A Neuro-inspired Adaptive Motion Detector

doi:10.4236/opj.2013.32B024

Paper Menu >>

Journal Menu >>

Optics and Photonics Journal, 2013, 3, 94-98

doi:10.4236/opj.2013.32B024 Published Online June 2013 (http://www.scirp.org/journal/opj)

A Neuro-inspired Adaptive Motion Detector

Xiaopin Zhong1, Lin Ma2

1College of Mechatronics and Control Engineering, Shenzhen University, Shenzhen, China

2Shaanxi international trust co.,ltd. Xian’an, China

Email: [email protected], [email protected]

Received 2013

ABSTRACT

In this paper, a novel motion detector is proposed to perceive the weak changes in a image sequence. This is inspired by

the mechanism of fixational eye movement and dynamics of vertebrate’s cortex. We realized respectively an artificial

model of visual attention selection, called dual-probe adaptive model (DPAM), and an active tremor operation (ATO)

approach. It is foun d that between them there exists a resonance phenomenon. The phenomenon is enhanced when the

ATO and the DPAM are in-phase and is suppressed when they are anti-phase. Based on this, we construct a novel mo-

tion detector combined by the ATO and the DPAM to resonate with the motion direction. This allows capturing moving

edges even in the image sequences with lighting change and noisy background. Simulation and Experimental results

demonstrate the effectiveness.

Keywords: Neuro-inspired; Motion Detector; Dual-pro be Adaptive Model; Active Tremor Operatio n

1. Introduction

Motion detection is an important basic process in many

video analysis tasks [1], such as object detection, behav-

ior recognition and video en coding. There exist a number

of typical approaches for motion detection. However,

they consider in an image sequence all of the areas in-

stead of the areas with motion.

The most used two well-known methods are the tem-

poral difference [2] and the background modeling [3].

They can work well only when the background is ap-

proximately stationary and the foreground is relatively

moving, i.e. they are sensitive to noise and variations in

illumination.

Marr-Ullman model [4] is an early motion detector

which realized a highly sensitiv e 1D directional-selective

detection by using the temporal derivative of zero-crossing

fragment of measurement. This detector passively re-

sponds to any significantly-moving object.

The optic ﬂow [5] is also a successful visual motion

detection technique. There have recently emerged some

specific computing techniques for highly accurate optic

flows, but then they are still computationally inefficient.

In fact, human and primate’s visual system can not

only localize accurately moving objects, perceive their

moving direction and velocity. This offers the cognitive

ability to make use of the limited computing resource.

Bouzerdoum and Pinter [6,7] proposed a directional se-

lective multiplicative in hibitory motion de tector (MIMD)

under steady lighting condition. Based on MIMD, Wang

and Zheng [8] further developed a multiplicative inhibi-

tory velocity detector (MIVD) by replacing the low-pass

filter with a band-pass filter to determine the detectors

temporal feature. This replacement allows obtaining the

selectivi t y of motio n v ec t or for a mot i o n d etector.

Inspiring by the Sterlings model of the retinal nerve

circuits [9], a dynamic differential equation groups has

been set up by Ma et al. [10], which could be used to

capture the weak changing signals ex actly in noisy back-

ground with few parallel computing steps.

In the authors’ previous research [11], a group of dy-

namic differential equations has been set up to capture

exactly the weak changing signals in noisy backgrounds.

It is so-called the Dual-Probe Adaptive Model (DPAM).

Inspired by the mechanism of fixational eye move-

ments of human vision [13,14], we simulated an active

tremor operation (ATO). It is found that there is a reso-

nance phenomenon between the ATO and the DPAM.

This allows perceiving the motion direction. Based on

this finding, we propose th en an adaptiv e motion de tector.

This detector can perceive edges with specific moving

directions and adapt to changes of background and lighting.

The paper is structured as follows. In Section II, the

detailed adaptive motion detector is introduced and dis-

cussed. In Section III simulation and experimental results

are shown and analyzed and we conclude in Section IV.

2. Adaptive Motion Detector

2.1. Dual-Probe Adaptive Model (DPAM)

In fact, the so-call DPAM is a family of dynamic spatial

X. P. ZHONG ET AL. 95

temporal filters, and the parameter settings d etermine the

actual performance of a corresponding detector in re-

sponse to the input video. We denote first the model by

the equations as follows.



()( 1)(1)()

()=,( 1)

issishi s

ihs hihhi

kaskahk buk

hk akahk

 









sK

 (1)

()( 1),(( 1)())

iAiAi

kk Akhk



 AK (2)



()() ()

() ()

iii

bkskhk

bk bk

bk k

















(3)

()()( ()())

()() (()())

ii ii

GkbkAkbk

GkbkAkb k













D

(4)

where , denotes the component

of the input intensity vector .

()

uk 1, ,ith

()()k

and

stand for the component of the state variable feed-

forward vector and feedback vector respec-

tively. is the unit vector of coefficient of the

lateral inhibition effect of to

()



,hi K

in the surrounding

of pathway. This is to simulate a local grade po-

tential of nerve cells. denotes the gain coefficient

of photoreceptor. and

()

Gk 0

b

a represent the transfer co-

efficient of ()

k and th e feedback co effici ent of (1)hk



to (

khs



. and are the coefficient of re-

newal equation and confined to the convex combination

0hh

a0

The formula 2 is an adaptive threshold equation, where

()

k is the component of vector of , denotes

the adaptive threshold;

i()kA





,Ai ,Ai

K is the coefficient of re-

newal equation; and () is the unit vectors

of coefficient of the lateral inhibition effect of

K0()

k to

()

k in the surrounding of pathway, which is

similar to and wider than it.

()

,hi

For convenience, we define in formula 3 four interme-

diate variables , ,

()

bk ()

and ()

build respectively the corresponding vectors ,

, and b. They are also known as

bipolar varia bles named after bi p ola r cell of ret i na.

()

Hkb

()

Dkb(

Hkb) ()

The formula 4 is exactly the dual push-pull probes

output equation of the model, where and

are the output dual probes of the model. The readers are

referred to [11] for the analysis and discussion of the

model parameters.

()k



G()k



2.2. Active Tremor Operation Based Detector

In the neurobiological research, three types of uncon-

scious eye movements are regarded during gazing a tar-

get. They are high frequency tremor, microsaccades and

slow drifts [12,16]. Further research illustrates that the

role of microsaccades and drifts is to latch down objects

and to compensate the noisy control of muscle [12]. Only

the tremor is believed to be related to visual perception,

i.e. visual fading on retina is inhibited by tremors [15].

Inspired by this point, we introduce an active tremor

operation (ATO) into the DPAM model [10,11]. Ac-

cording to the analysis in reference [11 ], when the update

coefficient is close to zero, the DPAM is reduced to a

typical image change detector. It is called DPAM-m de-

tector because the outpu t is like an M type cell of human

retina [18]. With this additional ATO, the selective-in-

hibitory visual fading is then realized in a DPAM-m de-

tector. We call this the ATO-DPAM-m detector in the

followings. See Figure 1 for the processing structure.

For convenience, we redefine the ATO and introduce

the basic process.

Definition 1: Active Tremor Operation (ATO) is a pe-

riodic translational operation of global image on digital

video flows. This operation repeats a same translation of

all pixels of a frame with an identical direction, ampli-

tude and frequency.

As shown in Figure 2, for video flow constructed

by image sequence ()uk

()

k, suppose that the translation

amplitud e and t h e fr equency o f A TO are and 0.25

round/s respectively, i.e. (dk)

()

() (),

ppdk

uk uk





 (5)

where denotes the coordinates of a pixel in image

and (dk)



is a function of time k.

In the case of ATO-DPAM-m detector, the push-pull

outputs denoted by and mean the On-

type response and the Off-type response respectively.

()k





G()k





We found that there exists a resonan ce in the response

of an ATO-DPAM-m detector. This is because in the

DPAM-m detector, a visual fading occurs for the part

without change in the image sequence and an enhance-

ment response occurs for the part with change. In other

words, ATO enhances DPAM if they are in-phase and

ATO DPAM-m

()k





()k





()uk



()uk

Figure 1. Structure of ATO-DPAM-m detector.

Figure 2. An example of ATO process.

X. P. ZHONG ET AL.

ATO inhibits DPAM if they are anti-phase. Therefore,

the output of ATO-DPAM-m detectors has the selective

ability and is highly related to the directio n and the phase

of ATO. This provides a solid foundation of motion de-

tectors with directional selectivity.

To differentiate the directions, we use subscript

for the responds, i.e. and ,

where P means the positive operational direction and N

for the negative one. As shown in Figure 3 for example,

it is shown in a synthesized image sequence



()(,PN)()k













()

pkv

uk I







that there occurs in-phase enhancement or anti-phase

inhibition according to the relationship between d



and velocity v, and the ATO phases.

2.3. Adaptive Motion Detector (AMD)

We consider further a pair of ATO-DPAM-m detectors

with inverse directions, i.e. a positive d irectiona l detector

called ATOP-DPAM- m and a negative one called

ATON-DPAM-m. See in Figure 4 for the conﬁguration.

The parameters of both detectors are all identical except

that the difference of their phase is



and particularly

is set to 0.01. We then obtain a new ATO-based

adaptive motion detector.

(a) (b) (c) (d)

(e) (f) (g) (h)

Figure 3. The environmental adaptive threshold and the

response of left tremor of a horizontal ATO-DPAM-m de-

tector with different phases and amplitudes for video flow

pkv

uk I() 



 formed by a left moving image I

According to the property of in-phase enhancement and

anti-phase inhibition, we compare the different combina-

tions of output response and , yielding a

new set of outputs, i.e. ()k







G()k







PPNN

NNPP

() () () ()

()() () ()

kkkk

k kkk























 



GGGG

(6)

where and denote the positive On-type

output and Off-type output respectively. They selectively

respond to On-type and Off-type moving edges whose

moving directions are not i dent i cal to the posi ti ve di recti on

of the ATO. and represent the negative

On-type and Off-type output respectively. They selectively

respond to On-type and Off-type moving edges whose

moving directions are not identical to the negative direc-

tion of the ATO. As shown in Figure 5, the different

outputs of the proposed AMD can selectively respond to

the edges with specific motion characteristics.

P()k



P()k



N(k

N()k



In summary, the proposed AMD has realized a com-

binational selective detector to capture moving edges

based on the ATO resonan ce enhancement and inhibition

of edges in different moving directions. The method is

robust to backgrounds with change or switch due to the

environmental adaptive thresho ld.

3. Experimental Results and Analysis

In this section, we experiment using the KTH dataset [17]

and analyze the performance of the proposed adaptive

P()k





N()k





()k





()k





()k



()k



()k



()k



()uk



()uk





with ve-

locity v. (a) a stripe image example in the video flow; (b) the

environmental adaptive threshold; (c) and (d) demonstrate

the ATO-R response yielded by output for in-phase

edges and that yiel ded by output for antiphase edges

respectively when the ATO phase is

kG()







)kG(









; (e) and (f) show

the ATO-R response yielded by output for in-phase

edges and that yiel ded by output for antiphase edges

respectively when the ATO amplitude and the phase

d







kG()









. (g) and (h) show the visual fading on the output of

and

respectively when the ATO amplitude

and the ATO phase is

kG()





dv

kG()









Figure 4. The diagram of ATO-AMD’s configuration and

its output connection.

(a) (b) (c) (d)

Figure 5. The AMD response to a synthesized video input.

(a) is the synthesized video input; (b) shows the adaptive

threshold ; (c) and (d) shows respectively the positive

and negative direction responses, where the red regions

stand for On-type and the blue regions for Off-type.

Ak()

X. P. ZHONG ET AL. 97

motion detector (AMD). A number of behavior analysis

results have been reported based on the KTH dataset.

However their analysis involves only motion perception

and motion feature ex traction.

We construct a four-directional motion detector com-

bination which is formed by a pair of horizontal AMDs

and a pair of vertical AMDs to detect motion in KTH

videos with the movements of jogging, arm lift, and arm

down. See in Figures 6, 7 and 8 for detection results: (a)

a frame of input video; (b) the environmental adaptive

threshold A(k); (c) and (f) show the vertical edge re-

sponses, upward and downward respectively; (d) and (e)

illustrate the horizontal edge responses, towards the left

and the right respectively. What’s more, the red regions

denote On-type moving edges and the blue regions de-

note Off-t ype movi ng e d ges.

(a) (b) (c)

(d) (e) (f)

Figure 6. Detecting results of ATO-AMD on a video with a

body moving to the right.

(a) (b) (c)

(d) (e) (f)

Figure 7. Detecting results of ATO-AMD on a video with

arm lift movement.

(a) (b) (c)

(d) (e) (f)

Figure 8. Detecting results of ATO-AMD on a video with

arm dropped dow n.

From the results, we find this four-directional detector

combination can detect effectively both On-type and Of f-

type motion with different moving directions. For the

movements of jogging towards the right, the detector

barely responds to the movements towards the left, e.g.

Figure 6(d), while the detector for movements towards

the right br ings obv ious output respon se, e.g . Figure 6(e).

On the other hand, there occurs a little upward and down-

ward movement simultaneously with left-right movement.

Therefore the detectors for upward and downward mo-

tion can respond a bit to left-right movement. Likewise,

the other detectors generate similar results shown in Fig-

ures 6, 7 and 8.

4. Conclusions

Inspired by the neuroscience research, in this paper we

proposed a so-called dual-probe adaptive model (DPAM)

and an active tremor operation (ATO) approach to simu-

late the visual attention selection process and the fixa-

tional eye movement. We found that they enhance each

other if in-phase and inhibit each other if anti-phase.

Based on this finding, we has constructed a framework

by combining DPAMs and ATOs in two opposite direc-

tions. This framework can agilely capture the edges with

different moving directions even in the image sequences

with lighting change and noisy background. Therefore,

the proposed method provide an important basis for the

further study on motion analysis system.

5. Acknowledgements

This work was supported by the grants from Natural

Science Foundation of Shenzhen University (grant no.

201206). The authors would also like to thank the anony-

mous reviewers.

X. P. ZHONG ET AL.

REFERENCES

[1] C. W. G. Clifford and M. R. Ibbotson, “Fundamental

mechanisms of visual motion detection: models, cells and

functions,” Progress in Neurobiology, Vol. 68, No. 6,

2002, pp. 409-437.

doi:10.1016/S0301-0082(02)00154-5

[2] A. J. Lipton, H. Fujiyoshi and R. S. Patil, “Moving Target

Classification and Tracking from Real-time Video,” in

IEEE Workshop on Applications of Computer Vision,

1998, pp. 8-14.

[3] Y. Ren, C.-S. Chua and Y.-K. Ho, “Motion detection with

nonstationary background,” Machine Vision and Applica-

tions, Vol. 13, No. 5, 2003, pp. 332-343.

doi:10.1007/s00138-002-0091-0

[4] D. Marr and S. Ullman, “Directional Selectivity and Its

Use in Early Visual Processing,” Proceedings of the

Royal Society of London. Series B, Vol. 211, 1981, pp.

151-180. doi:10.1098/rspb.1981.0001

[5] B. K. P. Horn and B. G. Schunck, “Determining Optical

Flow,” Artificial Intelligence, Vol. 17, 1981, pp. 185-203.

doi:10.1016/0004-3702(81)90024-2

[6] A. Bouzerdoum and R. B. Pinter, “A Shunting Inhibitory

Motion Detector that Can Account for the Functional

Characteristics of Fly Motion-Sensitive Interneurons,” in

1990 IJCNN International Joint Conference on Neural

Networks, Vol.1, 1990, pp. 149-153.

doi:10.1109/IJCNN.1990 .137560

[7] A. Bouzerdoum, “The Elementary Movement Detection

Mechanism in Insect Vision,” Philosophical Transactions:

Biological Sciences, Vol. 339, No. 1290, 1993, pp.

375-384.doi:10.1098/r stb.1993.0037

[8] A. Wang, N. Zheng, L. Yuan et al., “Multiplicative In-

hibitory Velocity Detector (MIVD) and Multi-Velocity

Motion Detection Neural Network Model,” Proceedings

of International Conference on Multi-Sensor Fusion and

Integration for Intelligent Systems, 1996, pp. 476-483.

[9] P. Sterling, “Microcircuitry of the Cat Retina,” Annual

Review of Neuroscience, Vol. 6, No. 1, 1983, pp.

149C185.

[10] L. Ma, N.-N. Zheng and Z.-J. Yuan, “A Dynamic Model

for Early Vision Processing,” in IEEE International Con-

ference on Computer and Electrical Engineering, Phuket

Island, Thailand, 2008, pp. 96-100.

[11] L. Ma, N.-N. Zheng, Z.-J. Yuan et al., “A Novel Dual

Probes Adaptive Model for Image Change Detection,”

IEEE Signal Proceeding Letters, Vol. 17, No. 10, 2010,

pp. 863-866.

doi:10.1109/LSP.2010.2063023

[12] S. Martinez-Conde, S. L. Macknik, and D. H. Hubel,

“The Role of Fixational Eye Movements in Visual Per-

ception,” Nature Reviews Neuroscience, Vol. 5, No. 3,

2004, pp. 229-240. doi:10.1038/nrn1348

[13] I. Murakami, “Fixational Eye Movements and Motion

Perception,” in Twenty-eighth European Conference on

Visual Perception, A Coruna, Spain, 2005, pp. 35-35.

[14] S. Moshel, A. Z. Zivotofsky, L. Jin-Rong et al., “Persis-

tence and Phase Synchronisation Properties of Fixational

Eye Movements,” European Physical Journal-Special

Topics, Vol. 161, Jul, 2008, pp. 207-223.

doi:10.1140/epjst/e2008-00762-3

[15] M. Greschner, M. Bongard, P. Rujan et al., “Retinal Gan-

glion Cell Synchronization by Fixational Eye Movements

Improves Feature Estimation,” Nature Neuro sci enc e, Vol.

5, No. 4, 2002, pp. 341-347. doi:10.1038/nn821

[16] Engbert, Ralf. “Computational Modeling of Collicular

Integration of Perceptual Responses and Attention in Mi-

crosaccades,” The Journal of Neuroscience, Vol. 32, No.

23, 2012, pp. 8035-8039.

doi:10.1523/JNEUROSCI.08 08-12.2012

[17] Z. Gao, M. Chen, et al. “Comparing Evaluation Protocols

on the KTH Dataset,” Human Behavior Understanding

(2010): 88-100.

[18] W. Hui, Q. Zuo and X. Guan, “Main Retina Information

Processing Pathways Modeling,” International Journal of

Cognitive Informatics and Natural Intelligence, Vol. 5,

No. 3, 2011, pp. 30-46.

doi:10.4018/IJCINI.20110701 02