Sparse codes for natural images

SPARSE CODES FOR NATURAL IMAGES
Davide Scaramuzza
(davide.scaramuzza@epfl.ch)
Autonomous Systems Lab (EPFL)
Final rapport of Wavelet Course MINI-PROJECT (Prof. Martin Vetterli)
ABSTRACT
The human visual system, at the primary cortex, has
receptive fields that are spatially localized, oriented
and bandpass. It has been shown that a certain
learning algorithm to produce sparse codes for natural
images leads to basis functions with similar properties.
This learning algorithm optimizes a cost function that
trades off representation quality for sparseness, and
searches for sets of natural images, which basis
functions lead to good sparse approximations. The
result of the learning algorithm is a dictionary of basis
functions with localization in space, direction and
scale.
In this paper, dictionaries for different set of images
are showed and their own properties are described
and verified. It will be showed that the learning
algorithm leads to overcomplete bases functions that
“capture” the intrinsic structure of the images. This
allows efficient coding of the images with good
representation quality. The results are applied to
image approximation and denoising.
1. INTRODUCTION
In 1962 and ’68, the Nobel prize winning discoveries
of Hubel and Weisel showed that the mammals primary
visual cortex (called V1) consists of cells responsive to
simple and complex features in the input. For example,
most cells in the visual cortex respond best to edges at
some particular angle or the other (thus they are spatially
oriented, see Fig.1). More generally, it has been proven
that V1 has receptive fields that are characterized as being
spatially localized, oriented and bandpass; that is, they are
selective to structure of the visual input at different spatial
scales [4]. One approach to understanding such response
properties of visual neurons has been to consider their
relationship to the statistical structure of natural images in
terms of efficient coding. In [1],[2] Olshausen et al.
showed that by designing an algorithm that attempts to
find sparse linear codes for natural scenes, develops a
complete family of localized, oriented and bandpass
receptive fields, similar to those found in the primary
visual cortex.
Fig.1 Orientation columns in a patch of the monkey visual cortex,
visualized with modern imaging techniques. Colors varying from red to
violet indicate orientation preference of cells varying from zero to 180
degrees. The overlaid white lines also show the orientation preference in
each area.
2. GENERATIVE IMAGE MODEL
In [1] the authors propose that the neurons in V1 model
,the structure of images I( yx ) in terms of a linear
superposition of basis functions φ plus noise ε(x, y) :i
, ,I( yx ) = ∑i
s φ( yx ) +ε( yx ) (1)i ,
One can think to these basis functions as a simple
“features vocabulary” for describing images in terms of
additive functions. Note also that these bases are not
necessarily orthogonal.
The goal of efficient coding is to find a set of φi
that
forms a complete code (that is, spans the image space) and
results in the coefficient values being as sparse and
statistically independent as possible over an ensemble of
1

natural images. Sparseness of coefficients allows for a
small number of bases to represent an image. This
resemble what found in V1, where neurons represent data
using a small number of active units. The statistical
independence, in terms of coding, obviously reduces the
redundancy of the code, while, in neurobiological words,
translates the fact that neurons are thought as independent
of the activity of their neighbours.
One line of approach to this problem, accounting for
sparseness and independence, is based on principal
component analysis (PCA), in which the goal is to find a
set of mutually orthogonal bases functions that capture the
directions of maximum variance in the data and for which
the coefficients si are pairwise decorrelated. However,
the receptive fields (bases) that result from this process
are not localized as it would be desirable. Moreover, it
assumes that the original data are well described by a
Gaussian distribution but natural scenes contain many
higher order forms of statistical structure. In this case the
independent component analysis (ICA) would be suitable
because accounts for non-Gaussian distributions. Two
limitations common to both of these techniques are that
they do not allow for noise to be modelled separately from
the signal structure and that they do not allow for
overcomplete codes in which there are more basis
functions than input dimensions.
Overcompleteness in the representation is important
because it allows for the joint space of position,
orientation, and spatial-frequency to be tiled smoothly
without artefacts. More generally, it allows for a greater
degree of flexibility in the representation, as there is no
reason to believe a priori that the number of “causes” for
images is less than or equal to the number of pixels.
3. CODEBOOK LEARNING ALGORITHM
The strategy for inferring the codebook from a set of
natural scenes is explained in [1],[2]. It applies a Bayesian
method for deriving an optimal basis that trades off
reconstruction quality for sparseness.
Rewriting the (1) in matrix form we get:
⋅x = sA + ε (2)
where A is a LxM matrix whose columns are the basis
functions, s is an M-element vector of basis coefficients
and ε is an additive Gaussian noise.
To derive the learning algorithm for finding the basis
functions it is possible to proceed in two steps: a) finding
the expansion coefficients si , given the image and the
basis set; b) learning the bases φi given a set of training
images and correspondent coefficients. The above stages
can be iteratively repeated until A converges towards a
stable solution. Following is the explanation of
coefficients extraction and the learning rule to update the
bases functions.
a) Finding expansion coefficients
To solve this, we assume we are given an image x and
a general overcomplete basis A. The goal is to infer the
vector s. This can be done by exploiting Maximum
Likelihood Principle, that states the most probable value
s’ for s is that satisfying the following:
),|(max' AxsPs
s
= )3(
By means of Bayes’ rule:
)|(),|(),|( AsPsAxPAxsP ∝ )4(
Because the model is assumed to be Gaussian:
−
( x−As)2
(5)2σ 2
( ,xP | sA ) ∝ e
where σ is the standard deviation of the additive noise.
The last term in (4), sP | A) , is the prior probability(
distribution over the basis coefficients. If the prior is
assumed to be independent of A, then sP | A) = sP ) .( (
At this point observe that, in order to impose sparseness
of coefficients, the prior probability P(s) has to peak at
zero. The prior P(s) chosen by the authors is the Laplacian
distribution (Fig 2):
−ϑi |s|
(6)(sP i ) ∝ e
)( isP
Fig.2 The probability distribution of the coefficients is peaked at zero.
Such a distribution would result from a sparse activity distribution over
the coefficients.
And by imposing statistical independence among
coefficients:
( −ϑT
|s| (7)(sP ) = ∏ sP i ) ∝ e
i
After substituting (4,5,6,7) in (3) we have:
2

2
+ϑT ⎞ (8)s' = max AxsP ) = min
⎛
⎜
1
( | , x − As s ⎟2s s
⎝ 2σ ⎠
Thus, in order to find the coefficients si for each image
presentation, (8) has to be minimized with respect to s.
b) Learning basis functions
The goal in learning basis vectors is to obtain a good
model of the distribution of natural images. That means,
finding a basis matrix which can optimally fit all image
presentations. In order to do this, what we want is to
maximize the average probability distribution aver as
many images as possible. The cost function to be
maximized is defined with respect to A is:
(=Γ log xP | A) (9)
where the distribution xP | A) is obtained by
marginalization:
( ( , ) ( )
(
xP | A) ∫= xP | dssPsA (10)
The maximization problem can be solved by gradient
ascent. This leads to the following updating term for the
basis matrix A:
while the attenuation at high spatial-frequencies
eliminates the artifacts of rectangular sampling.
Fig.3 The 2D Fourier transform of the zero-phase whitening lowpass
filter used to preprocess the natural pictures.
After these two necessary steps (random initialization of
A and preprocessing), I run the algorithm separately on
different set of pictures.
Let us suppose to have chosen to learn a complete
codebook of 64 bases, each one of dimension 8x8 pixel.
Then, in order to compute the updating rule in (11), the
algorithm randomly selects a large number (actually 100)
of 8x8 image patches from the training set.
⋅∆A ∝ se T
(11)
Where the error e = . Details about the learningx − As
algorithm can be found in [1],[2] and [5].
4. LEARNING CODES FOR NATURAL SCENES
I applied the learning algorithm proposed by B. A.
Olshausen to different sets of natural scenes (pictures of
nature, pictures of buildings in Amsterdam and Lausanne
and paintings of Van Gogh). I also tested it by varying the
number of bases (from complete to overcomplete
representations) and their dimensions.
Before starting the algorithm, the initial values for
basis functions were generated to random values evenly
distributed between [-1,1].
Then, the training set was preprocessed, following the
suggestions of Olshausen [1], by filtering all the grey-
scale images with a zero-phase whitening/lowpass filter
having the following 2D Fourier transform:
−
⎛ f ⎞
4
⎜ ⎟
⋅ ⎝ 200 ⎠
( fR ) = ef (12)
where f = fx2
+ fy2 . See also Fig.3. This is because
whitening filter is able to counteract the fact that the error
computed at (11) preferentially weights low frequencies,
A stable solution arrived after about 10,000 – 40,000
updates, that is, 1000,000 – 4000,000 image presentations
(between 1 and 4 hours of computation for each set of
natural scenes).
5. RESULTS OF LEARNING
The learning algorithm was applied to three sets of ten
512x512 pixel images taken from natural surroundings
(Fig.4), Van Gogh paintings (Fig.5) and Amsterdam and
Lausanne buildings (Fig.6). The results of the learned
codebooks are also shown there.
I derived different codebooks distinguishable for the
number of basis functions used (between 64 and 192) and
for basis dimensions (between 8x8 and 16x16 pixels).
The results show, as wanted, the basis patches to be well
oriented and spatially localized. Moreover, they seem to
“capture” the intrinsic structure of the pictures. For
instance, in the case of Van Gogh paintings, the extracted
bases resemble the brushstrokes of the painter, while, in
the pictures of cities, they capture the basic elements of
the buildings, mainly composed of vertical and horizontal
edges and corners.
3

Fig.4 Results from training a system of 192 bases functions on 16x16 image patches extracted from scenes of nature (here only several bases are
displayed for reasons of space). The red squares are displayed just for explaining the learning procedure, by randomly selecting image patches from the
training set. The scenes were ten 512x512 images of natural surroundings in the American northwest, preprocessed by the mentioned zero-phase
whitening/lowpass filter. The results shown were obtained after 40,000 iteration steps (4 hours of computation). Note well that, as desiderated, the learned
bases result to be oriented along specific directions and spatially well localized.
a) b) c)
Fig.5 Results from training a 2x-overcomplete system of 128 bases functions of 8x8 pixels (b) and a system of 192 bases of 16x16 pixels (c) (in the latter
only several bases are shown). The codebooks were extracted from ten 512x512 pixel images of Van Gogh paintings (a), preprocessed, as usual, by the
zero-phase whitening/lowpass filter. The results shown were obtained after 20,000 (b) and 40,000 (c) iteration steps (2-4 hours of computation
respectively). Note also here the learned bases result to be oriented along specific directions and spatially well localized. Moreover the bases seem to
capture the intrinsic structure of Van Gogh brushstrokes (this is well visible in (c)).
4

Fig.6 Results from training a complete system of 64 bases functions of 8x8 pixels. The codebook was extracted from ten 512x512 pixel images Lausanne
and Amsterdam buildings. The results shown were obtained after 20,000 iteration steps (2 hours of computation). Note also here the learned bases result
to be oriented along specific directions and spatially well localized. Moreover the bases seem to capture the intrinsic structure of the building elements,
that result to be mainly composed of vertical and horizontal edges and corners.
As mentioned at the beginning, we are also interested in
verifying the spatial-frequency localization of basis
functions. This can be easily seen by analyzing the bases
in the 2D Fourier Transform domain (see Fig.7).
Fig.7 The Power spectrum of basis functions correspondent to Fig.5.
Fig.7 shows that bases functions are well localized in the
frequency domain. Another point of interest, not visible
in Fig.7, is that the basis functions largest in magnitude
also have the lowest peak spatial-frequency tuning. The
latter is just what would be found by using PCA
technique, which assumes that the data have Gaussian
structure. This could reflect an attempt of the model to
capture small-amplitude noise in the images. To see
better the frequency localization characteristics, the next
figures show the polar plots of the peak spatial
frequency tuning of all bases.
Figures 8 and 9 show the frequency tiling of basis
functions for different degrees of completeness. You can
note that increasing the number of bases results in a
denser tiling of frequency space (and also in a denser
tiling of orientations and positions). Moreover, this
frequency tiling is higher at mid-high frequencies (to see
this, look at frequency histograms in Fig. 11).
One trend that appears immediately evident is that
the preferred orientation tends to align vertically and
horizontally, but Olshausen adverts that this is an
artefact due to having used a rectangular sampling grid
to digitize the images, rather than a reflection of an
intrinsic property of the images themselves.
Conversely, for the bases learned from pictures of
buildings, basis functions result to be disposed along a
few preferential directions of frequency space (Fig.10).
This strongly support the mentioned conjecture that they
try to capture the intrinsic structure of original images
they come from, that are mainly composed of vertical,
horizontal and slanting edges.
5

Fig.12 Histogram of overall
coefficients and the procedure
the frequency domain (red lines). These preferential directions are due0.35 0.35
to the localized orientation of the correspondent bases in the spatial
0.3 0.3
domain. As seen in Fig.6, they undergo the direction of horizontal,
vertical and slanting edges that frequently occur in the pictures.0.25 0.25
0.2 0.2
18 20
1816
0.15 0.15
14
16
140.1 0.1
12
12
10
0.05 0.05
10
8
8
0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 6
6
4
4
2
a) b)
2
Fig.8 Spatial-frequency localization of bases functions in natural 0 0
0
surroundings of Fig.4. The circle diameter is proportional to the a)
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
b)20
18
bandwidth. a) Complete system of 64 bases of 8x8 pixels, b) System of
192 bases of 16x16 pixels. Note the denser tiling of frequency space by
16
increasing the number of bases (b).
14
12
0.4
10
8
0.45
0.4 0.35
6
40.35
0.3
2
0.3 c)0.25 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
0.25
0.2
Fig.11 Histograms of peak spatial-frequency bandwidths for the0.2
0.15
complete 64 learned basis function of natural surroundings (a), Van0.15
Gogh paintings (b) and pictures of buildings (c). Note the higher0.1
0.1
density of bases at mid-high frequencies.
0.05
0.05
00
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
6. RECONSTRUCTION FROM BASES
0.4
b) FUNCTIONSa)
0.35
By using the generative model defined in (1), each
single image may be simply reconstructed as a linear
0.3
0.25
combination of learned basis vectors. Given the0.2
probabilistic nature of the approach, this is not a perfect
0.15
reconstruction of the original picture but, conversely, is
0.1
its best approximation by means of the learned bases.
0.05
1800
-3 -2 -1 0 1 2 3 4
c) Remember that the
16000
learning algorithm0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
1400
Fig.9 Spatial-frequency localization of bases functions in Van Gogh searches for those
1200
paintings (Fig.5). a) Complete system of 64 bases of 8x8 pixels, b) 2x bases that better trade
overcomplete system of 128 bases of 16x16 pixels, c) 192 bases of off representation
1000
800
16x16 pixels.
quality for sparseness 600
0.4
of coefficients, where 400
0.35 200the latter is essential
0
for efficient coding. To0.3
achieve sparseness, we of thresholding.
0.25
imposed the probability
0.2 distribution of expansion coefficients to peak at zero
(see Fig.12). Thus, to get the best non-linear0.15
approximation, we can get the coefficients with higher
0.1
absolute values, which we interested in, and discard the
0.05 others (Fig.12). In this way, each image will be
represented by the smallest number of bases0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
corresponding to the coefficients with higher activity.
Fig.10 Spatial-frequency localization of bases functions in pictures of
buildings (Fig.6). Here it is learnt a complete system of 64 bases of 8x8 Then, to have a quantitative measure of the
pixels. Note that bases are distributed along preferential directions in approximation-reconstruction quality, I compared the
6

Peak-Signal to Noise Ratio (PSNR) among the
reconstructed images from the learned codebooks
(Fig.13, 14).
a) Original b) Preprocessed by whitening/lowpass c) Approximation using 40 bases
d) Approximation using 10 bases e) Approximation using 5 bases
PSNR 64 bases 40 bases 30 bases 20 bases 10 bases 5 bases 2 bases
Complete 37.06 dB 36.62 dB 35.64 dB 33.49 dB 29.84 dB 27.24 dB 25.00 dB
Fig.13 Approximation of the pictures of buildings by using a decreasing number of basis functions of the complete codebook. The bases are selected by
taking the correspondent number of higher value coefficients. Note that e) reveals the 8x8 blocks of basis functions.
7

a) Original b) Preprocessed image
c) Approximation by 5 bases of the complete codebook d) Approximation by 5 bases of the overcomplete codebook
PSNR 128 bases 64 bases 40 bases 30 bases 20 bases 10 bases 5 bases 2 bases
Complete - 31.92 dB 31.43 dB 30.42 dB 28.51 dB 25.60 dB 23.62 dB 22.00 dB
Overcomplete 32.14 dB 31.75 dB 30.61 dB 29.37 dB 27.45 dB 24.77 dB 23.00 dB 21.62 dB
Fig.14 PSNR values computed for different approximations of the original image a). The values are displayed in the case of a complete and overcomplete
codebook. Note that d) reveals the 8x8 blocks of basis functions.
8

7. IMAGE DENOISING two global objectives are placed on a linear coding of
To demonstrate the ability of the adapted bases to
capture typical structures in the data, I applied the
algorithm to the problem of noise removal in images.
This task is well suited to the algorithm because
Gaussian additive noise is incorporated into the
specification of the image model. A set of bases that
characterizes the probability distribution of the data well
should improve noise removal properties, because they
are better at inferring the most probable image in the
face of uncertainty. Indeed, there results a good
improvement in PSNR value between the denoised and
noisy image (see Fig. 15 and 16).
Fig.15 Noisy image. PSNR=28.56 dB
Fig.16 Denoised images. PSNR=30.06 dB
8. CONCLUSIONS
Results demonstrate that localized, oriented and
bandpass receptive fields (as the ones found in the
simple cells of primary visual cortex) emerge only when
natural images:
• that information be preserved
• and that the representation be sparse
That means, an attempt of understanding the response of
primary visual neurons has been to consider their
relationship to the statistical structure of natural images
in terms of efficient coding.
By using Bayes’ rule and imposing some constraints of
the probability distributions of data, Olshausen et al. [1]
derived a learning algorithm that provides a set of basis
functions minimizing a certain cost function.
Applying the learning algorithm I verified that the
learned bases behave as feature detectors able to capture
the intrinsic structure of natural images (as seen in Van
Gogh paintings and pictures of buildings).
Moreover, increasing the degree of completeness results
in a higher density tiling of frequency space.
Because of sparseness and statistical independence
among coefficients it was possible to achieve efficient
and good quality representations of natural images.
Actually, decreasing the number of coefficients up to 10
elements of higher value, still gives a good quality of
reconstruction.
Besides, I applied the algorithm to the problem of noise
removal in images. Because Gaussian noise is already
incorporated into the specification of the image model,
the reconstruction stage infers the most probable image
in the face of uncertainty, carrying to good results.
Now, the last question, in neurobiological terms,
would be: “How close are we to understanding the
human visual cortex?”
What we learnt is that:
• Sparseness of coefficients resemble the sparse
activity of neuronal receptive fields.
• Learned bases from natural scenes reveal the
intrinsic structure of the training pictures: they
behave as feature detectors (edges, corners) like
V1 neurons.
But:
• The learned bases show higher density in tiling
the frequency space only at mid-high
frequencies, while the majority of recorded
receptive fields appear to reside in the mid to
low frequency range.
• Receptive field reveal bandwidths of 1 - 1.5
octaves, while learned bases have bandwidths
of 1.7 – 1.8 octaves.
• Finally, neurons are not always statistically
independent of their neighbours, as it is instead
assumed in our probabilistic model.
9

Moreover, there still remain several challenges that have
to be won by computational algorithms, that are:
• accounting for non-linearity, as shown by
neurons at later stages of visual system.
• accounting for forms of statistical dependence.
5. REFERENCES
[1] B. A. Olshausen and D. J. Field, Emergence of
simple-cell receptive field properties by learning a
sparse code for natural images, in Nature, nr. 381,
pp. 607-609, 1996.
[2] M. S. Lewicki and B. A. Olshausen Probabilistic
framework for the adaptation and comparison of
image codes, Journal of the Optical Society of
America, 16(7), pp. 1587-1601, 1999.
[3] R. M. Gray, Vector quantization, IEEE ASSP
Magazine, pp. 4-29, April 1984.
[4] B. A. Olshausen and D. J. Field, How close are we
to understanding V1?, submitted to Neural
Computation, 2005.
[5] B. A. Olshausen, and M. S. Lewicki, Sparse Codes
and Spikes, Probabilistic Models of the Brain:
Perception and Neural Function. R. P. N. Rao, ,
Eds. MIT Press. pp. 257-272, 2002.
10

Sparse codes for natural images

More Related Content

What's hot (16)

Similar to Sparse codes for natural images (20)

Recently uploaded (20)

Sparse codes for natural images