vertical, is T = d and the length of observation vectors is also d.
The parameters of these basic HMMs are CPTs A, CPDs B and
the initial distribution . CPTs A are associated to nodes Xti , t  2,
CPDs B to observed nodes Yti 1 and the initial state distribution  is
associated to node X1 . They are written for each stream i, vertical
(i = 1) or horizontal (i = 2), and for t  2 as
i
i
A = P(Xti = k | Xt1
= j), k, j [1, Q ]

j,k

Bi (yi ) = P(Y i = yi | X i = k)
t
t
t
k t
(1)
i ; i , i ),

=
N(y

t k k

i
1 (k) = P(X1i = k), k [1, Q ].

Q is the global number of hidden states. CPTs A are state transition

matrices of size Q Q . As in classical HMMs, we constrain A to allow
only left-right state transitions for parameter reduction purposes:
i
the value of Xti is either equal to the value of Xt1
or equal to that
i
of Xt1
+ 1. Each observation variable Yti follows a single Gaussian

Because of the stationarity assumption, all nodes Xti share the same CPT A
and all nodes Yt share the same matrix B.
1

probability density function (pdf) N(yti , ik ): ik is the mean vector

of length d of the Gaussian pdf associated to the current state k, ik

is a full covariance matrix of size d d (see Section 3.3).
A first limitation of HMMs is the observation independence assumption conditionally to hidden states. However, we can bypass it
by building auto-regressive (AR) architectures where observations
are explicitly dynamically linked in time. An auto-regressive HMM
is determined by its type and the order p of the regression. There
are two types of auto-regressive HMMs: linear predictive models [1]
and switching Markov auto-regressive models [27]. The AR models
proposed here are switching Markov models and the model order
is one. An observed node Yti depends on both the current state Xti
i . The two resulting vertical-AR
and the previous observed node Yt1
and horizontal-AR single stream architectures remain however independent (Fig. 4a and b). The only parameters which differ from
basic HMMs are the CPDs B. The mean ik of the Gaussian probability density function associated to the current state k is shifted
i
i
by Wki yt1
according to the previous observation yt1
and the re-

gression matrix W. Each regression matrix Wki is of size d d, with

d being the length of observation vectors. Regression matrices for
each stream and each state are estimated during training. Each observation variable Yti follows a Gaussian probability density function

i
N(yti ; ik + Wki yt1
, ik ). CPDs B are written for each stream i and
for t  2 as

 i i
i
i
i
Bk (yt , yt1
) = P(Yti = yti | Xti = k, Yt1
= yt1
), k [1, Q ]
i
i
i
i
i
= N(yt ; k + Wk yt1 , k ).

(2)

The matrix A and the initial distribution  remain the same as for
basic HMMs. The matrix A is still constrained to be left--right. Note
that basic HMMs are a particular case of AR-HMMs with Wki = 0.

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

3095

X11

X21

XT1

X12

X22

XT2

Y11

Y21

YT1

Y12

Y22

YT2

Fig. 4. Independent auto-regressive AR-HMMs represented as DBNs: (a) AR-vertical and (b) AR-horizontal.

Fig. 5. Observation and state sequences for simple O and H shapes on a 5 5 grid. Joint configurations of long bars of pixels (observations a) and short bars of pixels
(observations b) occur for different state configurations.

3.2. Coupled architectures

Starting from the previous single stream and independent HMMs,
we now construct several coupled architectures. They are obtained
by adding directed edges between the two streams within the same
time-slice. Edges are directed from the vertical stream to the horizontal one in order to enhance the influence of the vertical stream.
Experiments of Section 5 show that the vertical HMM is more reliable than the horizontal one since vertical strokes are predominant
for the shapes considered [28,29]. The coupling proposed here requires that both observation sequences have the same length since
streams are synchronized at each time slice: each image column is
associated with one row. The observation length is T = d as the character image is previously normalized to a square of size d d with
d = 28 pixels.
At each time, coupled models are in two states, the state corresponding to the column observation (the vertical state) and the state
corresponding to the row observation (the horizontal state). A transition to the vertical state Xt1 depends only on the value of the pre1
like classical left--right HMMs. But a transition to
ceding state Xt1
the horizontal state Xt2 depends on both the value of the preceding

2
and the value of the current vertical state Xt1 . This destate Xt1
pendence between the horizontal and the vertical states expresses
the dependence of the observations, i.e. between row t and column
t. Although row t and column t share only one pixel in common, the
whole row and the whole column of pixels may be correlated. The

more they are correlated the higher the probability of observing one
column configuration captured by the vertical state, in conjunction
with one row configuration captured by the horizontal state. As an
example, consider simple shapes on a 5 5 grid belonging to two
classes: H and O shapes, as shown in Fig. 5. We set the number of
states to three and we consider two discrete observation symbols
a and b: Yti = a when the number of pixels in column (or row) t is

> 3, else Yti = b. For H shapes the long central bar (row observation
a) is correlated with short bars (column observation b) in the central area of the image. For O shapes, long bars (row observations
a) are correlated with long bars (column observations a) at the top
and bottom of the image. The state/observation sequences shown
for both models in Fig. 5 express these correlations. For O shapes,
when (Xt1 = 1, Xt2 = 1) or (Xt1 = 3, Xt2 = 3), the probability of observing
long bars (a) in both row and column is high. For H shapes, when
(Xt1 = 2, Xt2 = 2) the long horizontal bar (a) is observed jointly with
a short bar (b).
To obtain the first coupled architecture, called the state-coupled
model (ST_CPL), we add directed edges between the hidden state
nodes of the vertical and horizontal HMMs as shown in Fig. 6a. The
parameters of ST_CPL are the CPDs bi and CPTs A and U. The conditional probability table A capturing the HMM left-right structure
of the vertical sequence {X 1 } can be written:
1 = j)
Aj,k = P(Xt1 = k | Xt1

k, j [1, Q ],

3096

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

where Aj,k is a left-right state transition matrix as for classical leftright HMMs. The value k of the current state is either equal to the
value j of the preceding state or to j + 1. For t  2, we write:

1
Y1

1
Y2

1
YT

bik (yti ) = P(Yti = yti | Xti = k) for i = 1, 2.

(3)

the CPD bik is a single Gaussian pdf N(yti ; ik , ik ) as for basic
HMMs. The CPTs U are more complex and of larger size than the
CPTs A: The CPTs U are a set of Q matrices (one for each value of Xt2 )
of size Q Q . Left--right transitions are allowed for state transitions
within stream 2 while ergodic transitions are allowed for state
transitions from stream 1 to stream 2 as shown in Fig. 7(a). All state
values increase through time because of the left--right constraint.
On the other hand, the value of Xt2 can be equal to, greater than or

X1

X2

XT

X1

X2

XT

Y2

YT

Y1

2 = j, X 1 = k) k, j, l [1, Q ],
Uj,k,l = P(Xt2 = l | Xt1
t

less than the value of Xt1 because of the ergodic property (all state

transitions are allowed between Xt1 and Xt2 ). In practise, the values
of Xt2 and Xt1 follow each other as can be seen in Fig. 7(b) during the
decoding of a sample digit, but without predefined order. Sample
values from CPT U of the digit four model are shown in Fig. 8. A
2 , X 1 ) only if
transition can be made to state Xt2 =8 from states (Xt1
t

2 equals 7 or 8 because of the left--right assumption. To transit

Xt1

to state Xt2 = 8, the highest transition probability is 0.9052 from

1
Y1

1
Y2

YT

X1

X2

XT

2
X1

2
X2

2
XT

Y2

YT

Y1

Y1

Y2

X1

X2

X2

X1

Y1

Y2

YT

XT

XT

YT

Fig. 6. Coupled architectures represented as DBNS: (a) state-coupled: ST_CPL, (b)

general-coupled: GNL_CPL, and (c) auto-regressive coupled: AR_CPL.

2 = 7, X 1 = 8).
states (Xt1
t

The initial state distributions 1 and 2 for the vertical and the
horizontal streams, respectively, can be written:
 1
 (k) = P(X11 = k),

k [1, Q ],

2 (j, k) = P(X12 = k | X11 = j),

k, j [1, Q ].

(4)

The conditional probability table 1 is of length d while CPT 2

is of size d d. 2 expresses interdependence of the states of the
horizontal stream and those of the vertical stream, at t = 1.
Starting from the previous architecture, the second coupled architecture is obtained by adding an edge from hidden states of
the horizontal stream Xt2 to observation variables of the vertical

stream Yt1 . This architecture is called the general coupled model

(GNL_CPL, see Fig. 6b). In the GNL_CPL model, the importance of
vertical observations is stressed as they are controlled by the states
of both horizontal and vertical streams in the same time slice. The
mathematical form of CPTs A, U and  are identical for ST_CPL and
GNL_CPL. The difference lies in the Gaussian CPDs which are
written:
1 1
b (y ) = P(Yt1 = yt1 | Xt1 = j, Xt2 = k)

j,k t
= N(yt1 ; 1j,k , 1j,k ),
(5)

2 2
2
2
2
2
2
2
bk (yt ) = P(Yt = yt | Xt = k) = N(yt ; k , k ).
The form of the distribution of the horizontal observations is the
same as in the previous ST_CPL model. The distribution of vertical
observations b1 is a single pdf with mean 1j,k of length d and d d

covariance matrix 1j,k . But one needs Q Q mean vectors and

covariance matrices to account for the cartesian product of states

to describe b1 , and only Q mean vectors and covariance matrices
to describe b2 .
Last, we construct an auto-regressive coupled architecture
(AR_CPL) by coupling the vertical and horizontal AR HMMs as
shown in Fig. 6c. Parameters for this model are CPTs A, U and 
which are identical for all coupled models. The CPDs are represented by two Gaussian pdfs, which are defined for each stream i
and for t  2 by
 i i i
i
i
Bk (yt , yt1 ) = P(Yti = yti | Yt1
= yt1
, Xti = k) for i = 1, 2
(6)
i
i
i
i
= N(yt ; k + Wk yt1 , ik ).

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

3097

Fig. 7. (a) Types of transitions (ergodic or left-right) allowed between the different state variables. (b) Example of state value sequences for both streams (horizontal and
vertical) during decoding of digit 4. Horizontal and vertical state values increase through time (left--right assumption) but at time t the value of the horizontal state can be
equal, greater or less than the value of the vertical state.

Table 1
Space and time complexity of independent and coupled models as functions of Q
(number of states) and d (length of observation vectors)

Fig. 8. Sample values from CPT U which includes state transition probabilities to
2
and Xt1 .
state Xt2 from states Xt1

As in the case of AR independent models, the mean associated to

i
the current state k is shifted by Wki yt1
according to the previous

i
and the regression matrix W. This model benefits
observation yt1
from both the predicting abilities of AR- models and the fusion of
observations performed by the coupling through states.

3.3. Complexity
The above architectures, associated to their parameters (CPTs and
CPDs) provide character models. In order to limit the number of parameters and to follow the HMM paradigm, we assume that the matrix A is left-right for all models. We also assume that for each CPT
U (which can couple up to three states together), state transitions
within the same stream are left--right whereas state transitions between different streams can be ergodic (see Section 3.2). Covariance

Model

Cov + Mean

Single HMM
Single AR-HMM
ST_CPL
GNL_CPL
AR_CPL

Qd2 + Qd
Qd2 + Qd
2(Qd2 + Qd)
(Q 2 + Q)(d2 + d)
2(Qd2 + Qd)

Q
Q
Q
Q
Q

U
1
1
1
1
1

W
Qd2

2Q 2
2Q 2
2Q 2

2Qd2

Decoding
O(Qd)
O(Qd)
O(Q 2 d)
O(Q 2 d)
O(Q 2 d)

matrices for Gaussian pdfs may be full matrices. The space complexity for all models is given in Table 1 as a function of the common
number of states Q and of the length d of observation vectors. Because of character size normalization, the length of the observation
sequence is T = d. Since the number of states Q is inferior to the
length of the observation sequences T, Q < T as in classical HMMs,
the coupled model with lowest complexity is ST_CPL: its complexity
is of order O(Qd2 ), similar to that of the AR_CPL model. Though the
AR-coupled model has the most dependence between observations,
the GNL_CPL model is the one with highest space complexity: its
complexity is of order O(Q 2 d2 ), since the dimension of the space of
conditioning states for the observations {Yt1 } is Q Q in this case. The
computational time complexity is dominated by inference. Indeed,
inference complexity depends both on the size of the cliques in the
junction tree and on their number [23,30]. Since all coupled models
share the same number of cliques, only clique sizes may be different. Time complexity is of order O(TQ p+1 ) where p is the maximum
number of parents for hidden state variables in the original graph.
This complexity is reduced by a factor Q in our models because the
state space is reduced in the cliques which include the hidden state
variables: there is always one hidden variable related to its parent in a left--right transition. Inference time complexity is shown in
Table 1 for the decoding (likelihood of an observation set) of a single
character. Our time estimation does not include the computation of
the observation pdfs for all states in each time-slice.

3098

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

4. Datasets and training

We apply the DBN architectures to broken character recognition.
We first consider artificially broken handwritten digits. Breaks are
created within characters according to a degradation model. We then
consider real degraded characters. These characters are extracted
from an historical printed book [31] and are naturally broken due to
ink fading. This section describes the two datasets and the training
process.
4.1. Artificially degraded handwritten digits
We start from the MNIST database of handwritten digits [32]
which provide separate training and test sets. A training set of 5000
digits is used to train DBN models (see Section 4.3) and the test set includes 10,000 samples. Degradations are obtained by creating breaks
within digit strokes. The degradation model we propose shares some
similarities with the process related to the `sensitivity' parameter of
Baird's image defect model [33]. Random pixel values are added to
original ones within a 5 5 window. Window position is randomly
chosen, following a uniform distribution in the 28 28 character image. If the resulting window is centered on a background pixel, the
nearest writing pixel is searched and the window is moved toward
this pixel. The values added to each pixel within the window are
distributed according to a Gaussian pdf, with mean  and standard
deviation . The number of windows applied to each character is w.
In the following experiments, we set  = 0.015,  = 0 and w = 0, 1
or 2. The value  = 0 corresponds to changing the pixels within the
window to background pixels as normalized pixel values vary from
0 to 1. The value w = 0 corresponds to the original handwritten digits. When w = 1, one break is created and when w = 2, two breaks
are created within digit strokes. Fig. 9 shows samples of original and
degraded characters.

Fig. 9. Pairs of original (left) and degraded (right) characters. Two breaks are created
within each digit (w = 2).

4.2. Real degraded old printed characters

The set of old printed characters is extracted from the British Library's collection of digitized Renaissance festival books [31]. This
collection describes the ceremonies that took place in Europe between the 15th and 17th centuries. The book we have selected is
written in French and describes the reception in 1636 of the Duke
of Parma in Fontainebleau by King Louis XIII. It was printed in 1656
in Paris and written in Roman type. The set of lowercase characters
then included long s which were used instead of usual `s' if occurring at the beginning or in the middle of a word. Characters `j' and
`k' were not in use and v was often printed instead of u. There were
also many ligatures such as (long s + t, or two long s) as can be seen
in the sample document in Fig. 10.
Characters from seven pages were extracted and manually labeled. It should be noted that ligature characters such as `fi' (f + i),
`long s t' (long s + t), etc. were considered as single characters
and were assigned to additional classes. The first five pages were
standard pages, well contrasted, whereas the other two were degraded, including many broken characters due to ink fading. This led
to two sets of characters: a standard set including 2796 characters
from the standard pages and a degraded set including 1216 characters from the degraded pages. Characters were then binarized, normalized to size 20 20 and placed in 28 28 images.2 It should
be noted that character normalization and image size follow the
MNIST paradigm [32] so that white borders are added around character images. In word recognition methods such as in [34], white

2 We intend to make this database publicly available (with permission of the

British Library).

Fig. 10. Sample document from the Renaissance Festival Books collection.

borders are added to training characters in order to deal with intraword spaces. In other cases, white borders can be removed as well
as the states representing each border: models with lower complexity are then obtained by reducing the number of states from Q
to Q 2.
Because some classes had very few samples, we selected the
classes which had enough samples (around 50) within the first three
pages dedicated to training. This led to 16 classes: a, b, c, d, e, i, l,
m, n, o, p, r, s, long s, t and u. Sample characters from standard and
degraded sets are shown in Fig. 11.

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

3099

Fig. 11. Sample old printed characters from standard and degraded pages.

Vertical and Horizontal ARHMMs

100

95

95
Recognition rate (%)

Recognition rate (%)

Vertical and Horizontal HMMs

100

90
85
80

vertical
horizontal

75
70
65

90
85
80

vertical
horizontal

75
70

10

14

18

65

22

Coupled DBNs

18

22

100
Recognition rate (%)

Recognition rate (%)

14

AR Coupled DBN

100

95

90
STCPL
GNLCPL

85

80

10

Number of states

Number of states

10

14

18

22

Number of states

95

90

85

80

ARCPL

10

14

18

22

Number of states

Fig. 12. Performance of independent and coupled models according to the number of states.

4.3. Training and recognition

Observation sequences are obtained by scanning character
images from left to right and top to bottom. Characters are first
preprocessed by a 3 3 Gaussian mask with standard deviation
0.5. The resulting pixel values are then normalized in [0., 1.]. Two
observation sequences of length T = 28 are obtained from the respective vertical and horizontal streams. All character models share
a single DBN architecture but their parameters differ for each class.

Parameters are learnt using the EM algorithm and inference. For

independent HMMs, observation parameters are initialized by assigning observations to states linearly. For AR models (independent
and coupled), observation parameters are initialized randomly. For
all other models, observation parameters are initialized to a common value for all states and each stream i.e, the empirical mean and
covariance matrix obtained from the sample data.
The common number of states for hidden variables is denoted
by Q. We study the effect of varying Q on the digit recognition task

3100

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

with a training set of 4000 digits and a test set of 1000 digits. The
value of Q ranges from 6-state models to 22-state models. Results
in Fig. 12 show that recognition performance increases with Q until
reaching a maximum for Q = 14 or 18. There is no improvement for
values of Q > 18 and a slight improvement is obtained for coupled
models by increasing Q from Q =14 to 18. The price for this improvement is higher space and time complexity: in the following, we set
Q = 14 which offers the best compromise between complexity and
performance.
Fig. 12 also shows that the best performances are always reached
by the AR_CPL model whatever the number of states Q. This shows
the superiority of the AR_CPL model over all independent and other
coupled models (see also Section 5.1).
For training digit models, we used a subset S of 5000 samples (500
per class) from the MNIST training database. Then, we conducted
cross validation experiments in the following way: the subset S was
split into F = 5 sets of 1000 characters. Each DBN architecture was
trained on F 1 sets and tested on the remaining set, F times. Within
each model, the parameters yielding the best cross-validation recognition performance were selected for testing. Testing was performed
on the 10 000 digits of the MNIST test set.
For training old printed character models, 50 characters per class
were selected from the first three standard pages. Then character
models were tested on two test sets: a standard test set (test-s) from
the two remaining standard pages and a degraded test set (test-d)
from the two degraded pages. The standard and degraded test sets
include 1009 and 1079 characters, respectively, for the 16 classes
considered.
During recognition, each character was assigned to the class with
the highest log-likelihood value. We use the BayesNet toolbox [35]
which provides general MatLab source code for training and inference in static and dynamic Bayesian networks.
5. Experimental results
We evaluate the various DBN architectures, independent and
coupled, for the problem of the recognition of degraded characters.
Two off-line recognition tasks are considered. We first evaluate DBN
architectures for the recognition of artificially broken handwritten
digits: their robustness to degradation is evaluated against different degradation parameters. We also test these architectures for the
recognition of real degraded printed characters.
5.1. Handwritten digits
5.1.1. Comparison between independent and coupled architectures
Independent and coupled architectures are first tested on the set
of artificially broken digits. We consider three levels of degradation:
no additional degradation using the original MNIST test set (w = 0),
one break created within digits (w=1) and two breaks created (w=2).
Recognition accuracies for each model are given in Table 2.
For each degradation level, vertical models perform better than
horizontal ones within each type of independent models (HMM/AR).
This means that columns of character images are more discriminating than rows for handwritten digits. This is also observed for
old printed characters: there is a predominance of vertical strokes
for these forms of letters and digits [28]. Comparing between independent models, the auto-regressive vertical model performs better
than the basic vertical-HMM. This is due to the fact that the basic
HMM assumes conditional independence of observation variables
with respect to hidden states, whereas the AR model assumes explicit dynamic dependence between observations. For horizontal independent models, performances of the basic horizontal-HMM and
the horizontal-AR model are comparable. The prediction of image
rows is less efficient than the prediction of image columns.

Table 2
Recognition rates (%) for handwritten digits under different levels of degradation
(w = 0: no additional degradation, w = 1: one break, w = 2: two breaks)
Model

w=0

w=1

w=2

Vertical-HMM
Horizontal-HMM
Vertical-AR
Horizontal-AR
ST_CPL
GNL_CPL
AR_CPL
Combination of HMM scores
Combination of AR-HMM scores
SVM

90.2
87.4
93.2
87.7
92.4
93.4
94.9
93.1
94.7
96.1

86.9
82.8
89.8
81.6
90.8
90
93.4
90.6
91.9
91.1

83.8
75.3
85.3
75.6
87.4
86.2
90.9
87
89
85.4

Our results show that coupled models perform significantly better

than basic HMMs. Although the horizontal stream is less reliable, its
coupling with the vertical stream improves any corresponding single
stream representation.
The general coupled model (GNL_CPL) differs from the other coupled models because it uses state--observation relations in addition
to state--state relations to express the interdependence of streams.
This model requires more observation parameters, but provides
little recognition improvement compared with ST_CPL. Achievement of coupling between streams using state--state relations as in
ST_CPL and AR_CPL, rather than state--observation relations as in
GNL_CPL, leads to more efficient models in terms of complexity and
performance. Last, the AR coupled (AR_CPL) model emphasizes the
importance of observations through dynamic linking in time.
Coupled architectures behave better than any independent HMM
(basic and AR) as the level of degradation increases (w = 1 and 2).
Moreover the auto-regressive coupled architecture performs best.
This is because missing observations (such as in broken characters)
may be predicted through auto-regressive models. Coupled architectures may also include at least one uncorrupted stream, horizontal
or vertical, within each time slice and thus better cope with missing
information.
5.1.2. Comparison with the combination of HMM scores
We also compared coupled models with the weighted combination of HMM scores. The combined score for a pattern given a class
model results from the weighted sum of the log-likelihoods (scores)
provided by each HMM, vertical and horizontal. The weights i , i=1, 2
must satisfy the constraints: i  0 and 2 = 1 1 . Thus, only the
value  = 1 dedicated to the vertical HMM needs to be optimized.
We search for the optimal  on a validation set of 1000 digits. Fig. 13
shows recognition rates versus  for the combination of basic HMMs.
For digits, the maximum is reached for  = 0.5. We have observed
that log-likelihoods provided by the vertical HMM are in average
higher than those provided by the horizontal HMM. Consequently,
 = 0.5 gives more weight to the vertical HMM.
Results for the test set using the optimal value of  are given
in Table 2. The AR-coupled model outperforms the combination of
HMM scores whatever the level of degradation. When w = 0 (no
degradation), the combination of HMM scores performs better than
the state-coupled model and worse than GNL_CPL and AR_CPL models. When the level of degradation increases (w = 1, w = 2), both the
ST_CPL and the AR_CPL models perform better than the combination
of HMM scores. The performance of individual HMMs as well as the
linear combination of them, deteriorates more rapidly as degradation
increases than when they are combined in a coupled DBN model.
5.1.3. Comparison with the combination of auto-regressive HMM scores
We can also compare the AR_CPL model with the weighted combination of auto-regressive HMMs. As previously, the combined score
for a pattern given a class model results from the weighted sum of

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

Table 3
Recognition rates (%) for standard and degraded old printed characters

100

Recognition rate (%)

3101

95

90

Model

Standard (test-s)

Degraded (test-d)

Vertical-HMM
Horizontal-HMM
Vertical-AR
Horizontal-AR
ST_CPL
GNL_CPL
AR_CPL
Comb. of HMM scores
Comb. of AR-HMM scores
SVM

98.3
93.7
97.9
96.2
98.7
98.6
98.8
98.4
98.7
98.4

93.8
88.1
94.5
91.2
95.5
94
96
95.4
95.5
94.9

old printed characters

handwritten digits

100

85
0.2

0.3

0.4 0.5 0.6

weight factor

0.7

0.8

0.9

1
96

100
98

Comb. ARHMM scores

ARCPL

96

Recognition rate (%)

0.1

Fig. 13. Combination of HMM scores: recognition rate versus weight factor .

Recognition rate (%)

Comb. ARHMM scores

ARCPL

98

94
92
90
88
86

94

84

92

82

90

80
tests

88
86

testd
Degradation Level

testh

Fig. 15. Comparison of the AR_CPL model with the combination of AR-HMM scores
for old printed characters and several degradation levels.

84
82
80
0

1
2
Degradation Level

Fig. 14. Comparison of the AR_CPL model with the combination of AR-HMM scores
according to degradation.

the scores provided by each AR-HMM, vertical and horizontal. The

optimal weight , searched on the validation set, is equal to  = 0.45
for the AR-HMM combination. Recognition accuracies are given in
Table 2. The AR-coupled model outperforms the combination of
AR-HMM scores whatever the level of degradation. The improvement
brought by the AR_CPL model is enhanced for degraded characters
(w > 0). This is also observed in Fig. 14 where the performances of
the AR_CPL model and the combination of the AR-HMMs scores are
compared for w ranging from w =0 to 3. An additional level of degradation is provided here: when w = 3, three breaks are created within
characters. Results in Fig. 14 show that the improvement brought by
the AR_CPL model increases as the level of degradation w increases.
5.2. Old printed characters
5.2.1. Model comparison
Recognition accuracies for old printed characters are given in
Table 3. Similarily to handwritten digits, the horizontal stream is less
reliable but its coupling with the vertical stream increases performance. The vertical-AR model shows some advantage over the basic
vertical-HMM on the set of degraded characters but performances
are comparable on the standard set. The GNL_CPL model deteriorates

more rapidly on the degraded set because defects in the vertical observation disturb both horizontal and vertical state sequences.
The auto-regressive coupled model (AR_CPL) always performs
better than independent models, the state-coupled model and the
combination of HMM scores for which the optimal value of  = 0.65
was found on a validation set of 1038 characters (Fig. 13). As before,
the ST_CPL and AR_CPL coupled architectures better cope with degraded characters (test-d) than independent HMMs (basic and AR)
and the combination of HMM scores.
5.2.2. Comparison with the combination of auto-regressive HMM scores
For the combination of auto-regressive HMMs (AR-HMMs), the
optimal weight  searched on the validation set is equal to  = 0.4.
Recognition accuracies are given in Table 3. The AR_CPL model performs better than the combination of AR-HMMs whatever the level
of degradation. However, several classifiers have high performance
on the set of non-degraded characters (test-s): coupled DBN classifiers and the combination of AR-HMM scores perform accurately on
such characters. The improvement brought by the AR_CPL model is
enhanced for degraded characters. To highlight this property, an additional set of highly degraded characters is provided by lowering
the binarization threshold of degraded characters by 20%. This leads
to the test h set (highly degraded) including highly fading characters. Recognition accuracies are compared in Fig. 15 for all test sets.
The improvement brought by the AR_CPL model over the combination of AR-HMM scores increases as degradation increases as seen
previously for handwritten digits (see Section 5.1.3).
When characters are broken due to natural fading process or
low binarization threshold, the AR_CPL model performs better than
the other models. This coupled architecture is thus particularly

3102

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

convenient for recognizing old printed characters since broken

characters are often found in old printed books.
5.3. Comparison with SVM classifier
However, higher accuracies can be achieved on the MNIST digit
database with discriminative classifiers such as SVMs as reported in
[36,37]. We compare below the influence of defects such as broken
characters on DBN and SVM classifiers respectively. The SVM classifier is implemented with the LIBSVM toolbox [38] with a RBF kernel
and parameters C = 26,  = 25 as suggested in [37]. SVM recognition
accuracies are given in Tables 2 and 3 for handwritten digits and old
printed characters respectively, under different levels of degradation.
For handwritten digits and without any additional degradation
(w = 0), the SVM classifier outperforms other classifiers. When the
level of degradation increases (w = 1), the AR-coupled model outperforms the SVM classifier. But all coupled models, outperform the
SVM classifier in case of high degradation (w = 2).
For old printed characters (see Table 3), the SVM classifier obtains slightly lower performances than coupled architectures on the
standard data set. On the degraded set (test-d) which includes many
broken characters, SVM performance decreases significantly more
than that of coupled architectures ST_CPL and AR_CPL. Also, performances of these coupled architectures remain higher than that of
the SVM on the degraded test set. This shows that state-coupled and
auto-regressive coupled architectures are more robust to degradation than the SVM classifier in case of highly broken characters.
6. Conclusion
We have presented a new approach for off-line character recognition, based on DBN. The modeling consists of coupling two HMMs
in various DBN architectures. The observations for these HMMs are
the image rows and the image columns, respectively. Interactions
between rows and columns are modeled through state transitions or
state/observation transitions. This results in finer representations of
character images and in improvement of the basic HMM framework.
We first investigated independent HMM and AR models. We
showed that vertical models perform better than horizontal ones
since columns of character images are more discriminating than
rows. Secondly, we coupled these independent models into single
models providing better performance than for the non-coupled models, as well as for the combination of the scores of the independent
HMMs. We also demonstrated that the coupling through states such
as in ST_CPL is more efficient than the coupling from state to observation as in GNL_CPL. The AR-coupled architecture which dynamically links observations in time gives the best recognition results.
We applied this approach to the recognition of handwritten digits and old printed characters. We demonstrated the robustness of
this approach in the presence of artificial and real world degradations. Our experiments show that coupled architectures cope better
with highly broken characters than both basic HMMs and discriminative methods like SVMs. This is because coupled architectures are
able to predict missing information and may provide at least one
uncorrupted stream within time slices.
The proposed coupled DBN architectures are thus particularly
efficient for the recognition of broken characters. We expect further
improvements from an accurate initialization of the parameters.
Acknowledgements
The authors wish to thank the reviewers for their constructive
comments. They are also grateful to Chafic Mokbel from Balamand
University and Franck Lebourgeois from INSA Lyon for fruitful discussions.

References
[1] L.R. Rabiner, A tutorial on hidden Markov models and selected applications in
speech recognition, Proc. IEEE 77 (2) (1989) 257--286.
[2] R. Plamondon, S. Srihari, On-line and off-line handwriting recognition: a
comprehensive survey, IEEE PAMI 22 (1) (2000) 63--84.
[3] C. Bahlmann, H. Burkhardt, The writer independent online handwriting
recognition system frog on hand and cluster generative statistical dynamic time
warping, IEEE PAMI 26 (3) (2004) 299--310.
[4] M. Schenkel, M. Jabri, Low resolution degraded document recognition using
neural networks and hidden Markov models, Pattern Recogn. Lett. 19 (1998)
365--371.
[5] A. Senior, A. Robinson, An off-line cursive handwriting recognition system, IEEE
PAMI 20 (3) (1998) 309--321.
[6] A. Vinciarelli, S. Bengio, H. Bunke, Offline handwriting recognition of
unconstrained handwritten texts using HMMs and statistical language models,
IEEE PAMI 26 (6) (2004) 709--720.
[7] J.-C. Anigbogu, A. Belaid, Recognition of multifont text using Markov models,
In: Proceedings of the Seventh Scandinavian Conference on Image Analysis,
Aalborg (Denmark), 1991, pp. 469--476.
[8] H.S. Park, S. Lee, Off-line recognition of large-set handwritten characters with
multiple hidden Markov models, Pattern Recogn. 31 (1998) 1849--1864.
[9] N. Arica, F.T. Yarman-Vural, Optical character recognition for cursive
handwriting, IEEE PAMI 24 (6) (2002) 801--813.
[10] H. Baird, The state of the art of document image degradation modeling, in:
Proceedings of the Fourth Workshop on Document Analysis Systems, DAS, Rio
de Janeiro, 2000, pp. 1--16.
[11] A. Whichello, H. Yan, Linking broken character borders with variable sized
masks to improve recognition, Pattern Recogn. 29 (8) (1996) 1429--1435.
[12] B. Allier, N. Bali, H. Emptoz, Automatic accurate broken character restoration
for patrimonial documents, IJDAR 8 (4) (2006) 246--261.
[13] A. Antonacopoulos, D. Karatzas, Document image analysis for World War II
personal records. In: First International Workshop on Document Image Analysis
for Libraries, DIAL 04, Palo Alto, 2004, pp. 336--341.
[14] M. Droettboom, Correcting broken characters in the recognition of historical
printed documents, in: Proceedings of Joint Conference on Digital Libraries,
JCDL'03, 2003.
[15] J. Kittler, M. Hatef, R. Duin, J. Matas, On combining classifiers, IEEE PAMI 3 (20)
(1998) 226--239.
[16] W. Wang, A. Brakensiek, G. Rigoll, Combining HMM-based two pass classifiers
for off-line word recognition, in: Proceedings of ICPR, Quebec, 2002,
pp. 151--154.
[17] A.J. Elms, S. Procter, J. Illingworth, The advantage of using an HMM based
approach for faxed word recognition, IJDAR 1 (1998) 18--36.
[18] K. Hallouli, L. Likforman-Sulem, M. Sigelle, A comparative study between
decision fusion and data fusion in Markovian printed character recognition, in:
Proceedings of ICPR, Quebec, 2002, pp. 147--150.
[19] X. Xiao, G. Leedham, Signature verification using a modified Bayesian network,
Pattern Recogn. 35 (2002) 983--995.
[20] S. Cho, J. Kim, Bayesian Network modeling of hangul characters for on-line
handwriting recognition, In: Proceedings of ICDAR, 2003, pp. 297--211.
[21] R. Sicard, T. Artieres, E. Petit, Modeling on-line handwriting using pairwise
relational features, In: Proceedings of IWFHR, La Baule, 2006.
[22] J. Pearl, Probabilistic Reasoning in Intelligent Systems: Networks of Plausible
Inference, second ed, Morgan Kaufman, Los Altos, CA, 1988.
[23] G. Zweig, Bayesian network structures and inference techniques for automatic
speech recognition, Comput. Speech Language 17 (2003) 173--193.
[24] K. Daoudi, D. Fohr, C. Antoine, Dynamic Bayesian networks for multi-band
automatic speech recognition, Comput. Speech Language 17 (2003) 263--285.
[25] M. Brand, N. Oliver, A. Pentland, Coupled hidden Markov models for complex
action recognition, in: Proceedings of the IEEE Conference CVPR 97, 1997,
pp. 994--999.
[26] N. Friedman, D. Koller, Being Bayesian about network structure: a Bayesian
approach to structure discovery in Bayesian networks, Mach. Learning 2001,
pp. 201--210.
[27] J.D. Hamilton, Analysis of time series subject to changes in regime, J. Econometr.
45 (1990) 39--70.
[28] C. Sirat, Handwriting and the writing hand, in: W.C. Watt (Ed.), Writing
Systems and Cognition: Perspectives from Psychology, Physiology, Linguistics,
and Semiotics, Kluwer Academic Publishers, Dordrecht, 1994, pp. 375--459.
[29] A. Tonazzini, S. Vezzosi, L. Bedini, Analysis and recognition of highly degraded
printed characters, IJDAR 6 (2003) 236--247.
[30] G. Bilmes, Dynamic Bayesian multinets, in: UAI '00: Proceedings of the
16th Conference in Uncertainty in Artificial Intelligence, Stanford, CA, 2000,
pp. 38--45.
[31] British Library, British Library Digitised Festival Books. Available on the web
at: http://www.bl.uk/treasures/festivalbooks/homepage.html.
[32] Y. LeCun, C. Cortes, The MNIST handwritten digit database, 1998. Available on
the web at: http://yann.lecun.com/exdb/mnist/.
[33] H. Baird, Document image defect models, in: H.S. Baird, H. Bunke, K. Yamamoto
(Eds.), Structured Document Image Analysis, Springer, New York, 1992,
pp. 546--556.
[34] S. Procter, A.J. Elms, J. Illingworth, A method for connected hand-printed
numeral recognition using hidden Markov models, in: IEE European Conference
on Handwriting Analysis and Recognition, Brussels, 1998.

L. Likforman-Sulem, M. Sigelle / Pattern Recognition 41 (2008) 3092 -- 3103

[35] K. Murphy, BayesNet Toolbox for Matlab, 2003. Available on the web at:
http://www.ai.mit.edu/murphyk/Bayes/bnintro.html.
[36] C.-L. Liu, K. Nakashima, H. Sako, H. Fujisawa, Handwritten digit recognition:
benchmarking of state-of-the-art techniques, Pattern Recogn. 36 (2003)
2271--2285.

3103

[37] K.-M. Lin, C.-J. Lin, A study on reduced support vector machines, IEEE Trans.
Neural Networks 14 (2003) 1449--1559.
[38] C.-C. Chang, C.-J. Lin, LIBSVM: a library for support vector machines, 2001.
Software available at: http://www.csie.ntu.edu.tw/cjlin/libsvm.

About the Author---LAURENCE LIKFORMAN-SULEM is graduated in engineering from ENST-Bretagne (Ecole Nationale Suprieure des Tlcommunications) in 1984,
and received her PhD from ENST-Paris in 1989. She is Associate Professor at TELECOM Paris Tech (former ENST) in the Department of Signal and Image Processing where
she serves as a senior instructor in Pattern Recognition and Document Analysis.
Her research area concerns document analysis dedicated to handwritten and historical documents, document image understanding and character recognition. Laurence
Likforman and co-researchers won the first place at the ICDAR'05 Competition on Arabic handwritten word recognition.
Laurence Likforman is a founding member of the francophone GRCE (Groupe de Recherche en Communication Ecrite), association for the development of research activities
in the field of document analysis and writing communication. She chaired the program committee of the last CIFED (Conference Internationale Francophone sur l'Ecrit et le
Document) held in Fribourg (Switzerland) in 2006.
About the Author---MARC SIGELLE was born in Paris in 1954. He graduated from Ecole Polytechnique Paris in 1975 and from Ecole Nationale Suprieure des
Tlcommunications Paris in 1977. In 1993 he obtained a PhD from Ecole Nationale Suprieure des Tlcommunications. He worked first at Centre National d'Etudes des
Tlcommunications in Physics and Computer algorithms. Since 1989 he has been working in image and more recently in speech processing at Ecole Nationale Suprieure
des Tlcommunications.
His main fields of interests are restoration and segmentation of signals and images with Markov Random Fields (MRFs), hyperparameter estimation methods and relationship
with Statistical Physics. His work has been first devoted to blood vessel reconstruction in angiographic images and then to the processing of remote sensed satellite and
synthetic aperture radar images. His most recent interests deal with a MRF approach to image restoration using level sets for Total Variation and its extensions. He is also
devoted to speech and character recognition using MRFs and Bayesian Networks. M. Sigelle is IEEE Senior Member since Fall 2003.