運動視差における視差量変化の検出(第23回大会 優秀発表賞抄録)

The Japanese Psychonomic Society

NII-Electronic Library Service

The JapanesePsychonomic Society

ThE

_.tapanese

_fournalof Ps),chonotnic Science

2005,Vol.24,No,1,113-l14

Summaryof

Awarded

PresentationIPI3

Change

detection

from

motion

_parallax

Kenchi

HosoKAwA*,

Satoko

OHTsuKA**,

and

Takao

SATo*

The bbeiversity

of

TOleyo*

and

Saitama

institute

of

7kechnolog),*'

We

investigated

thc temporal-charactcristicsof

depth

perception

from

motion parallax

by

presenting stimuli with an abrupt change

in

motion _parallax. In the

first

experiment the

participantscompared the depth _perceived from two stimuli. one ef which changed itsmotion

paraLlax abruptly. The results indicated thatthe same

depth

was _perceived when the averaged

parallax was

the

samc. In thesecond experiment, theparticipants reported the change

in

depth

perceived

from

a stimulus with a variety of motion _parallaxchanges. The reaction time of the

subjects was reversely _proportionaltotheamount of _parallaxchange. These results imply thatthe

depth

perception evoked by motion paral]ax

is

averaged

in

a periodof time mere

than

one second.

Key words: motion _para]]ax,motion indepth

Rogers and Graham

<1979)

have

reported that

reti-nal relative motion synchronizing toan observer's

head

movement can provide efficient cues for

per-ceiving depth,

The

studies on the structure from

rnotion without head mevement suggest that an

ac-cumulation of motion information isnecessary

for

the reconstruction of three-dimensional

(3-D)

sur-facesfrom retinal motion

(Ullman,

1984).

Inthis study we investigatedthctemporat

charac-teristicsof

depth

_perception from motion parallax

by

presenting

stimuli with changes inmotion _parailax

(Le,,

relative velocities to

head

movernent).

Experiment

1

The basicsettings were

based

on the way of

Ro-gcrs and Graham

(1979).

Inthe present study

how-ever, rnotions of

the

stimuli wcre intermittent,and

themotions stopped and restarted suddenly during a

presentation.

Method

Eachofthefourparticipant$reciprocally

moved theirhead sideways. The participants

ob-served two stirnuli: a constant and an intermittent

stimulus which were presented side

by

side. The

participantswere required tornatch the perceived

* _{Department of}

_Psychology,

_{Faculty of Letters,}

The University of

Tokyo,

7-3-1

Hongo,

ku,

Tokyo 113-O033

**

_Saitama

_Institute

_{of Technolog}', 1553-1}

_Fusaiji,

Okabe,

Saitama

369-0203

Copyright2005

depth of both stimuli by adjusting thevelocity of the

constant stimulus by a staircase rnethod. When the

percieved

depth

of

both

stimulus was equaL the

ve-locityof the constant stimulus was

defined

as the

pointof $ubjective equality

(PSE).

Both

of the stimu}i consisted of random dots.The

dots

in

the constant stimulus were synchrenized

with allof the head movement

but

the

dots

in

inter-mittent stimulus were only partially synchronized.

When

a participant's

head

was

located

in

the

"mov-ingpart" of the reciprocation, the

dots

were

synchro-nized but in the remaining

part

the

dots

did

not

move,

The

ratios of the `'moving

part"tothe tetal

length of the reciprocation, were from 1/6 to 5/6.

The

location

of the "moving part" was defined as either

the

"center"

or "periphery''.

In

the "center" condition, the "moving

part"was located at the

cen-ter of the rcciprocation and the non-moving parts

were at each periphery,

The

"periphery"

condition was opposite. The velocities of theintermittent stim-uli

in

the "moving

parts"were 40, 60, and 80 arc

seconds,

in

equivalent _parallax.

The head movement was measured

from

the

loca-tion of chin rest which was mounted on a rail20

centimetres long. The reciprocation Nv'as _performed at

least

three times,and cued

by

u tone. Each

one-way movement occurred

in

a period of

1

second

(Le.

reciprocated at O.5Hz). The velocity _gradient of the

dot

motion was sinusoidal.

The Japanese Psychonomic Society

NII-Electronic Library Service

TheJapanesePsychonomic Society

114 The

_Japanesc

_Journalof

Psych

(a)

s

`D velocity:4o"

"I.1'

L

,nt

F

g

i

g,,

...･

P

]'

t

a

'..,･,"

1 e "'･---T--Lr'r

lfs 21e 31e 4I6 srs alG 11ey6 4reSrS

Moving-part

ratio

Results A noticeable change in depth was not

reported by any of

the

_{participants,}

In

addition,

the

PSE

wcre clearly _proportional to the ratio of the

"rnoving

part",as shown

in

Figure

la.

In other

words, the _perceived

depths

were equal when the

averaged velocity between both stimuli was equaL

The

correlation coefficients of _the "moving

part"

ra-tioand thc

PSE

velocity wcre allabove

O.72,

for

all of

the individualdata sets

Cnot

the avcraged data}.

Experiment

2

Experiment 1suggested thatpercieued depth from

motion parallax isaveraged over a tirneperiod. To

examine thispossibilitythe amount of the velocity

changes were manipulated.

It

was proposed that

if

the motion _parallax was averaged temporally, the

amount of the velocity change would beina reversc

proportion to the time to exceed the threshold

to

noticc thc change indepth,

Methods

The

basic

settings were similar to those

in

expcriment

1,

but

only one stimulus was

pre-sented

in

thisexperiment.

The

ve]ocity of the

stimu-luswas suddenly changed after some cycles of

head

reciprocation. This velocity change was varied in11

steps

frem

-100%

to+100% of the velocity

before

the change, The times of thevelocity changes were

at 2,4,6,and 2.25,4.25,and 6.25cycles of

reciproca-tion.

The

first

three conditions

did

not

include

an

abrupt change because the velocity change occurred

at a point where

the

stimulus djd not have any

motion. The other conditions however, included an

abrupt change ofvelocity. The participantsreported a change of the depth which was perceived

by

press-ing

kcy, and

their

reaction time after the change

onomic Sclence Vol.24,No. 1

(b)

Ti.o

--･･1

4S F

.,..

84n

t' '･,

'gl: "'' 'x,,,,'s,.

:.s

',

tt

p L.' 020

Ei.,III

i-･

'･i.'li'{

US 716 3te 4V6 5fe / {'t' M.P/ta

i'za des -bk

-!ca.

'

.2os

.4'es'

i'os .e'os

,Edm

VeloGityeltange

occurred was recorded,

Results

The

reaction times of the subjects were

almost constant

for

all of the times of velocity

change and the reaction time was inreverse

propor-tionto theamount of velocity change, as shown in

Figure lb. This result agreed with the expectation

that the reaction _time would

be

reversely

propor-tionaltethe amount of velocity change. Moreovcr, a

differencebetween the abrupt and non-abrupt

cha-nges was not reflected

in

the

reaction time.

The

resu!ts indicatedthattheternporalcharacteristics of

thechange of motion _parallaxhad a weak cffect on

the _perccption of a

depth

change.

Discussion

The proportionality found inthe two experiments suggests that the fluctuatingvelocity signal

is

aver-aged over a presentation period,

The

lack

of

detec-tion of abrupt motion changes found in thesecond

experirnent also supports

this

_proposaL

The

integra-tion time should

be

one second or

longer,

because

there was one switch inthe motion velocity within

each _periodof head movement. However. the

pre-sent experiments were not able to

determin

which of

both,

motion and

depth

system,

lead

tothe response characteristics.

References

Rogers,

B.

&

Graham,

M. 1979

Motion

paral]ax as an

independent cue for depth.

_perception.

Ilerception,

8,

125-139.

Ullman,

S.

1984 The incremental recovery of

3-D

structure from rigid and nonrigid motion.