A great variety of neuroscience studies and results concerning motion estimation and perception in primates can be found in the literature. In this work, we focus on two aspects of motion estimation, namely the dynamics of motion integration and the respective influence between 1D versus 2D cues. For these well-known questions, representative stimuli can be chosen. These stimuli fit into two classes: gratings and line drawings.
In order to get comparable results, all the stimuli were generated from their psychophysical counterparts using the following conversions: 1° = 8 pixels, and 1°/s = 0.5 pixels/frame, thus 1 second = 16 frames.Motion estimation benchmark
Stimuli database
Gratings
Grating size
In the grating size stimuli, a translating gratting is viewed through a circular aperture. In each experiment the size of the circular aperture varies.
The grating size stimuli allows to quantify experimentally the receptive field properties of neurons involved in motion integration. In particular, it allows to quantify the maximal spatial receptive field size of neurons involved in motion integration.
Details
Related publications:
barthelemy-vanzetta-etal:06
Expected result: The early change in eye position increase with stimulus diameter before a slightly decreasing saturation.
Parameters:
sfreq: 0.27cpd
speed: 37deg/s
orient: 0deg
tfreq: 10Hz
Size: 350x350 px
Duration: 22 frames
Scoring:
dynamic
(PDF)
Barberpole
In the barber pole stimuli, a translating grating is viewed through a rectangular aperture. The orientation of the aperture is varied in each stimuli to change 2D motion cues.
The different orientation of the rectangular aperture lead to different 2D motion cues since the angle between the aperture edges and the grating orientation varies. Moreover the 2D motion cues are grouped into two sets with orthogonal motion directions, corresponding to the elongated and the short edges of the apertures.
Details
Related publications:
masson-rybarczyk-etal:00
Read-out: Final perceived motion.
Expected result: Global motion direction is given by the orientation of the longest edge of the barberpole.
Quality assessment: Angular direction error at the end.
Parameters:
speed: 2px/f
orient: 0deg
Size: 100x100 px
Duration: 20 frames
Scoring:
Line-drawings
Translating diamonds
In the translating diamonds stimuli, an elongated diamond to which two different rotations are applied is presented with a translating motion. The diamond translations are upward, downward, leftward, and rightward.
The peculiar shape of the diamonds permits to unravel the mechanisms involved in integration of 2D cues. Indeed, one cannot simply use 1D (edges) motion information for solving the aperture problem as it would lead to incoherency in the edge displacements.
Details
Related publications:
masson-stone:02
Expected result: Perceived direction moves from vector-average to object-motion with an exponential decay.
Size: 250x250 px
Duration: 20 frames
Scoring:
dynamic
(PDF)
static
(PDF)
Translating dashed bar
In the translating dashed bar stimuli, one or several translating bars are presented to the subject, corresponding to a single cutted bar, and thus leading to varying number of 2D cues.
Affecting the number of 2D cues, here the line endings, allows to study their influence on the time course of motion integration. Such a simple stimulus already necessitates some spatial motion integration to solve the aperture problem.
Details
Related publications:
lorenceau-shiffrar-etal:93
Read-out: Direction error between SPEM and the true bar motion.
Expected result: Adding more line endings to the stimuli fasten the convergence towards the true motion.
Quality assessment: Amplitude of fitted DoG decreases with breaks.
Parameters:
speed: 1px/f
orient: 135deg
Size: 100x100 px
Duration: 20 frames
Scoring: