Using Blur to Affect Perceived Distance and Size.

Area of Science:

• Computer Vision
• Perception Psychology
• Image Processing

Background:

• Defocus blur is a pictorial cue that aids in estimating absolute distances and perceived scale.
• Existing models often use approximations for blur, potentially impacting accuracy.
• Understanding how viewers utilize blur is crucial for realistic image manipulation.

Purpose of the Study:

• To develop a probabilistic model of defocus blur perception for estimating absolute distances and scene scale.
• To create a semiautomated algorithm for manipulating perceived distance and scale using blur.
• To experimentally validate the model's predictions against human viewer performance.

Main Methods:

• Developed a probabilistic model integrating defocus blur with relative depth cues.
• Created a semiautomated algorithm to apply blur for altering perceived scale and distance.
• Conducted experiments with human viewers comparing distance estimates to model predictions for accurate and approximated blur.

Main Results:

• The model demonstrated excellent agreement with human viewer estimates of absolute distance.
• Viewer responses to different blur types aligned with model predictions.
• The model accurately predicted which blur approximations would be effective.

Conclusions:

• The probabilistic model effectively captures how viewers use defocus blur for depth and scale perception.
• The developed algorithm precisely manipulates perceived scale and distance through blur.
• This work provides a robust method for controlling perceived scene properties in digital images.