Journal of Signal and Information Processing, 2013, 4, 102-108
doi:10.4236/jsip.2013.43B018 Published Online August 2013 (http://www.scirp.org/journal/jsip)
Segmenting Salient Objects in 3D Point Clouds of Indoor
Scenes Using Geodesic Distances
Shashank Bhatia, Stephan K. Chalup
School of Electrical Engineering and Computer Science, The University of Newcastle, Callaghan 2308, NSW, Australia.
Received May, 2013.
ABSTRACT
Visual attention mechanisms allow humans to extract relevant and important information from raw input percepts.
Many applications in robotics and computer vision have modeled human visual attention mechanisms using a bottom-
up data centric approach. In co ntrast, recent studies in cognitive scien ce h ighligh t adv antages of a top-do wn ap pro ach to
the attention mechanisms, especially in applications involving goal-directed search. In this paper, we propose a top-
down approach for extracting salient objects/regions of space. The top-down methodology first isolates different objects
in an unorganized poin t cloud, and compares each object for uniqueness. A measure of saliency using the properties of
geodesic distance on the object’s surface is defined. Our method works on 3D point cloud data, and identifies salient
objects of high curvature and unique silhouette. These being the most unique features of a scene, are robust to clutter,
occlusions and view point changes. We provide the details of the proposed method and initial experimental results.
Keywords: Saliency Detection; 3D Image Analysis; Image Segmentation
1. Introduction
Traditionally, methods of landmark extraction for the
purpose of robot localization have been dependent on the
type of environment and nature of landmarks. Such
methods follow the standard procedure of sequentially
scanning the input percept and aim to match pre-defined
patterns for recognition of landmarks. The necessity of
pre-defining patterns associated with landmark locations,
limits the use of the rob ot to a specific environment. Fur-
thermore sequential processing of data necessitates high
computational power to be ported on a mobile plat- form.
This sequential processing of pix els or image windows is
in contrast to human visual mechanism. The latter incor-
porates an attention mechanism that helps humans to fo-
cus on the most relevant stimuli based on the task at hand
[1-2]. Incorporation of similar strategy in computational
vision systems, especially for application of robotics, can
have many advantages. Computational Attention (CA),
commonly known as “Saliency Detection” or “Interest
Point Detection”, aims to identify the regions of sensory
input that stand out from their neighbors and attract the
attention of the subject [3-4]. Due to the conv enience that
CA offers, it has been adapted in many applications re-
quiring either judicial selection of inputs [5], minimizing
computational cost [6], or to achieve invariance to clutter
[7].
Computational attention in the 2D image domain has
been investigated from past five decades. C. Koch and S.
Ullman [8] were the first to provide theoretical founda-
tions of visual attention mechanisms. The authors pro-
posed creation of different conspicuity maps each select-
ing locations in visual space, which d iffer from their sur-
roundings in terms of color and orientation. Further, a
Winner Take All (WTA) neural network was proposed to
combine different conspicuity maps and select the most
salient region. Most of the current visual attention sys-
tems are based on the implementation of the WTA net-
works, developed by L. Itti et al. [9]. In their implemen-
tation, the authors extended the existing theoretical con-
cepts by adding intensity as another feature for comput-
ing the conspicuity map. However, it should be noted that
most of the existing approaches use the intrinsic proper-
ties of an input image. These properties depend on factors
like the presence of ambient light, amount of re- flections,
visibility of colors and presence of occlusions and are
therefore unstable. This factor has motivated re- search-
ers to utilize 3D depth information (which is inde- pen-
dent of ambient light) in the process of determining sali-
ent regions. Furthermore, the availability of the low- cost
3D capturing devices in recent years, has motivated the
usage of 3D depth informatio n, especially for systems on
mobile robotic agents.
To id entif y salie nt reg ions in a scene, mechanisms that
evaluate intrinsic properties of raw data elements to spot

Segmenting Salient Objects in 3D Point Clouds of Indoor Scenes Using Geodesic Distances 103
the regions of potential interest are said to follow bot-
tom-up approaches [10]. Such methods explore the
neighborhood of each data point present in the input, and
assemble points into small clusters having salient char-
acteristics. In this process, the clusters of higher saliency
thus obtained, may comprise of arbitrary points that may
belong to multiple objects. On the other hand, methods
that instead of operating on individual data elements
evaluate a collection (with all data elements belonging to
same object) are known as top-down approaches. Here,
the evaluation is performed after isolation of the points
into different obj ects. Many cognitiv e science studies and
experimental evaluations described in [11] have shown
that bottom-up methods are well suited to applications
involving explorative tasks, but may not be suitable for
goal-directed searching. These studies encourage ideas of
taking an “object” as a unit for attention selection and
support suitability of top-down approaches for applica-
tions involving goal directed search. However, since most
of the existing methods of 3D saliency detection are
based on bottom-up approaches, their usage has re-
mained limited. Only the most simplistic methods like
[12] and [6] have been used in robotic applications.
In this paper, we propose a simple top-down approach
to extract salient regions from raw 3D point cloud data.
The top-down nature of our approach segments the scene
into physically disconnected regions and then compares
properties of each region for saliency. We define saliency
measures that capture variations in curvature and silhou-
ette (an outline of an object/scene consisting of feature-
less interior) of the corresponding regions, and compare
them with other objects present in the surroundings. We
report the initial experiments and results of by testing it
in an environment containing objects of different shapes
and degrees of curvature. Section 2 of the paper provides
a short review of related work, followed by the motiva-
tion behind this research. Details of the proposed ap-
proach are provided in section 3 and 4. Section 5 de-
scribes the measures and initial experiments conducted,
and concludes the paper.
2. Related Work
Available methods in majority either cannot handle large
size point clouds, or to achieve computational efficiency,
reduce the dimensionality of the point cloud. In [13], a
multi-scale filtering operator is derived by the convolu-
tion of a Gaussian kernel with the operating surface. The
operator has the property of being proportional to the
curvature of the local area at which it is applied. In effect,
it is directly applicable to 3D point clouds and captures
the variation in shape of the neighborhood of point of
application. However, the need of processing over multi-
ple scales renders its utility limited to very small point
clouds. To overcome this disadvantage, J. Stuckler and S.
Behnke [7] extended the interest operator to work on
depth images. Similar to the historical intensity-driven
visual attention algorithms, their approach builds a multi-
scale pyramid representation of the depth image to be
used with the operator. However, approximation of the
depth image limits the factual description of the interest
points limited to detection of blobs and corner-like fea-
tures. It is important to note that in goal-driven applica-
tions, salient regions are useful only if they can be rec-
ognized as important features. Common to most applica-
tions, extracted regions should be invariant to noise, scale,
and viewpoint transfor m. Following the same conven tion
of bottom-up mechanisms provided by [7] and [12] limits
the input space to depth images, and pro- vide a simplis-
tic method to extract boundary regions as areas of interest.
While this approach is computationally efficient, it does
not take account of the 3D shape of the objects in deter-
mining the saliency values.
Cole and Harrison were one of the first to incorporate
the 3D curvature information directly to identify regions
of interest for the application of robot Simultaneous Lo-
calization and Mapping (SLAM) [6]. The auth ors utilized
an information-theoretic entropy measure to identify the
regions of maximum random curvature. In contrast to [7],
authors of [12] used spherical regions to define the scale
space. The degree of saliency was based on the entropy
of normals in each spherical region and its variation over
multiple scales. Using a spherical shape for defining a
scale space leads to the selection of points with the high-
est variation in curvature as the most salient regions.
These points in general may correspond to more than one
object, and therefore do not provide any recognizable
information of the selected region.
Flint et al. [14] defined an interest point to be the one
having the largest principal curvature in all three Euclid-
ian axes. The magnitude of all three principal curvatures
was estimated by calculating the Hessian matrix con-
volved with a Gaussian kernel. Finally, the areas having
the highest determinant of the Hessian matrix were
deemed to be most salient regions. The experimental re-
sults revealed that the proposed method extracts all cor-
ner and edge points, representing the areas with the high-
est variations in curvature. Usin g spatial properties of the
cloud data, Akman and Jonker [4] proposed to include
the depth as a criterion for saliency. In their approach,
two different saliency maps were used in combination to
obtain the final saliency map. The first saliency map was
calculated using values of curvature. The second map
was taken as being inversely proportional to the depth of
a region. The farther a region, the lower its saliency
would be. In the application of saliency for object classi-
fication, Potapova and Zillich [15] proposed the extrac-
tion of the orientation of objects in point clouds relative

Segmenting Salient Objects in 3D Point Clouds of Indoor Scenes Using Geodesic Distances
104
to the surface on which they are located. This relative
orientation was used to create a saliency map, and was
combined with the traditional 2D saliency approach to
obtain a complete saliency map.
As evident from our review, most of the present re-
search on visual attention mechanism on 3D data has
been focused on bottom-up approaches. These ap-
proaches assume homogeneity of input data, and the de-
tection of saliency is mostly based on the intrinsic prop-
erties. There is no mechanism that suggests the grouping
of raw data into identifiable objects/artifacts present in
the input space. Some of these methods require multi-
scale processing, while others approximate 3D data to
depth images. For this research, a particular highlighted
drawback is the grouping of multiple objects into a re-
gion of interest. This potentially can degrade the effi-
ciency of goal directed search applications. Research in
cognitive science has shown the suitability of top-down
attention mechanisms in goal directed applications [11].
Further, top-down mechanisms are more computationally
efficient, as they do not necessarily process all the data
sequentially. In view of these findings, in this paper we
propose a top-down approach for salient region detection.
Our approach first clusters objects present in an input
percept, and then evaluates each separated object for the
degree of attention that it may acquire. With the applica-
tion of our approach, goal directed search applications
have the possibility of increasing computational per-
formance, and time efficiency. The maj or contribution of
our approach is its top-down nature of saliency estima-
tion, which considers an object as an elemental unit for
attention selection.
3. Planar Region Extraction
Indoor scenes constitute planes as a major part of their
point cloud input. The planar part of the 3D point cloud
does not contribute to any variations in curvature, and
therefore become distractions and increase computational
cost. Additionally, removal of planar regions is required
for isolation of salient objects from other artifacts present
in the scene. Ther efore, while seeking to identify regions
with higher curvature, we choose to remove these planar
regions using the RANSAC method as described in [16].
3.1. Local Surface Normals
The planar regions have low or zero curvature associated
with them. Leveraging on these regions can be identified
if the surface normal to each point is known. These LSNs
are estimated using the method described in [17]. The
surface normal to a query point can be estimated using
the eigenvalues and eigenvectors of the of the covariance
matrix comprising the k-nearest neighbors of the query
point (equation 1).
C1
k
i1
k

p
i
p

p
i
p

T

(1)
C


v
j


j


v
j
,j{0,1,2}
 

0

0
 
1
 
2
(2)
Here
p
is the 3D centroid of the nearest neighbors.

j
is the eigenvalue, and j
th

v
j
its corresponding
eigenvector. The eigenvector

v
0
corresponding to smal-
lest eigenvalue 
0
is the approximation of the normal
at the query point, and the ratio of the eigenvalues (equa-
tion 2) provides an estimate of variation in curvature at
the query point.
3.2. Iterative RANSAC
Plane extraction using Random Sample Consensus
(RANSAC) method described in [16], identifies best
fitting planar region in the input point cloud. As a result,
only one largest p lanar region is extracted fro m the input
point cloud, leaving the rest intact. In order to remove
subsequent planar regions, we adapted recursive use of
RANSAC. The input point cloud is processed multiple
times, separating one best planar surface at a time.
Recursive RANSAC works by feeding back the residual
cloud obtained in previous iteration. The extraction is
executed until 95% of the point cloud is processed. As a
result, a list of all extracted planar regions is obtained.
This list contains all possible planar reg ions present in the
input percept. In addition, the list may also contain parts
of planar regions embedded on objects present in the
environment. The resultant cloud may thus contain
occlusions and holes. These holes also lead to incorrect
object based clustering. To overcome this disadvantage,
we employ a statistical high pass filter, which removes
only significantly large planar region s present in the input
cloud. The high pass threshold value is set to



,
where

and

are the mean and standard deviation
of the total number of points contained in all the
candidate planes. Finally the candidate planes with
number of points above the threshold are removed from
the original cloud.
Figure 1 displays raw point cloud data of The
Newcastle Robotics Lab, with different objects placed on
the ground. The objects comprise of a toy bear, a
humanoid robot, a carton box, and a basketball. These
objects have different properties of curvature and were
chosen to demonstrate the behaviour of the proposed
method with different types of objects. The extracted
planar region points are marked different shades of grey.
Objects that remain after plane extraction are marked in
black. The data was collected using the Kinect RGB-D
camera [18]. More details of the experimental setup are

Segmenting Salient Objects in 3D Point Clouds of Indoor Scenes Using Geodesic Distances 105
Figure 1. Complete point cloud (top), corresponding
detected planar regions marked in different shades of grey
(bottom), residual objects after plane extraction marked in
red (bottom).
provided in Section 4.
4. Salient Region Extraction
In this section we provide details of how saliency is
measured, and relevant areas with high saliency extracted.
There are two major phases involved. First, the Euclid-
ean clustering, and second the ranking of the extracted
clusters for saliency. Euclidean clustering divides the
point cloud into smaller objects/regions to be considered
as elements for saliency computation. Finally, the sepa-
rated objects are evaluated for saliency. The following
subsections pr ov ide detail s of eac h ph ase.
4.1. Euclidean Clustering
The residual cloud obtained after removal of planar re-
gions contains unlabeled points, some belonging to iso-
lated objects and others being noisy residual of planar
extraction. In order to perform a top-down object-level
comparison of uniqueness, these objects have to be iden-
tified as separate entities. In other words, the point cloud
needs to be divided in multiple parts, each containing one
isolated object. This is achieved by comparing the
Euclidean distance between neighboring points. Cluster-
ing is performed using a k-nearest neighbor search [19].
Nearest neighbor search starts by selecting a random
point from the residual cloud, and computes the Euclid-
ean distance of the point from its nearest neighbors.
Points that fall under a threshold value are labeled to be
part of the object. The search stops when no nearest
neighbor falls under the threshold distance. At this stage
the points found so far are labeled into one group, an d the
search starts again by removing the object from the cloud,
and randomly selecting another un-labeled point. The
clustering stops when all poin ts are labeled. For the ne ar-
est neighbor search, we make use of a binary KdTree
implementation as in [20]. This approach divides the re-
sidual cloud into a binary tree structure, enabling easy
and fast nearest neighbor searches. The result of the clus-
tering can be seen in Figure 2. The distance threshold
used here is 0.2 m and as evident, the method identifies
four different objects present in the scene. In the figure,
each object is presented by different color of the points it
contains.
4.2. Saliency Ranking
Multiple point clouds obtained as a result of clustering
are evaluated for uniqueness in two aspects: 1) Variance
of curvature on the object’s surface, 2) Shape of silhou-
ette formed from the object. These properties are cap-
tured together in one measure, defined using the differ-
ence between geodesic and Euclidean distances between
all sets of points in the object point cloud. The geodesic
distance between any two points of a cloud is the length
of the shortest curve on the surface, connecting these
points. Due to the embedding of the curve on the surface,
in Euclidean space the geodesic distance between two
points having non-zero curvature is always greater than
or equal to their Euclidean distance. Additionally, sur-
faces with high amount of variation in the curvature of
their boundary/silhouette may also have their geodesic
Figure 2. Clusters obtained after removal of planar regions,
and performing Euclidean Cluster ing on the residual cloud.

Segmenting Salient Objects in 3D Point Clouds of Indoor Scenes Using Geodesic Distances
106
distance between any two points on their boundary
granter than corresponding Euclidean distance. Figure 3
illustrates these facts by means of two simple examples.
First, a half sphere is presented with the geodesic dis-
tance (green curve) and Euclidean distance between two
points on the surface of the half sphere. The grater value
of geodesic distance is evident from the figure. Secondly,
a curved silhouette of an arbitrary object is presented.
Again, the geodesic distance between two points on the
boundary is grater than their corresponding Euclidean
distance.
Figure 3 conveys that for any object with curved sil-
houette and higher curvature on the surface, the values of
sum of geodesic distance between all points would be
high. More precisely, the difference between geodesic
and Euclidean distances between all points of the surface
can be used to identify objects with higher curvature and
complex shape of silhouette. Exploiting the properties of
geodesic distance, we formulate a saliency measure that
captures variation in the curvature as well as the silhou-
ette of the object under study. Consider an object point
cloud
k
comprising of a total
C n
k
points, for each
point in we define the following:
p
i
G
ij
k
p
i
,p
j
, ij,i,jC
k
E
ij
k
 p
i
,p
j
, ij,i,jC
k
P
k
G
ij
k
E
ij
k

2
ij

VkPkPk

2
nk1
ij

S
k
V
k
V
k
k

(3)
Gij
k
p
denotes the geodesic distances from point
i
to
j
in the
p
k
th
object point cloud.
ij
denotes corre-
sponding Euclidean distance between the same points.
E
k
Figure 3. Comparison of geodesic distance (dashed line) and
Euclidean distance (solid line) in presence of curvature (left),
on a curved silhouette (right). Note that in both cases, the
geodesic distances are higher than Euclidean distances.
Since the geodesic distances are always grater or equal to
the Euclidean distances for surfaces with higher curva-
tures, any object having higher variability in the differ-
ence between
ij
and
ij
will stand out from its sur-
roundings. This variance is captured by
G
k
E
k
V
k
and finally,
the geodesic saliency
S
k
is defined as the normalized
value of
V
k
(normalized over all clusters). The graph in
Figure 4 presents the quantity
k
normalized by the
size of the cluster. This normalization factor also ensures
that the value of saliency does not depend on the size and
number of points contained in the point cloud of the ob-
ject. The graph illustrates the variation in the values of
the proposed saliency measure against changes in dis-
tance. There are four different objects present in the input
cloud namely a toy bear, Nao humanoid robot, a basket-
ball, and a flat box object. It can be noticed that as the
distance increases, the value of saliency reduces. This
happens due to the addition of noise. Moving away from
objects, the curvature is less observable. This particular
noise addition is sensor dependent, and current experi-
ments report the results obtained using Microsoft Ki-
nect-RGBD Sensor.
S
Geodesic distances between all pairs of points in each
object point cloud are computed using Floyd-Warshalls
algorithm [21]. The point cloud of each object is con-
verted into a fully connected graph, with each point
treated as a vertex. The algorithm compares distance-
minimizing paths between two vertices in the given con-
nected graph and incrementally improves the estimate of
geodesic distance between two vertices iteratively. A
more detailed explanation of the algorithm can be found
in [21].
Figure 4. Values of the quantity k
ik
j
i
j
GE, normalized over
the size of cloud. Note that as the distance increases, the
saliency values of ball and box converge. This is due to
increasing noise in the calculation of curvature with
increase in distance.

Segmenting Salient Objects in 3D Point Clouds of Indoor Scenes Using Geodesic Distances 107
5. Experimental Evaluation
In order to evaluate our approach we captured point
clouds from different view-angles in the laboratory. Four
objects of varying shapes and curvature were placed on
the floor. 3D point clouds were recorded with 3D
Time-Of-Flight (TOF), Microsoft Kinect Sensor [18] that
was mounted on a tr ipod. The viewpo int was varied fr om
-25 to 25 degrees, in steps of 5 degree. The distance was
varied betw een 1.5 m to 2 m, in steps of 0.1 m. The input
clouds were processed on a Dell workstation equipped
with Intel Xeon® 3.40 GHz processor and 16 GB of
RAM.
5.1. Performance Measures
We utilized the existing measures of repeatability and
overlap, as described in [7] to evaluate our approach.
These measures are known to evaluate qualitative and
quantitative performance of saliency extraction method.
The repeatability of detection of salient regions is de-
fined as the frequency with which the same cluster is
ranked with a si milar level of salien cy. This measures the
stability of the approach with variations in distance and
viewpoint changes. Overlap rate on the other hand, is
calculated by comparing the location of salient regions
found with variations in distance and viewpoint. If the
salient regions belong to same location, overlap is incre-
mented and vice versa. Location of the salient region was
calculated as the centroid of the point cloud clus ter. Sin ce
the position of the sensor was changed, these cen- troids
were transformed from the local frame of reference into
the global frame of reference of the environment. Heat-
maps are used to gr aphically represent this perfor- mance
measure. The heatmap used consists of cells, with each
cell representing the value of repeatability/overlap
(scaled between 0 and 1). The rows of the heatmap rep-
resent the distance of evaluation, and the columns repre-
sent the angle in degrees. All together, presented heat-
maps are a visual representation of robustness of pro-
posed method.
5.2. Discussion
Figure 5 shows the resultant heat map reflecting the
overlap and repeatability of salient object detection. The
values of repeatability and overlap are scaled (between 0
and 1) to provide an accurate account of the performance.
The figure demonstrates the repeatability of the toy bear
is higher than that of the rob ot. This is due to complexity
of silhouette of the toy bear. Reason behind higher sali-
ency values of the toy bear are depicted in Figure 3. Ad-
ditionally Figure 4 conforms to results in Figure 5,
where the toy Bear has attained highest values of sali-
ency. The humanoid robot, having highly curved surface
Figure 5. Bear (top), Ball (row 2), Robot (row 3), and Box
(bottom) performance (left: overlap and right: repeatability)
wrt. view angle and distance. This performance measure
was adapted from [7]. We can see that the proposed method
is robust to viewpoint and scale changes.
follows next in saliency ranking. Despite being smaller as
compared to the flat box, it has higher values of sali-
ency. Finally, the robustness is demonstrated by the high
values of repeatability, which in most cases ranges be-
tween 0.7 to 1. It should be noted that lower values of
overlap and repeatability in case of bear and ball are due
to the restricted exposure of the objects with change in
angle of the sensor. Moving beyond 10 degree, the bear
was not completely visible in the Field of View (FOV) of
the sensor. Similarly the ball, that was not visible in the
FOV while changing the angle of the sensor below 0 de-
gree. Apart from the missing values, all other observa-
tions presented high values of the two measures, which
are most desirable characteristics of salient region ex-
traction methods [7].
6. Conclusions
In this paper, we present a top-down approach for ex-

Segmenting Salient Objects in 3D Point Clouds of Indoor Scenes Using Geodesic Distances
108
tracting salient regions/objects from indoor environments.
Our method segregates significant planar regions, and
extracts isolated objects present in the residual point
cloud. Each object is then ranked for saliency based on
higher curvature complexity of the silhouette. These
properties are captured together using the proposed geo-
desic distance measure (Figure 4). The paper has re-
ported initial experiments and demonstrates capacity of
the method in identifying objects/regions of higher cur-
vature. Further, testing with variations in viewpoint and
distance, reveal stability of propo sed saliency criterion.
These initial experiments demonstrate the advantages
of adapting top-down clustering for the purpose of sali-
ency ranking. A possible limitation of the method could
be identified as lack of using RGB informatio n to suppo rt
the selection of salient regions, and future developments
of this research aim to include variations in color for sa-
liency computation.
REFERENCES
[1] T. N. Vikram, M. Tscherepanow and B. Wrede, “A Sali-
ency Map Based on Sampling an Image Into Random
Rectangular Regions of Interest,” Pattern Recognition,
Vol. 45, No. 9, Sep. 2013, pp. 3114-3124.
doi:10.1016/j.patcog.2012.02.009
[2] S. Frintrop, E. Rome and H. I. Christensen, “Computa-
tional Visual Attention Systems and Their Cognitive
Foundations,” ACM Transactions on Applied Perception,
vol. 7, no. 1, Jan. 2010, pp. 1-39.
doi:10.1145/1658349.1658355
[3] N. Riche, M. Mancas, B. Gosselin and T. Dutoit, “3D
Saliency for Abnormal Motion Selection: The Role of the
Depth Map,” in J. Crowley, B. Draper, and M. Thonnat,
Eds. Computer Vision Systems, Springer Berlin/Heidel-
berg, 2011, pp. 143-152.
doi:10.1007/978-3-642-23968-7_15
[4] O. Akman and P. Jonker, “Computing Saliency Map from
Spatial Information in Point Cloud Data,” Advanced
Concepts for Intelligent Vision Systems, Vol. 6474, 2010,
pp. 290-299. doi:10.1007/978-3-642-17688-3_28
[5] D. Simon, “Fast and Accurate Shape-Based Registration,”
PhD thesis, Robotics Institute, Carnegie Mellon Univer-
sity, Pittsburg, PA, 1996.
[6] D. Cole and A. Harrison, “Using Naturally Salient Re-
gions for SLAM with 3D Laser Data,” In Proceedings of
International Conference on Robotics and Automation,
Workshop on SLAM, 2005
[7] J. Stückler and S. Behnke, “Interest Point Detection in
Depth Images Through Scale-Space Surface Analysis,” In
2011 IEEE International Conference on Robotics and
Automation (ICRA), 2011, pp. 3568-3574.
doi:10.1109/ICRA.2011.5980474
[8] C. Koch and S. Ullman, “Shifts in Selective Visual Atten-
tion: Towards the Underlying Neural Circuitry,” Human
Neurobiology, Springer-Verlag, Vol. 4, No. 4, 1985, pp.
219-227.
[9] L. Itti, C. Koch and E. Niebur, “A Model of Saliency
Based Visual Attention for Rapid Scene Analysis,” IEEE
Transactions on Pattern Analysis and Machine Intelli-
gence, Vol. 20, No. 11, 1998, pp. 1254-1259.
doi:10.1109/34.730558
[10] C. E. Connor, H. E. Egeth and S. Yantis, “Visual Atten-
tion: Bottom-Up Versus Top-Down,” Current Biology,
Vol. 14, No. 19, 2004, pp. R850-R852.
doi:10.1016/j.cub.2004.09.041
[11] M. Begum and F. Karray, “Visual Attention for Robotic
Cognition: A Survey,” IEEE Transactions on Autonomous
Mental Development, Vol. 3, No. 1, 2011, pp. 92-105.
doi:10.1109/TAMD.2010.2096505
[12] B. Steder and R. Rusu, “NARF: 3D Range Image Fea-
tures for Object Recognition”, In Workshop on Defining
and Solving Realistic Perception Problems in Personal
Robotics at the IEEE/RSJ International Conference on
Intelligent Robots and Systems (IROS), 2010.
[13] R. Unnikrishnan and M. Hebert, “Multi-Scale Interest
Regions From Unorganized Point Clouds”, In IEEE
Computer Society Conference, Computer Vision and Pat-
tern Recognition Workshops, 2008, pp. 1-8.
[14] A. Flint, A. Dick and A. v. d. Hengel, “Thrift: Local 3D
Structure Recognition,” In Society on Digital Image
Computing Techniques and Applications, 9th Biennial
Conference of the Australian Pattern Recognition, 2007,
pp. 182-188.
[15] E. Potapova, M. Zillich and M. Vincze, “Calculation of
Attention Points Using 3D Cues,” 35th Annual Workshop
of the Austrian Association for Pattern Recognition
(OAGM/AAPR), May 2011.
[16] R. B. Rusu, “Semantic 3d Object Maps for Everyday Ma-
nipulation in Human Living Environments,” PhD Thesis,
Computer Science Department, The University of Roch-
ester, Oct. 2009.
[17] D. Holz, S. Holzer and R. Rusu, “Real-Time Plane Seg-
mentation Using RGB-D Cameras,” In Proceedings of the
15th RoboCup International Symposium, Isntanbul, Tur-
key, 2011, pp. 306-317.
[18] Microsoft kinect sensor,
http://www.xbox.com/en-au/kinect?xr=shellnav," 2013.
[19] E. Fix and J. L. Hodges, “Discriminatory analysis. Non-
parametric discrimination: Consistency properties,” In-
ternational Statistical Review/Revue Internationale de
Statistique, 1989, pp. 238-247.
[20] R. Rusu and S. Cousins, “3D Is Here: Point Cloud Library
(PCL),” In IEEE International Conference on Robotics
and Automation (ICRA), 2011, pp. 1-4.
[21] R. W. Floyd, “Algorithm 97: Shortest path”, Communica-
tions of the ACM, Vol. 5, No. 6, Jun. 1962, p. 2.
doi:10.1145/367766.368168